tics.umls.server.term-server-model.1.2.1.source-code.rm_tty.doc Maven / Gradle / Ivy
??ࡱ? > ?? ???? ? ????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????R o o t E n t r y ???????????? ???? ???????????? ???????????? ???????????? R o o t E n t r y ???????? ? F p s? ? D a t a
???????? # ?Z 1 T a b l e ???????? ? Bp C o m p O b j ???? r ???????? ????????????????
???? ???????????????? ! " # $ % & ' ( ) * + , - . / 0 1 2 3 4 5 6 7 8 9 : ; < = > ? @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z [ \ ] ^ _ ` a b c d e f g h i j k l m n o p q r s t u v w x y z { | } ~ ? W o r d D o c u m e n t ???? ???? .? S u m m a r y I n f o r m a t i o n ( ???? ???? D o c u m e n t S u m m a r y I n f o r m a t i o n 8 ???????????? O l e
???????????? ???? ???????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????? ??
???? ? F Microsoft Word 97-2003 Document
MSWordDoc Word.Document.8 ?9?q ?? ??՜.?? +,??D ??՜.?? +,??d h p ? ? ? ? ? ? ? ?
? ? ? MSD Inc. ? t 6? / Rationalizing Inversions Workgroup: Term Types Title , 8 @ _PID_HLINKS ? A ? B g # 0 h t t p : / / w w w . n l m . n i h . g o v / r e s e a r c h / u m l s / m e t a 2 . h t m l l z ] h t t p : / / u n i m e d . n l m . n i h . g o v / v q w i k i / j s p / W i k i ? a c t i o n = a c t i o n _ v i e w _ a t t a c h m e n t & a t t a c h m e n t = M R T T Y . x l s , - b h t t p : / / u n i m e d . n l m . n i h . g o v / v q w i k i / j s p / W i k i ? a c t i o n = a c t i o n _ v i e w _ a t t a c h m e n t & a t t a c h m e n t = T e r m g r o u p s . x l s g # 0 h t t p : / / w w w . n l m . n i h . g o v / r e s e a r c h / u m l s / m e t a 2 . h t m l q G 5 h t t p : / / w w w . n l m . n i h . g o v / r e s e a r c h / u m l s / a b o u t _ u m l s . h t m l M e t a t h e s a u r u s 6 h t t p : / / u n i m e d . n l m . n i h . g o v / M E M E / T r a i n i n g / g l o s s a r y . h t m l
t e r m g r o u p 6 h t t p : / / u n i m e d . n l m . n i h . g o v / M E M E / T r a i n i n g / g l o s s a r y . h t m l
t e r m g r o u p 6 h t t p : / / u n i m e d . n l m . n i h . g o v / M E M E / T r a i n i n g / g l o s s a r y . h t m l a t o m B f . . / A p p D a t a / L o c a l / T e m p / T e m p 1 _ L M . z i p / s h e r e r t d / L o c a l S e t t i n g s / T e m p o r a r y I n t e r n e t F i l e s / O L K 2 A / r m _ t t y . h t m l f 2 K } 1 h t t p : / / w w w . n l m . n i h . g o v / r e s e a r c h / u m l s / m e t a b 3 . h t m l s b 3 _ 0 l T m a i l t o : b c a r l s e n @ a p e l o n . c o m ?? ?????Oh?? +'??0 ? ? ? ? ? ? 8 D
d p
| ? ? ? ? ? 0 Rationalizing Inversions Workgroup: Term Types Brian Carlsen Normal.dotm Brian Carlsen 3 Microsoft Office Word @ ^в @ ??p?@ T?s? ?! ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?
! " # $ % & ' ( ) * + , - . / 0 1 2 3 4 5 6 7 8 9 : ; < = > ? @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z [ \ ] ^ _ ` a b c d e f g h i j k l m n o p q r s t u v w x y z { | } ~ ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ????????? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? _? ?? n? bjbj???? .? ??5\??5\?? ? ?? ?? ?? ? | | | | | ???? ? ? ? 8 ? ? t
? ?j ? ?# ? @$ ( h$ h$ h$ C% ?
& l y& 8 j j j j j j j $ jl ? o ? ;j | ,( C% C% ,( ,( ;j | | h$ h$ ? Pj 4 Z+ Z+ Z+ ,( 0 | h$ | h$ j Z+ ,( j Z+ Z+ ? ?d L ?g h$ ???? 0?s? ???? \) F Ag 4 j ?j < ?j ug | ?o ?) $ ?o h ?g ?g ( ?o | h ? ?& Z ' @ Z+ K' 4 ' ? ?& ?& ?& ;j ;j ?* ? ?& ?& ?& ?j ,( ,( ,( ,( ???? ???? ???? ???? ???? ???? ???? ???? ???? ???? ???? ???? ???? ???? ???? ???? ???? ?o ?& ?& ?& ?& ?& ?& ?& ?& ?& > B : Executive Summary
Problem
The UMLS has never had a clear model or definition for a term type to guide inversions or tcoo present to users.
Term types are a very important aspect of the homogenization of sources into the UMLS view.? They are always based on source data but asserted by the UMLS. Although sources typically indicate or label collections of names in ways that are roughly equivalent to the UMLS term types, the methodology for assigning UMLS TTY values from this source information is not always evident.
Consequence
As a result of the first two points, term types are very confusing to users, and choices regarding their values may be perceived as capricious or arbitrary.
Users are not able to fully appreciate or leverage the value of UMLS TTY assignments.
Solution
Formal modeling of term types makes explicit a whole world of implied structure and semantics. This information can be used to inform the inversion processing of this kind of source data to make it more systematic, reliable, and predictable.? Furthermore, it facilitates NLM's ability to clearly articulate to users what term types are and how they are used by the UMLS to reflect source information. Recommendations are made below for two sets of changes to formalize the modeling of term types.
Adopting the proposed minimal set of recommendations would have a small impact on users.? A small number of TTYs would be merged, split, or changed, and these changes would cascade to a small number of MRCONSO.RRF and MRRANK.RRF entries.? For atoms that change because of these TTY changes, there would be a corresponding AUI change, which could be documented in MRAUI.RRF (as safe replacement entries).
Adopting the proposed additional set of recommendations (which includes the minimal set) would have a more significant impact on users.? An extra file (MRTTY.RRF) would be added to the distribution.? In addition, the use of locale values instead of the three letter language values would change the LAT field of MRCONSO.RRF.? Thus, any users relying on values like "ENG" would have to adjust their applications to use new-style values, like "en" or "en_US".
Recommendations
Minimal Recommendations
Create a formal stand-alone term type model based around these properties: CVT, NVT, HT, USE, STYLE (See Defining a Term Type section below for details of what these properties mean, and what their allowable values are):
Use this model in data representations across internal life cycle stages: .src files, MID.
Formalize use of this model across life-cycle internal processes: inversion, insertion, editing.
Use this model to clean up existing TTY assignments to conform to it.
Improve stand-alone term type documentation internally and for users.
Employ quality assurance checks to validate the model's representation.
Provide tool support.
Create a formal context-specific term type model based around the current MRRANK and new REFTTY and CODETYPE properties:
Use this model in data representations across internal life cycle stages: inversion/insertion (.src files) and editing (MID).
Formalize use of this model across life-cycle internal processes: inversion, insertion, editing.
Use this model to clean up existing associations between sources and term types to conform to it.
Improve context-specific term type documentation internally and for users.
Employ quality assurance checks to validate the model's representation.
Provide tool support.
Make formal term type model data available to users (i.e. MRTTY.RRF).
Additional Recommendations
Adopt a formal term type model as official part of UMLS Metathesaurus.
Deprecate the use of ?OBS? as a USE property value.? This goes hand in hand with the "Atoms" recommendation of separating "obsoleteness" and "suppressibility".?
Revise the representation of language variants:
Model language values on "locale" model (e.g. en_US) instead of current LAT values (e.g. ENG).
Clean up existing cases of TTY overloading based on language variant.
Extend the source metadata model to accommodate multiple languages within a single source.
Employ quality assurance checks to validate this new use of languages.
Thoroughly document this change internally and for users.
Support MMSYS options to produce legacy-style language values.
Problems
This section provides a little more detail regarding the problems described in the Executive Summary. As of 2007AC, there are lingering problems with existing TTY assignments. Clearer guidance for the future is needed to understand how to best manage these assignments over time.? These are the issues as we understand them today:
Term types are difficult or impossible for users to understand:
The rules are not discernible from the data.
There is some inconsistency in the use of values:
Some changes are needed (e.g. ICPC/CC => ICPC/MTH_CP).
Some splits are needed (e.g. {LNC,ICPC}/CX => LNC/CX, ICPC/CPX).
UMLS documentation regarding TTYs is lackluster.
There is significant overloading of TTY values to capture other atom properties:
Overloading of TTY for NLM additions and transformations (e.g. SNOMEDCT MTH_PT).
Overloading of TTY for NCI Thesaurus sub-source atoms (e.g. NCI FDAPT).? This is particularly problematic because then SABs overloaded in TTY are not versioned and have no metadata.
Overloading of?proper TTY properties within the name (e.g. PS, PT, PX, and MH all refer to something that is "preferred" within its group - whether it be concept or descriptor).
Overloading of TTY to represent language variants (e.g. SNOMEDCT PTGB).
The structural nature of "semantic" TTYs?is underspecified (e.g. what role does DI play within a concept?)
Overlapping of "structural" and "semantic" TTY assignments within the same source (e.g. MSH/N1 vs. MSH/MH).
Lack of uniformity with respect to qualifier term types:
Some sources assert that atoms are "qualifiers" (e.g. MSH).
Some sources assert that atoms are "attributes" (e.g. RCD).
Some sources assert that atoms are "modifiers" (e.g. HCPCS).
In some sense, these are really all the same kinds of names.
The representation of TTY properties is awkward for users (e.g. tty_class in MRDOC.RRF).
Goals
This section provides a little more detail regarding the goals behind the solution summarized in the Executive Summary.
Avoid unnecessary proliferation of term types.
Consistently handle TTY assignment across sources:
Same value should mean same thing.
Same ?thing? should not use different values.
Same set of TTY properties should apply correctly across all sources.
TTY properties should capture similar kinds of structural aspects across sources where structure may be different (e.g. PT can mean preferred name of concept or preferred name of descriptor).
Add formal and explicit properties to term types.
Avoid SAB references within stand-alone term type properties (e.g. expanded forms) or values.
Avoid unnecessary overloading of TTY values.? With explicit property representation, TTY values do not have to encode all term type properties.
Improve documentation:
Include information on how we arrived at TTY values from incoming source information.
In each case, be able to answer why we represented some feature of a source name as a TTY instead of as an attribute (e.g. SNOMEDCT INITIALCAPITALSTATUS vs. DESCRIPTIONSTATUS).
Make TTY properties more explicit and easy to access programmatically.
Understand when to use structural vs. semantic term types. This will involve characterizing sources as either being "semantically" or "structurally" organized.
Support source transparency, i.e. use values and expanded forms that are meaningful to a knowledgeable user of the source.
Specification
Minimal Recommendations Specification
Create a formal stand-alone term type model based on these properties: CVT, NVT, HT, USE, STYLE (These properties are described further in the Defining a Term Type section below.):
Expand the inversion data model (add MRTTY.RRF file to inversion or add fields to termgroups.src).
Expand MEME data model (add term_types core and staging area tables).
Add inversion and MEME QA checks for term type inconsistencies:
Flag cases of different TTY values that are both Structural but have identical properties.
Flag cases of different TTY values that are both Structural and have a very similar set of property value assignments.
Flag cases of different TTY values that are both Semantic and have identical properties and are used within the same VSAB.
Specify and develop tools to report and compare stand-alone term types for internal use:
Implement a web-based form with fields for TTY properties.? The user enters as much information as desired and matching candidate TTYs are returned (in rank order).? Make sure to match on sub-types of NVT (e.g. AB is a kind of S).
Implement a command-line tool that takes a MRTTY.RRF file for a new source inversion and reports potential QA problems (if these TTYs were inserted).
Create a formal context-specific term type model based around the current MRRANK and new REFTTY and CODETYPE properties:
Expand the inversion data model (add REFTTY, CODETYPE fields to termgroups.src).
Expand MEME data model (add REFTTY, CODETYPE columns to termgroup_rank core and staging area tables).
Formalize procedures for choosing term types during source inversion (e.g. develop flowcharts). Use tools to compare and report term types. See Choosing Term Types section below for further discussion.
Use model to clean up existing context-specific term type assignments.
Improve documentation internally and for users:
Describe how term types are used within the context of a source and have different properties.
Explain how term types are derived from source information.
Explain how different term types are supposed to be used within a source.
For any AUIs that change as a result of TTY changes, add AUI safe-replacement facts to MRAUI.RRF to help users.? Document this change for UMLS users.
Add inversion and MEME QA checks for context-specific term type inconsistencies:
Flag changes in use of REFTTY across versions of source.
Flag changes in use of CODETYPE across versions of a source.
Report cases of different REFTTY used across sources for same TTY (include TTY property info).
Make formal term type model data available to users:
Create a draft MRTTY.RRF for all term types.? Use this specification to identify TTY-level changes to existing term types that need to take place.?
Present MRTTY.RRF as de-normalized view of stand-alone and context-specific term type properties.
Consider merges, splits, additions, deletions, and changes:
Deprecate use of inexact synonym variants.
Deprecate use of term types containing unversioned sub-sources. Instead use the conventions employed for the versioning of the MDS sub-source within LNC. TTY values should never include SAB values, whether versioned or unversioned.
Consider updates to existing TTY expanded_forms. In particular, eliminate use of SAB values in expanded forms. Certain cases of source-specific structural TTYs that apply specifically to a SAB may be valid exceptions to this rule (e.g. LOINC).
Consider changes to context-specific term types required as a result of these changes.
Compute the scope of the changes before approving them (e.g. how many AUIs will change, etc).
Author a comprehensive proposal document for all TTY changes.
Release MRTTY.RRF as an adjunct file to the UMLS Metathesaurus.? This is to be produced outside the scope of the normal production system and posted to KSS.
Additional Recommendations Specification
Adopt a formal term type model as an official part of UMLS Metathesaurus release:
Officially create MRTTY.RRF data as part of UMLS Metathesaurus production process, and include MRTTY.RRF as part of each release.
Deprecate the use of "tty_class" in MRDOC.RRF.
Update UMLS Metathesaurus documentation to ensure term types are fully defined and explained to UMLS users:
Document the formal stand-alone term type model (e.g. _VT, NVT, HT, USE, STYLE)
Document the formal context-specific term type model (e.g. RANK, SUPPRESS, REFTTY, CODETYPE)
Document conditions where SY TTY is used without a REFTTY; the implication is that it is a synonym of the same-CODE atom labeled as (the meaning of the CODE is defined by CODETYPE).
Author a "bridge" document for users to explain changes.
Deprecate the use of ?OBS? as a USE property value:
Prerequisite - atoms have independent OBSOLETE and SUPPRESS attributes.
Deprecate all "obsolete" term types:?
For all atoms with these term types use non-obsolete term types instead and set those atoms ' OBSOLETE flag.
Remove any entries from MRRANK.RRF/MRTTY.RRF based on obsolete term types.
Remove any term types (MRTTY.RRF) with USE=OBS
For any AUIs that change as a result of term type changes, add AUI safe-replacement facts to MRAUI.RRF to help users.? Document this change for UMLS users.
Revise representation of language variants:
Update the inversion data model (ensure the language field in sources.src can support multiple languages). In particular, update the SNOMEDCT reversion (source transparency) scripts to accommodate these new language values.
Update the MEME data model:
Ensure language field in classes (core and staging area tables) can support full locale value.
Ensure language field in source (core and staging area tables) can support list of locale values.
Need to look through code and identify things that use 'ENG' code (e.g. we only normalize English strings).
Need to look through QA checks and identify queries that use 'ENG'.
Update MRD data model:
Ensure MRCONSO.LAT can accommodate full locale values.
Ensure MRSAB.LAT field can accommodate a list of locale values.
Ensure MRCOLS field definitions are correctly set for these fields.
Ensure MRDOC.RRF "LAT" expanded forms exist for all locale values that are used.
Update UMLSKS data model.
Formalize procedures for choosing locale values during source inversion (develop flowchart):
Use source asserted language values, do not attempt to infer language from the actual string value (same with locale, don't infer en_GB from en).
Transliterations of UTF-8 language strings will share same language as original.
Convert all three-letter language values to the appropriate two-letter ISO language code (with country code where known). For example, "en" or "en_US".
Convert all term types that use language variants to non-language variant forms.? Ensure language fields are appropriately updated to include language with country code.
Remove all term types that use language variants.
For any AUIs that change as a result of these term type changes, add AUI safe-replacement facts to MRAUI.RRF to help users.? Document this change for UMLS users.
Convert all language values in MID atoms_ui table to conform to new standard (ensures AUIs will not change).
Update UMLS documentation to clearly indicate the correct set of language values now used (including an appendix if necessary).
Add inversion, MEME, and production QA checks for TTY inconsistencies:
Every LAT value used by an atom of a source must appear in the list of LAT values associated with that source.
Thoroughly document this change internally and for users. Make it clear (in UMLS documentation) to users that "en" doesn't mean "en_US", it means "en, not otherwise specified".
Support MMSYS options to produce legacy-style language values:
Output stream should support "Language Style" option: 3-letter, 2-letter, full locale.?
Warn users who choose 3-letter option that they may lose data in the case of full locale language settings.
More Information
This document is about term types.? Contained below is a brief history and analysis of term types on their own and in the context of particular sources. This should provide additional information behind the formal recommendations for creation, management, and release of term types
Background
When categorizing Metathesaurus data, it is important to make the distinction between UMLS data and source data.? For example, CUI, LUI, SUI, AUI, and STY (semantic types) are all UMLS data, while?ATV, STR, and SAB values are source data.? The Metathesaurus attempts to organize names into native source data structures (e.g. concepts, descriptors) and in parallel organize them into UMLS concepts (represented by CUI).? This division is rigidly adhered to in the sense that one could take each file or each field of each release file and indicate whether it contains UMLS data or source data.? In some cases the actual values of source data have been modified to achieve the kind of uniformity that the Metathesaurus demands. In an important way, atoms are the point of intersection between the UMLS and source views of the world.? The rows of MRCONSO.RRF indicate a lot about the Metathesaurus view of the world and the source view, as delineated in the description of each field of MRCONSO.RRF below:
FieldCategoryMeaning / OriginCUIUMLSAssigned as a result of Metathesaurus editor editingLATSourceLanguage (standardized to a formal set of values) of the STR source nameTSUMLSAssigned based on relative rankings atoms within the CUI,LUI pairLUIUMLSAssigned based on outcome of passing STR through LVGSTTUMLSAssigned based on relative rankings of atoms within the triple CUI,LUI,SUISUIUMLSAssigned based on case-sensitive STR ValueISPREFUMLSAssigned based on relative rankings of atoms within the quartet CUI,LUI,SUI,AUIAUIUMLS Assigned atom identifierSAUISource Assigned atom identifierSCUISource Assigned concept identifierSDUISource Assigned descriptor identifierSABSource But abbreviated to fit established UMLS naming conventionsTTY*UMLSsee belowCODESource Assigned codeSTRSource NameSRLUMLSLicensing restriction informationSUPPRESSUMLSAssigned either by TTY in inversion, or atom-by-atom during editingCVFUMLSAssigned by membership into certain "content views"TTY is a Metathesaurus asserted entity whose value is based on source data.? While a source may not call it a term type, a collection of names (corresponding to a term type) will generally be labeled in such a way as to indicate a simple categorization.? This could be done by field names, XML tag names, or other kinds of labels. ? The inversion involves a transformation of this source data into an official UMLS TTY value. Where possible, an effort is made to utilize the existing set of UMLS values.? However, for some sources (e.g. NCBI), inverters were unable to find a suitable existing TTY and wound up creating new TTY values based on the naming conventions within the source itself (e.g. SCN, scientific name). While "term type" continues to be an absolutely essential field in the understanding of the role of an atom, it has not been very clearly defined.? Here is the definition of the TTY field from the Metathesaurus documentation: TTY
Abbreviation for term type in source vocabulary, for example PN (Metathesaurus Preferred Name) or CD (Clinical Drug). Possible values are listed in HYPERLINK "http://www.nlm.nih.gov/research/umls/metab3.html" \l "sb3_0" Appendix B.3.HYPERLINK "../AppData/Local/Temp/Temp1_LM.zip/sherertd/Local Settings/Temporary Internet Files/OLK2A/rm_tty.html" \l "f2"2
A "term type" is not explicitly defined in the UMLS documentation.? The MEME4 glossary uses the following definition for term type:
term type
A value indicating the general nature of an HYPERLINK "http://unimed.nlm.nih.gov/MEME/Training/glossary.html" \l "atom" atom.? Term types include things like PT for "preferred term" or SY for "synonym" or MH for "main heading".? Term types can be found at the end of HYPERLINK "http://unimed.nlm.nih.gov/MEME/Training/glossary.html" \l "termgroup" termgroup values after the slash. [See HYPERLINK "http://unimed.nlm.nih.gov/MEME/Training/glossary.html" \l "termgroup" termgroup]
Term types are a way of categorizing atoms either structurally or semantically.?? Similar kinds of atoms within the same source or across sources should share the same term type values.? Different kinds of atoms within a source should have different values.? When a term type is applied to a source, its value should be meaningful to a knowledgeable user of the source itself.? From these vague and general descriptions, it seems clear we need a more formal definition of exactly what a term type is, what its properties are, and how it is used in the context of a particular source. These are key features of a term type:
Term types indicate the relationship of an atom to its organizing cluster (e.g. concept, descriptor or code).
Term types indicate as much as possible about the term (and its organizing cluster) without having to look elsewhere.? For example, it may provide domain specific semantic information.
Term types indicate the role of an atom in its source, based on the source's understanding of the varying roles of its atoms (or the conceptual analog of its atoms)
Term types support the aspect of source transparency that directs the UMLS to represent things in as semantically similar a way as possible to the native representation.? In this case, it means that term types should be labeled in ways that are similar to the ways names are labeled within native source files.
Term types have properties that differentiate them (within a type).
Term type properties are the same across all sources in which they are used.
A term type can be interpreted in a context-specific way where the context is a source.? Context-specific term types have properties in addition to their stand-alone TTY properties (e.g. SUPPRESS).
With this more formal notion of a term type, we can assess the problems with the current state of TTYs and the goals we hope to achieve in recommending changes.
Given the information above, we can begin to see what value TTYs provide.? In the context of a Metathesaurus that unifies different kinds of terminologies, thesauri, ontologies, and other vocabularies, term types provide a unifying structure that indicates what kinds of names are the same or can be used in the same way. It is worth noting that sometimes this attempt to provide a unified view across terminologies can be at odds with the goals of source transparency.
Perhaps surprisingly, despite the lack of a formal model for internal or external use, there is a great deal of consistency with respect to term type assignments within the UMLS as of 2007AC. In a sense, we used the current state of term types to derive the implicit model that the UMLS had been using all along and identified the small number of weaknesses, inconsistencies, and problems that had developed as a result of not having a formal model.
Defining a Term Type
Here we will look at the specific properties of a term type and then look at examples from some of the more interesting sources.
Sources assert a variety of different atom attributes. Sometimes these are first class attributes (e.g. SNOMEDCT DESCRIPTIONSTATUS) and sometimes they are connected attributes (e.g. RXNORM ORIG_SOURCE).?
Our formal model of a term type attempts to capture a subset of these attributes.? By doing so, we are making a choice to represent or encode those source attributes in the term type, rather than attaching them as connected attributes in MRSAT.RRF.? In some cases, term type properties may be based on an algorithm using more than one source asserted attribute (e.g. term types for SNOMEDCT based on LANGUAGECODE, DESCRIPTIONSTATUS, and DESCRIPTIONTYPE.). Here, we will also represent this source asserted data as connected attributes in MRSAT.RRF.
Below is a list of the various term type properties and the allowable range of values assignable to each.? Each property should be assigned only one value.? This list of properties was chosen because of their importance, applicability across a wide variety of sources, and the degree of value they provide as a means of grouping atoms by type.
CVT ? CODE Variant Type.? This property indicates the relationship of the atom to the primary organizing identifier used by the source. This will be the same identifier chosen for the CODE field (for atoms of this context-specific term type). Given that some sources have both concept and descriptor structure, the range of values can support indications of preferedness for both. Range:
P ? Preferred name (use with CODETYPE=SCUI or SDUI)
SY ? Synonym. If it is a synonym of something; use REFTTY to indicate what that something is. (use with CODETYPE=SCUI or SDUI)
ET ? Entry term (typically for a concept with descriptor-only structure) (use with CODETYPE=SDUI)
PET ? Entry term of descriptor, preferred name of its concept (use with CODETYPE=SDUI where source also uses SCUI)
SYET ? Entry term of descriptor, synonym of preferred name of its concept (use with CODETYPE=SDUI where source also uses SCUI)
NVT - Name Variant Type.? What kind of name this is.? If it is a variant of something, use REFTTY to indicate that. The most specific single value specified by the source should be used. Range:
V - Lexical variant (synonym)
S - Short form (synonym)
AB - Abbreviation or acronym (synonym)
X - Expanded form (synonym)
F - Fully specified name (synonym)
L - Language variant (synonym) - recommend avoiding this
C - Common, or colloquial form (synonym)
null - n/a
HT - Hierarchical Type.? This property gives the role the atom plays in the source hierarchy (if any). Range:
H - Hierarchical: (participates in subsumption/meronomy, etc.) used for hierarchical navigation, not a leaf node, not intended for the same kind of use as leaf nodes. Under some circumstances, we may know that a node marked as HT=H is really a purely navigational node (as indicated by the source). If this is the case, the inverter should indicate this with a source asserted atom attribute.
null - n/a
USE - Usage Directive.? Occasionally sources indicate how atoms are supposed to be used in some application context.? This property attempts to capture that in a cross-source manner. Range:
DISP - Displayable
NODISP - Not intended for display (e.g. a "lookup" term) - these may be candidates for suppressibility
QUERY - Representative of a query or intended to be used for querying (entry point into a graph intended for use in querying or building a query)
QUAL - Qualifier (modifier, attribute)
MTH - A UMLS transformation of a source atom, typically for 7-bit transliteration (relationship RELA will define more accurately)
OBS - Obsolete (this value overrides all others)
null - n/a
STYLE ? Style Directive.? Primarily intended to indicate whether the term type name represents a domain-specific semantic category, or a structural association. Range:
Semantic - term type has a specific meaning in a particular domain (e.g. CD is a clinical drug).
Structural - term type relates to the position in the class structure, variation with respect to the preferred form, or role in the concept or descriptor structure (e.g. PT is a preferred term).
Source-Specific Structural - term type name carries key structural information in addition to that which can be captured by the structural properties listed above.? Typically these are used to have source-relevant TTY names whose structural properties are very similar to the straight structural cases (e.g. SCN is a scientific name in NCBI, operates in the role of a normal PT).
Metadata - term type is associated with an atom whose concept captures metadata information for a Source, Content View, Subset, or Mapping set.
This list of properties serves to render the Metathesaurus view of a term type and allows the term type value itself to reflect more closely what the source has said.? Notice that the properties serve to identify the structural role of a term type.? This allows even purely "semantic" TTY values (like CD) to be related back via similar properties to purely "structural" TTYs (like PT).? Thus, for a semantic TTY, it is still possible to understand the structural role it plays within the source.
There are a number of sources that challenge the straightforward notion of structural vs. semantic term types.? These sources are good ones to consider when thinking about assigning term types for new source data.? In fact, in any case where we are unable to harmonize a source to a straightforward structural or semantic view, we should provider users with explicit reasons why those particular sources are exceptions.?
Examples
In the following sections are some sample sources that use structural and semantic term types in different ways. The codes for the TYPE values are:
SSS ? Source-specific Structural
STR ? Structural
SEM ? Semantic
META - Metadata
LOINC LOINC is a source that does have a specific domain focus (lab tests), but in general organizes its terms structurally.? Below is the historical list of LOINC TTYs with their property assignments from the list above:
TTYEXPANDED FORMTYPE_VTNVTHTUSECNLOINC official component nameSSSPCXComponent process in ICPC with abbreviations expandedSSSSYXHCHierarchical classSTRPHLNLOINC official fully specified nameSTRSYFLOObsolete official fully specified nameSTRSYFOBSLPDNLOINC parts display nameSSSSYDISPLPNLOINC parts nameSSSPLSExpanded system/sample typeSEMPLXOfficial fully specified name with expanded abbreviationsSTRSYFOLXExpanded LOINC obsolete fully specified nameSTRSYFOBS OOSNObsolete official short nameSTRSYSOBSOSNOfficial short nameSTRSYSSNOfficial component synonym in LOINCSSSSYSXMixed-case component synonym with expanded abbreviationsSTRSYV From this, the first thing to notice is that there is a mixture of Semantic, Structural, and Source-Specific Structural TTYs.? If not for "LS", we could declare LOINC a source with completely structural term types.? LS is considered Semantic because in this context "system" refers to a system of the body in which some measurement is being taken (e.g. "blood serum").? This is a domain-specific term type and not one that describes the relationship of the atom to the structure of the source.? Those things marked as Source-Specific Structural have to do with the fact that LOINC defines structural elements called "component" and "part" that are not recognized structural elements within the standardized term type property list.? We recognize that these things, which are derived from sections of the native LOINC files, are eventually represented as concepts and we can identify which term types characterize preferred and synonymous names of those concepts.? That is why these things are ultimately considered structural. All of those things labeled as "P" in the _VT column could just as easily have been called "PT", but then the term type names would not be as useful to a knowledgeable user of LOINC as the names we chose.? This is an important aspect of source transparency. MDR MeDRA is a source that does have a specific domain focus (adverse reactions), but in general organizes its terms structurally.? Below is the historical list of MDR TTYs with their property assignments from the list above:
TTYEXPANDED FORMTYPE_VTNVTHTUSEHGHigh Level Group TermSTRPHHTHierarchical termSTRPHLTLower Level TermSSSETOLNon-current Lower Level TermSTRETOBS OSSystem-organ class in the WHO Adverse Reaction TerminologySEMPHPTDesignated preferred nameSTRPSMQStandardized MedDRA QuerySEMPQUERYXMCross mapping setMETAP In this source, notice that there is both descriptor and concept structure.? As with the LOINC case, there is a mixture of Semantic, Structural, and Source-Specific Structural TTYs.? If not for "OS", we could declare MDR a source with completely structural term types.? OS is considered Semantic because the word "organ" is clearly a domain-specific word with semantic implications. The term type "LT" is marked as Source-Specific Structural because MDR defines the structural relationship of these atoms as being at a "lower level".? The standardized term type property list does not have an entry for this.? The inverter has determined that these things play an "entry term" role within the MDR descriptor, so the choice is to use this name. One other interesting thing to note about MDR is that it natively distinguishes between different aspects of the hierarchy.? It clearly labels some things as "classes", other things as "groups", and still other things as being merely "hierarchical".? This notion led us to consider having different types of "hierarchal" nodes.? Eventually, we settled on representing these all as HT=H and letting the "source specific structural" nature of the TTYs tease out the differences. MSH MeSH is a source that does not have a specific domain focus and in general organizes its terms structurally.? The only exceptions are N1 and NM which are semantic term types for chemical names. Below is the historical list of MSH TTYs with their property assignments from the list above:
TTYEXPANDED FORMTYPE_VTNVTHTUSEDEVDescriptor entry versionSTRSYDSVDescriptor sort versionSTRSYNODISPENMeSH non-print entry "term"STRSYETNODISPEPEntry "term"STRSYETDISPHSShort or alternate version of hierarchical termSTRSYHHTHierarchical termSTRPHMHMain headingSSSPN1Chemical Abstracts Service Type 1 name of a chemicalSEMSYETNMSupplementary chemical "term", a name of a substanceSEMPPCEPreferred entry "term" to a Supplementary Concept "term"STRPETPENPreferred MeSH nonprint entry "term"STRPETNODISPPEPPreferred entry "term"STRPETDISPPMMachine permutationSTRSYVPXQPreferred term in preferred qualifier concept.STRPQUALQABQualifier abbreviationSTRSYABQUALQEVQualifier entry versionSTRSYQUALQSVQualifier sort versionSTRSYQUALTQTopical qualifierSTRPQUALXQAlternate name for a qualifierSTRSYQUAL This source is primarily a Structural source but has two term types identified as Semantic.? The "supplementary" portion of MSH contains chemical names and as such, this is a domain-specific kind of term type.? It may not be possible to convert these to Structural term types (e.g. by referring to "supplementary" concepts instead of "supplementary chemicals"), as the values N1 and NM are meaningful to a user of MSH and should probably continue to be used. One other thing to note about this source is the presence of both concept and descriptor structure.? The term types clearly label those atoms that are preferred members of their descriptor, and others that are preferred members of their concept.? This source also has a great number of qualifier term types, and sometimes draws a distinction between displayable and non-displayable terms. The most recent addition to the term type list for MSH are the "abbreviation", "entry version", and "sort version" term types (DEV, DSV, QAB, QEV, QSV).? These are based on information in the MSH XML that previously had not been inverted as "names".? A decision was made to make atoms out of this information and label them with term types that are clearly derivative of the XML tags from which they came. NCBI NCBI is a source that does have a specific domain focus (organism taxonomy), but in general organizes its terms structurally.? Below is the historical list of NCBI TTYs with their property assignments from the list above:
TTYEXPANDED FORMTYPE_VTNVTHTUSECMNCommon nameSSSSY CEQEquivalent nameSSSSY SCNScientific nameSSSPSYDesignated synonymSTRSYUCNUnique common nameSSSSY CUEUnique equivalent nameSSSSYUSNUnique scientific nameSSSSYUSYUnique synonymSSSSY NCBI is a source with entirely structural semantic types.? One could probably argue that "scientific name" implied some kind of domain-specific semantics, but it is broad enough and not directly targeted to the domain coverage of NCBI to make a good case.? What is peculiar is that almost all of the TTYs are considered Source-Specific Structural.? This is because NCBI makes an explicit distinction between "scientific", "equivalent", and "common" names as well as sometimes indicating that a name is the "unique" form of the name.? These are labels that are meaningful in the context of NCBI but which are not directly represented in the term type property list.? From the property assignment, however, one can see that SCN plays a role very similar to "PT" and the others are different forms of synonym of the NCBI concept.
NCI NCI is a source that does not have a specific domain focus, and it freely mixes its terms structurally and semantically.? This terminology is targeted towards use in cancer research. Below is the historical list of NCI TTYs with their property assignments from the list above:
TTY EXPANDED FORM TYPE _VT HT USE ABAbbreviation in any source vocabularySTRSY ADAdjectiveSEMPBioCPTBioCarta PTSTRPBNFully-specified drug brand name that can not be prescribedSEMPCCNChemical code nameSEMPCCSFIPS 10-4 country codeSEMPCDCPTCenters for Disease Control and Prevention PTSTRPCDISCPTClinical Data Interchange Standards Consortium PTSTRPCDISCSYClinical Data Interchange Standards Consortium SYSTRSYCSNChemical Structure NameSEMSYDCPPTDivision of Cancer Prevention Program PTSTRPDNDisplay NameSTRSYDISPDTPSYDevelopmental Therapeutics Program SYSTRSYFBDForeign brand nameSEMPFDAABFood and Drug Administration ABSTRSYFDAPTFood and Drug Administration PTSTRPFDASYFood and Drug Administration SYSTRSYHDHierarchical descriptorSTRPHJAXPTNCI Mouse Terminology PTSTRPJAXSYNCI Mouse Terminology SYSTRSYKEGGPTKyoto Encyclopedia of Genes and Genomes PTSTRPNCI-GLOSSPTNCI-GLOSS PTSTRPNCI-GLOSSSYNCI-GLOSS SYSTRSYOPObsolete preferred termSTRPOBS PTDesignated preferred nameSTRPSYDesignated synonymSTRSY Like PDQ, this is another case of a source that clearly mixes Semantic and Structural term types.? The NCI thesaurus is an aggregation of a variety of other terminologies, some of which indicate the use of Structural term types and some of which indicate the use of Semantic term types.? Due to the nature of this thesaurus, and the fact that it contains varied sub-sources, it does not make sense to attempt to convert this to an entirely semantic or entirely structural view. Another thing to note about NCI is the inclusion of term types that are overloaded with SAB values.? A decision was made with this source to include sub-source atoms but label them with the NCI source instead of creating individual SABs for each sub-source.? The mechanism chosen to differentiate NCI atoms from sub-source atoms was to prepend an SAB-like value to the beginning of each TTY.?? This is problematic for several reasons: 1) the SAB-like values are never expanded and explained with SRC concepts or MRSAB.RRF entries; 2) it leads to unnecessary proliferation of term types that would otherwise be standard ones (like PT and SY), and; 3) it encodes information that cannot be made explicit in the UMLS domain model. PDQ PDQ is a source that does have a specific domain focus (cancer clinical trials) and it freely mixes its terms structurally and semantically.? Below is the historical list of PDQ TTYs with their property assignments from the list above:
TTYEXPANDED FORMTYPE_VTNVTHTUSEABAbbreviation in any source vocabularySTRSYABACRAcronymSTRSYABBDFully-specified drug brand name that can be prescribedSEMPCCNChemical code nameSEMPCHNChemical structure nameSEMPCUCommon usageSTRSY CETEntry "term"STRET FBDForeign brand nameSEMPHTHierarchical termSTRPHISObsolete SynonymSTRPOBS LVLexical variantSTRSYVOPObsolete preferred termSTRPOBS PSCProtocol selection criteriaSEMPPTDesignated preferred nameSTRPSYDesignated synonymSTRSY Here we have a source that clearly contains both Semantic and Structural term types. The various drug and chemical types are clearly domain-specific types with semantic meaning, and the remaining ones are the clear-cut structural cases, like PT, SY, and HT.? Due to the nature of the source itself, and the way it labels its own names in the native files, it does not make sense to attempt to convert this to an entirely semantic or entirely structural view. Special Role of HT and SY As of 2007AB, a review of the context-specific term types (i.e. SAB, TTY) reveals a kind of special role for HT and SY within this division of structural and semantic term types.? It is a desirable property for a source to have term type assignments that are entirely structural or entirely semantic.? This allows for a cleaner interpretation of the atoms and term types of a source than if there is a mixture of the two types. There are a variety of cases of sources that would otherwise have entirely semantic term types if it were not for the presence of HT.? HT is a term type used to indicate that atoms represent hierarchical structure above the leaf nodes of the source.? In a source with? semantic term types and a hierarchy, the term type assignments should be derivable from the hierarchy itself.? This is because a hierarchy subdivides the source into semantic categories.? However, as the atoms of the hierarchy define the semantic categories themselves, they typically do not actually belong to these semantic categories.? This is especially true of the highest levels of the hierarchy.? Given that, even a source with only semantic term types will need to make use of HT.? There are also a variety of cases of sources that would otherwise have entirely semantic term types if it were not for the presence of SY.? For sources that are semantically organized, SY defies a clean model because it prevents a user from being able to identify what kind of atom this thing is a synonym of.? In a purely structural source, SY will indicate synonymy with the concept structure, but there is no semantic distinction between particular kinds of concepts and other kinds, so there is no ambiguity.? Back to the semantic case, in a source with semantic term types, one would want to know what kind of SY it is.? Thus, it seems clear that semantic-SY term types should be used instead of the purely structural SY in these cases. Consider AIR93 as an example.? This source (as of 2007AB) makes use of four term types: DI, FI, SY, and HT.? Two of these are semantic: DI (disease), and FI (finding).? Two of these are structural: SY (synonym) and HT (hierarchical term).? Consider these sample atoms: DI atom: Carpal tunnel syndrome Context for DI atom: ?? AIR93/AIR93/FI1/DXCTS ?? AI/RHEUM ???? Examiner's Diagnosis (term type = HT) ?????? Neurological Problem (term type = HT) ??????? FI atom: Cardiac murmur Context for FI atom: ?? AIR93/AIR93/FI1/MURMR ?? AI/RHEUM ???? Menu of Other Findings: Systems Review (term type = HT ) ?????? Heart (term type = HT) ??????? ? One can see from these two examples that the levels of the hierarchy above the diseases and findings are labeled as HT.? That is because these atoms do not actually represent diseases or findings within the source, but are intended to provide a basis for distinguishing and organizing the leaf nodes, which are diseases and findings.? Now take a look at an SY case: SY atom: primary OA Here, it may not be immediately clear if this is a synonym of a finding or a disease.? To determine this, we need to look at what other AIR atoms share a concept with this one.? A lookup in MRCONSO.RRF shows these atoms: C0029408|ENG|S|L0085247|PF|S0003722|N|A0009717||||AIR|DI|DJD|Degenerative joint disease|0|N|256| C0029408|ENG|S|L0085247|VO|S0366815|Y|A0405887||||AIR|FI|DXDJD|Degenerative Joint Disease|0|N|| C0029408|ENG|S|L0313782|PF|S0424294|Y|A0487244||||AIR|SY|DJD|primary OA|0|N|256| This is even more confusing because the concept shares a finding and a disease.? By following the codes, one can determine that the SY is a synonym of the DI atom, thus it is a "disease synonym".? When representing synonyms within an otherwise semantically organized source, one thing to consider is using semantically-specific SY term types.? Thus, in the case of AIR93, we could use "SYDI" and "SYFI" to indicate that the various synonym atoms are synonyms of either DI or FI atoms.? This clarifies the nature of the term type without leading to an unnecessary proliferation of types (2 types instead of 1).? If this were the case, we would see: SYDI atom: primary OA This would much more clearly and transparently indicate what it is a synonym of.? In lieu of the use of semantically organized synonyms, the implicit assumption is that it is an SY of the matching CODE preferred term atom. Special Role of PT In rare cases, we may also consider PT to be a kind of term type like HT or SY. A source like RXNORM contains a number of semantic term types that define the semantic nature of an individual RXCUI.? Given the naming conventions of some RXNORM term types, it may be desirable for RXNORM to assert a better preferred name than that required by the model.? In these cases, otherwise entirely semantically-oriented source concepts will contain a purely structural TTY, namely PT.? It is possible that different kinds of RXNORM concepts will require these PT names.? Thus a PT may be a PT for an SCD, or maybe for an SBD.? In the same way as SY does not distinguish between what kind of SY it is, these kinds of PTs do not distinguish between what kinds of PTs they are.? To determine that, the user would have to do a same-code join within the source to see what it is a preferred name of.
Defining a Context-specific Term Type
A context-specific term type is a term type expressed in the context of a source.? This data is currently represented in the release in the MRRANK.RRF file.? Context-specific term types are ranked so that we can use this as a basis of ranking atoms within the UMLS to determine things like the:
Preferred atom of a concept
Preferred atom of a source concept
Preferred atom of a LUI within a concept
Preferred atom of a SUI within a concept,LUI pair
Term types have a set of property definitions that remain constant across sources.? Thus "PT" used in SABx would have the same properties as "PT" used in SABy.? Context-specific term types have properties that can be different from one source to the next, even if the term type is the same.? For example, the NCI/PT context-specific term type is not considered suppressible, but the RCD/AB one is.? Thus suppressibility is a property of a context-specific term type that changes from source to source. There are a couple of properties that can apply independently to term types within the context of a source.? We call these context-specific term type properties.? There are already two of these defined in MRRANK.RRF.? To fully and accurately represent the semantics of a context-specific term type, the following properties are needed:
RANK.? This is a numeric value indicating the relative rank of this context-specific term type to another one (higher values mean higher rank).
SUPPRESS.? This indicates whether or not the context-specific term type is suppressible.? Range:
Y - suppressible
N - Not suppressible
REFTTY.? In the case that the term type is a synonym or entry term, the field refers to another term type within this source that it is a synonym or entry term of.? For example PEN in MSH should have a REFTTY of MH. Alternatively, SY in SNOMEDCT has a REFTTY of PT. This applies as well to the sub-types of synonym (expansion, short form of, etc.).??
CODETYPE.? Indicates (for this context-specific term type) what kind of identifier the CODE is. Range: SCUI, SDUI, null (for other).
REFTTY is most useful for sources that use Source-specific Structural or Semantic term types because the TTY values themselves do not obviously reflect structural characteristics.? For example, in LOINC we find LNC/OLX.? Its term type properties indicate that it has a concept variant type of SY, but it may not be obvious exactly what it is a synonym of (as OLX is a term type only in LNC).? A REFTTY value of LO in this case indicates very clearly to the user that LNC/OLX are expanded synonyms of LNC/LO atoms.? This may also be represented in MRREL.RRF via SFO/LFO "has_expanded_form" relationships, but this broadens the assertion to apply to all atoms within those respective context-specific term types.
Choosing Term Types
When approaching a new source to assign term types to atoms within the source, there are a variety of considerations.? The inverter should start by assessing whether or not the term types for this source will need to be semantic or structural.? A review of term types that may be appropriate for reuse is in order as well. However, before that is even done, the first step is to determine how to define term type categories at all.? Below is one methodology for doing this.
Identify the various ways types or categories of atoms are indicated or labeled within the source:
XML tag names.
Relational file labels.
Explicit indications in documentation.
Are atom properties required to differentiate atoms within a category?
Does the organizing UI (concept or descriptor) always indicate a "preferred name"?
Does the preferred name always have one or more synonyms?
Are the indications or labels structural or semantic in nature?
Determine whether this source uses STRUCTURAL or SEMANTIC term types consistently.? This is primarily done by having a formal model of the source itself that characterizes things like its domain coverage and granularity.? It is also done by understanding whether the source organizes names by domain or sub-domain, or whether it organizes them relative to concept or descriptor structure.? For example, SNOMEDCT indicates (via DESCRIPTIONSTATUS) that a description is the "preferred" form of the concept, whereas RXNORM indicates that an atom is a "Clinical Drug" and says nothing explicit about its relation to the RXCUI.
The prior step should produce a set of categories: T1, T2, T3.?? For each of these placeholder term types, assign attribute values for each of these TTYs, including _VT, NVT, HT, USE, and TYPE.
Note: it is not required that each concept in the source have an atom with a TTY marked as _VT=P.? Consider the case of RXNORM/OCD.? In this example, there are multiple names for the concept that all have equal precedence.? In that case, there isn't one that is clearly indicated to be preferred, therefore the OCD TTY should be considered _VT=S without a REFTTY.? This is truly a concept without a source-asserted preferred name.
If there are key properties of the placeholder term types that are outside the scope of the standard UMLS term type property list, this is a red flag.? Consider re-categorizing the placeholder term types to avoid a dependency on these key properties.? Consider inverting these key properties as atom attributes instead.
If any two placeholder term types share substantially similar properties, consider consolidating them into a single term type.
Take each resulting placeholder term type and look up the property assignments to find existing UMLS term types with similar property sets.? If it is possible to reuse existing term types to capture the key properties of the placeholder term types and the existing term types have names that are relevant and meaningful to a user knowledgeable about the source, then reuse these existing term types, but:
Avoid unnecessary proliferation (reuse where the possibility to reuse exists).?
Consider modifying an expanded form (to make it more expressive) if there is a TTY that is "close enough".
Create documentation to track the TTY assignments back to the native source file indications.
This whole process would be greatly assisted by tool support.? One tool could perform term type matching based on similarly term type attribute assignments.? Another tool could produce a report of candidate term types for quick assessment. At the conclusion of this process, the inverter should have chosen a set of term types that:
Maximize reuse of existing term types where property lists match and naming conventions are appropriate
Avoid creation of new term types with VERY similar names or properties
Avoid unnecessary proliferation
Avoid overloading of term type with values that encode non-standard term type properties.
Capture information about any cases of source attributes being attached to all atoms with a term type where the attribute value is always the same.? This is a possible candidate for a new term type property. One other consideration is that if we choose to represent a group of atoms using a Structural TTY where the source seems to indicate a Semantic TTY, it is desirable to encode the semantic properties of those atoms with connected attributes in MRSAT.RRF.? One case where this may happen is a source that has mostly structurally organized atoms but one or two groups of semantically organized atoms. If the inverter decides to represent all term types for this source structurally (to avoid mixing Structural and Semantic term types), then this would have to be considered.? Finding Equivalent Term Types When attempting to determine which TTY values to use, an inverter?s greatest challenge is to figure out if any existing term types express enough of the right properties to reuse them.? In general, only Structural and Semantic term types will be able to be reused.? It is possible that within a particular domain space a Source-Asserted Structural term type will be a candidate, but only if both sources happen to name their constituent pieces in the same way. Suppose an inverter is at the step above where they have identified the candidate term type categories: T1, T2, T3, etc.? The next step is to take the properties assigned to these categories and compare them against existing term types.? This should definitely be a tool-assisted effort. Following are ways in which a term type definition may be equivalent to an existing term type:
Similarity of expanded form.? This is typically useful for Semantic term types (like "Disease" or "Finding").
Identical use of _VT.? For a TTY to be reused in a new source, the candidate must have the same values for both of these properties.? It would not be possible, for example, to reuse a _VT=ET term type in a source that did not have descriptor structure.
Similarity of NVT.? As NVT values are hierarchically organized, it is possible to compare across different levels (e.g. AB and S).? It is worthwhile also to consider cases where the candidate uses a value like O and an existing term type uses another of the values.
Identical use of HT.
USE may not be known in one case or another, so the inverter will have to use a judgment call to determine whether or not identical USE values are required if, for example, the candidate TTY has a value like DISP and an otherwise matching existing UMLS TTY has a value of null.? Such cases should definitely be considered.
Identical use of OBS.? It would not be appropriate to assign an obsolete TTY value to a candidate term type that is not indicated as obsolete (or vice versa).
When comparing TYPE, consider Structural and Source-asserted Structural to be equivalent, and avoid considering matches where the candidate is in one of these two categories and an existing UMLS TTY has TYPE=Semantic (or vice versa).
QA Issues
An effort to formalize TTY usage during inversion will result in changes to the data representation of certain sources.? This is likely to create a legacy data problem that will need to be addressed.? Furthermore, any changes in our notions of how to use TTY over time may lead to new legacy data problems. We will have to make a determination about whether to allow legacy sources to be brought into line as they are updated, or to undertake a massive effort to normalize representations of source semantics in UMLS Metathesaurus all at once. Periodic review should be undertaken to attempt to find potential merges and potential splits with respect to TTY use.? A potential split is a case of a single TTY value used across sources where the semantics of that TTY do not really apply in both cases (should be two values).? A potential merge is a case of two or more TTY values being used across sources where the semantics of those TTYs are identical and the source transparency considerations do not require the values to be different (should be one value). Periodic review should be undertaken, as well, to determine if there should be new formal UMLS term type properties.? We may want to consider having a "source asserted" term type property mechanism that allows sources to further describe the TTY values we use.? We notice trends or patterns over time across multiple sources with respect to properties they assign to term types.? These things should be considered candidates for inclusion into the formal UMLS model. Periodic review should be undertaken to identify any existing or new cases of sources that have both structural and semantic term types.? We may want to consider ways to make these sources have only structural or semantic term types.
References
Footnotes:
Definition of UMLS Metathesaurus from HYPERLINK "http://www.nlm.nih.gov/research/umls/about_umls.html" \l "Metathesaurus" http://www.nlm.nih.gov/research/umls/about_umls.html#Metathesaurus.
Definition of TTY from HYPERLINK "http://www.nlm.nih.gov/research/umls/meta2.html" http://www.nlm.nih.gov/research/umls/meta2.html.
HYPERLINK "http://unimed.nlm.nih.gov/vqwiki/jsp/Wiki?action=action_view_attachment&attachment=Termgroups.xls" MRTTY spreadsheet (on unimed wiki).
HYPERLINK "http://unimed.nlm.nih.gov/vqwiki/jsp/Wiki?action=action_view_attachment&attachment=MRTTY.xls" MRSABTTY spreadsheet (on unimed wiki).
Use the following references for more information.
HYPERLINK "http://www.nlm.nih.gov/research/umls/meta2.html" Metathesaurus Documentation
Rationalizing Inversions Workgroup: Term Types
Page PAGE 22Contact: HYPERLINK "mailto:[email protected]" Brian A. Carlsen
Created: 08/31/2007
Last Updated: 11/27/2007
M Q S \ u w ? ? ? ? ?
?
?
^ l ?
c
?
?
?
?
?
?
# 4 D N O v ? ? ? ? ? ? ? ? ? ` a q ? ? ? ? ? ? ? ? ? ? ???????????????????????????????????????????????????????????????????? h1P 6?h?#? h?z? h]? h?
hhWo h?.? hKo h?
h?? h VU h!? hKo h?TB 5?hL h x? h1P h?]? h?TB 6?]? h?TB D ?
? ? ? M ? P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB W
& F ? ?h?d ?? Eƀ ?+?f ?? [$^? `?hgd x? gd?TB gd gE
?
? ] X gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB
? ? a O
& F ?d ?? Eƀ ?+?f ?? [$gd?TB O
& F ?d ?? Eƀ ?+?f ?? [$gd?TB ? ` q ? f ? ? ? S P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB gd?TB gd?J? P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB c e ? ? & + U f p | , 7 C J ? ? d h m ? ? ? ? ? ? ? ? ? ? n o ? ? ? ? 9 ; I ? ? ? ? ? ? $ % ? ? ? u z ? ? ? ? ? ? G H W p ? ? ? ? ? ? ??????????????????????????????????????????????????????????????????????????????h? ? h?TB B*ph? h?kh hPy? h?TB 6?]? h;o( h? H h?[ h1P hhWo h?TB hRm? h?? Nf ? " ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB " h ? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ? ? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB d ? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ? Y ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB Y o ? ? ? ] X gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gdhWo P
& F ?d ?d Eƀ ?+?f o [$\$gd;o( ? ? ? ] P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB ? ? H ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB H ? ? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ? 0 j ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB j ? ? ? ? ? ? ? L P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ?d ?d [$\$gd?TB gd gE P
& F ?d ?d Eƀ ?+?f o [$\$gd?)t ? l ? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ? ? ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB I ? ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ? ? ? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ? S ? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ? r ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB r ? ? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ? ! " ^ _ ` o ? ? ? ? ? ? P [ u ? ? ? o! ?! " " ," /" 9# C# O# ?# ?# ?# p$ x$ y$ ?$ ?$ ?$ ?$ ?$ X% {% ?% ?% ?% ?% ?% ?% & & && <& >& Q& R& T& V& j& l& ?& ?& ?& ?& ?& ?& ?& ?& ' ????????????????????????????????????????????????????????۰??ϰ???? h?)t h?)t 6?h?D5 h?z? h]? h?\? h?? hhWo h?TB 6?]? h?)t h* ; h;o( h/v h?X? h| 6?h| h?.? h? ? h?kh hRm? h?TB B? # ` ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ` ? ? ? u ? ] X S gd?TB gd gE P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB u ? ? ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?kh ? ? )! ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB )! o! /" a" ? ? N P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?kh
& F ?d ?d [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB a" ?" O# ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?kh P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?kh O# g# ?# ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ?# p$ ?$ X% ? ] N
& F ?d ?d [$\$gd/v P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?kh X% ?% ?%
& ?& ? ? ? S P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB gd?TB gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?kh ' ' J' O' S' _' ?' ?' ?' ?' 5( ?( I( K( W( X( y( ?( ?( &) 2) <) M) N) ?) ?) ?* C+ D+ E+ L+ P+ j+ p+ z+ ?+ ?+ ?+ ?+ ?+ ?+ ?+ ?+ ?+ ?+ ?+ ?+ ?+ ?+ ?, ?, ?, ?, - - <- J. M. ?. ?. ?. @/ A/ B/ u/ w/ |/ ~/ / ?/ 0 0 0 ???????????????????????????????????????????????????????????????¾¾???ݷ h;o( h!? h?z? h 4 h;o( h?)t 6?]? h? ? hrt> hhWo h?5@ h?[ h/v h?^ h??8 h?TB 6?]? h?TB h?)t H?& #' i' ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB i' ?' ( ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ( {( ?( ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ?( O) 6* ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB 6* ?* E+ ? ] P
& F ?d ?d Eƀ ?+?f . [$\$gdhWo P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB E+ ?+ ?+ ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ?+ ?,
- ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?)t
- =- ?- ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ?- ?- ". ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ". ?. / ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB / B/ / ?/ ? ? N P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB
& F ?d ?d [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ?/ 0 ?0 ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?)t P
& F ?d ?d Eƀ ?+?f . [$\$gd!? 0 0 0 N0 W0 d0 g0 ?0 ?0 ?0
1 D1 E1 W1 ^1 a3 }3 ?4 ?4 5
5 5 )5 ,5 O5 W5 X5 u5 y5 ?5 ?5 ?5 ?5 t6 u6 v6 6 ?6 ?6 ?6 ?6 ?6 ?6 ?6 ?6 ?6 ?6 ?6 ?6 7 !7 "7 ,7 57 67 <7 =7 ]7 ?7 ?7 ?7 8 58 D8 E8 R8 U8 ?8 ?8 59 N9 a9 j9 ?9 ?9 ?9 ?9 l: p: ?????????????????????????????????????????????????????????????????????????????? h?$6 h?lx h]? h?qH h? ? h? ? h?TB h?TB 6?]? h;o( h?)t 6?]? h?
h?5@ h?)t N?0
1 F1 ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?)t P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?
F1 q1 Z2 ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?)t P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?)t Z2 N3 ?3 ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?)t P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?)t ?3 4 A4 ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?)t P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?)t A4 ?4 5 Y5 ? ? X P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?
Y5 ?5
6 ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB
6 v6 ?6 ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd;o( P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ?6 #7 ?7 ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd;o( P
& F ?d ?d Eƀ ?+?f ?? [$\$gd;o( ?7 8 F8 ? ] P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB F8 ?8 ?8 ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ?8 "9 m9 ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB m9 ?9 9: ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB 9: e: F; b; ? ? N P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB
& F ?d ?d [$\$gd?5@ P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB p: ?: ?: ?: ?: ?: ?: ?: ; +; 1; M; Q; ?; ?; ?; ?; ?; ?; ?; ?; ?; ?; ?< ?< ?< ?< k> l> ?? ?? |@ ?@ ?@ ?@ ?@ ?@ ?@ ?@ 8B ;B ?D ?D ?D E 2E ?E ?E ?E LF QF ZF `F ?F ?F ?F ?F ?F ?F 7G QG ?H ?H ?I ?I ?I J J CJ OJ ????????????????????????????????????????????????????????????????????? h?? h?;/ CJ aJ h?+? h?;/ h?? h?;/ 5?CJ aJ hmD hT`{ h1P hdpI h gE h?$6 h?3? h?? h?TB 6?]? h? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB > m> ?> ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ?> P? ?? ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ?? ?@ ?@ ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?? P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ?@ fA ?A ? ] P
& F ?d ?d Eƀ ?+?f
o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ?A SB ?B ? ] P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB ?B C ?C ? ] P
& F ?d ?d Eƀ ?+?f
o [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ?C ?C RD ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f o [$\$gd?TB RD ?D ?D ?E ?E ?I ?I ?I J ? ? ? ? ? ? ? ? ?d $If \$gd?? gd?TB gd gE gd?TB gd gE P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB J J J J CJ ` S S S ?d $If \$gd?? ? kd $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? CJ DJ HJ OJ ?J ` P P P ?d ?d $If [$\$gd?? ? kd? $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? OJ ?J ?J ?J ?J "K ,K wK ?K ?K ?K L L -L ;L TL bL ~L ?L ?L ?L ?L ?L ?L ?L M M M $M 1M 6M @M bM cM lM qM ?M ?M ?M ?M N N mN yN ?N ?N ?O P P P ?Q ?Q BR CR ?R ?R ?R ?R ?R ?R S S ?????????????????????????????????????????????????ɸɰɰ??ɰ?? ?jF h, [ Uh, [
h?TB 0J j h?TB Uh?TB 5?OJ QJ \?^J h_? h?TB h?? h?;/ CJ aJ #h?? h?;/ 5?B*CJ \?aJ ph? h?;/ h?+? h?;/ h?? h?;/ CJ aJ =?J ?J ?J ?J ?J ` P P P ?d ?d $If [$\$gd?? ? kd? $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? ?J ?J ?J ?J "K ` P P P ?d ?d $If [$\$gd?? ? kd@ $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? "K #K 'K ,K wK ` P P P ?d ?d $If [$\$gd?? ? kd $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? wK xK |K ?K ?K ` P P P ?d ?d $If [$\$gd?? ? kd? $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? ?K ?K ?K ?K L ` P P P ?d ?d $If [$\$gd?? ? kd? $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? L
L L L -L ` P P P ?d ?d $If [$\$gd?? ? kd@ $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? -L .L 3L ;L TL ` P P P ?d ?d $If [$\$gd?? ? kd $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? TL UL ZL bL ~L ` P P P ?d ?d $If [$\$gd?? ? kd? $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? ~L L ?L ?L ?L ` P P P ?d ?d $If [$\$gd?? ? kd? $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? ?L ?L ?L ?L ?L ` P P P ?d ?d $If [$\$gd?? ? kd@ $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? ?L ?L ?L ?L M ` P P P ?d ?d $If [$\$gd?? ? kd $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? M M M M $M ` P P P ?d ?d $If [$\$gd?? ? kd? $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? $M %M )M 1M 6M ` P P P ?d ?d $If [$\$gd?? ? kd?
$$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? 6M 7M ;M @M bM ` P P P ?d ?d $If [$\$gd?? ? kd@ $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? bM cM lM qM ?M ` P P P ?d ?d $If [$\$gd?? ? kd $$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? ?M ?M ?M ?M ?M ` P P P ?d ?d $If [$\$gd?? ? kd? $$If ?H ? ?F ?g "?? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? ?M ?M ?Q S ?S ?S ?U X ` [ R K [ B [ ??^??gd?TB ?? gd?TB ??^??gd? ? gd?TB ? kd?
$$If ?H ? ?F ?g " ? ?
t ??0 ? ? ? ? ? ? ? 6? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ?4? 4?
H a?hp? ? ? ? ? ? ? yt?? S S S S ?S ?S ?S ?S #T $T (T )T ?T ?T U U U U 4U 5U ?U ?U ?U ?U ?V ?V X X sX ?X ,Y @Y ?Y ?Y ?[ ?[ ?[ ?[ ?[ ?[ ?[ \ 3\ ?\ C\ O\ T\ ?\ ?\ A^ ?_ ?_ ?` ?` ?c ?c }e ~e ?e ?e ?e ?e ???????????????????????????????????????????????????ż?豪??? he"? 5?\? h?qH 5?\? h?TB 5?\? h?' 5?\? hu
? hdpI h1P 6?h1P h(J? h?? h?;/ hT`{
h?TB 0J h?TB 0J 5?\? h?TB h?TB 0J CJ H*aJ j h?TB U =X sX ,Y ? a P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB M
& F ?d Eƀ ?+?f . \$gd(J? ,Y ?Y [ ? ] P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB [ L[ ?[ ? ] P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB ?[ _\ ] ?^ ?` ?` 3a b %d }e ? ? ? ? ? ? ? ? ? gd?TB gdu
? gd1P ?d ?d [$\$gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB ?e ?e ?e ?e pf qf ?f g g g g 5g 9g :g ;g Bg Wg cg dg |g ?g ?g ?g ?g ?g ?g ?g h Th ?h ?h ?h ?h i i "i 5i :i Ri ^i di ji ?i ?i ?i ?i ?j ?j ?j ?j ?j ?j |k }k ?k ?k ?k ?k ?k ?k ?k ?l ?l ?l ?o ?o ?o 0p 8p ?p ?p ?p ?p Cq ???????????????????????????????????????????????????????????????????????? hY? 5?\? h?R? h?TB 6?B*]?ph? hY? h?? h?w} hu
? h?TB 6?]? h?B? h?qH h?TB h?TB 5?\? h?qH 5?\?I}e g 6g ? U P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB X
& F
? ?h?d ?d Eƀ ?+?f ?? [$\$^? `?hgd x? 6g ?g h ?h i ?i ? ? ? ? N P
& F
?d ?d Eƀ ?+?f ?? [$\$gd?TB
& F
?d ?d [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB ?i ?i j ? ] P
& F
?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB j *j Fj ? ] P
& F
?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F
?d ?d Eƀ ?+?f ?? [$\$gd?TB Fj ij ?j ? ] P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB P
& F
?d ?d Eƀ ?+?f ?? [$\$gd?TB ?j ?j ?j ? ] P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB ?j Dk ?l ? ] P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB P
& F
?d ?d Eƀ ?+?f ?? [$\$gd?TB ?l ?l ?m ? ] P
& F
?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB ?m ?m n ? ] P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB n ?n ?n ? ] P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB ?n Lo }o ? ] P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB }o ?o 0p ? ] P
& F
?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB 0p ?p Tq ? ] P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB Cq Qq Tq oq *r -r ?r ?r ?r ?r ?r ?r t t ?t u
u u v v ?v ?v w w pw ?w ?w ?w ?w ?w ?w ?w ?w ?x ?x ?x ?x ?x ?x ?x y y y y Ty Xy _y `y vy zy ?y ?y ?y ?y ?y ?y ?y ?y ?y ?y z z z z /z 3z 8z 9z Xz \z ???????????????????????????????????ÿ÷??÷??÷??÷??÷??÷??÷??÷?? hAXj h?qH 5?h?qH CJ aJ h?qH h?qH 5?\? h?K? h?TB 5?\? hY? 5?\? hAXj h?.? hdpI 5?hdpI hY? h?$6 h?TB h?TB 6?]?ETq ?r `s Qu ?v w ?w ? ] X Q Q Q ?? gd?TB gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB P
& F
?d ?d Eƀ ?+?f o [$\$gd?TB ?w ?w ?w ?w ? k ! I
& F Eƀ ?+?f ?? gd?.? I
& F Eƀ ?+?f ?? gd?.? I
& F Eƀ ?+?f ?? gd?.? ?w ?w ?w ?x ?x ?x ?x ?x ?x ?x ?x ? ? ? ? ? ? ? ? ? ? $If $$If a$ gdAXj I
& F Eƀ ?+?f ?? gd?.?
?x ?x ? kd? $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?x ?x y y y y y y ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If y y ? kd? $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH y y Ty Xy [y ]y ^y _y ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If _y `y ? kdw $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH `y cy vy zy |y }y y ?y ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?y ?y ? kd] $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?y ?y ?y ?y ?y ?y ?y ?y ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?y ?y ? kdC $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?y ?y ?y ?y ?y ?y ?y ?y ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?y ?y ? kd) $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?y ?y z z z z z z ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If z z ? kd $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH z z /z 3z 5z 6z 7z 8z ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If 8z 9z ? kd? $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH 9z { ?{ @{ ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If @{ A{ ? kds $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH A{ D{ h{ l{ o{ p{ q{ r{ ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If r{ s{ ? kdY $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH s{ v{ ?{ ?{ ?{ ?{ ?{ ?{ ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?{ ?{ ? kd? $$If ? ? ֞ ???N??U ?!?? ?? ?? ? ? ?? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?{ ?? ?? ?? ?? Á ǁ ʁ ? ? ? ? ? ? ? ? $$If a$ ?? gd?TB ρ ? kd% $$If ? ? ֞ ??.$>N? 8 ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ρ ҁ ? ? ? ? ? ? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ? ? ? kd? $$If ? ? ֞ ??.$>N? 8 ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ? ? ? kd? $$If ? ? ֞ ??.$>N? 8 ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ? ? '? +? .? /? 0? 1? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If 1? 2? ? kd_ $$If ? ? ֞ ??.$>N? 8 ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH 2? 5? R? V? Y? Z? [? `? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If `? a? ? kd $$If ? ? ֞ ??.$>N? 8 ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH a? d? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?? ?? ? kd? $$If ? ? ֞ ??.$>N? 8 ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ǂ ˂ ͂ ς Ђ ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If Ђ т ? kd?! $$If ? ? ֞ ??.$>N? 8 ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH т Ղ ? ? ?? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?? ?? ? kdW" $$If ? ? ֞ ??.$>N? 8 ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ? ? ? ?? ?? ?? ?? ?? >? F? ?? ل ? ?? ? ? ? ? ? ? ? ?
? +? ,? 3? 4? Q? U? W? [? \? x? |? ?? ?? ?? ?? ?? ?? ˉ ω ۉ ܉ ? ? ? ? 0? 4? :? ;? K? O? T? U? ?? ?? ?? ?? Ҋ ֊ ۊ ܊ ? ? $? %? N? R? _? `? {? ? ?? ?? ?? ?? ?? ??????????????????????????????????????????????????????????????????????????????? h?Y. 5?\? hAXj h?qH 5?h?qH h?TB 5?\? h?K? h?TB 6?]? h?TB h?qH CJ aJ h?qH 5?\?O ? ? ? kd# $$If ? ? ֞ ??.$>N? 8 ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?
? ? ? $? (? ,? /? 3? ? ? ? ? ? ? ? ? $If $$If a$ ?? gd?TB 3? 4? ? kd?# $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH 4? 8? Q? U? X? Y? Z? [? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If [? \? ? kd?$ $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH \? `? x? |? ? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?? ?? ? kdU% $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?? ?? ? kd& $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ˉ ω ԉ Չ ։ ۉ ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ۉ ܉ ? kd?& $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ܉ ߉ ? ? ? ? ? ? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ? ? ? kd?' $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ? ? 0? 4? 6? 7? 9? :? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If :? ;? ? kdM( $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ;? >? K? O? Q? R? S? T? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If T? U? ? kd ) $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH U? X? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?? ?? ? kd?) $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? Ҋ ֊ ؊ ي ڊ ۊ ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ۊ ܊ ? kd?* $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ܊ ?? ? ? !? "? #? $? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If $? %? ? kdE+ $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH %? )? N? R? V? W? X? _? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If _? `? ? kd, $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH `? d? {? ? ?? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?? ?? ? kd?, $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?? ?? ? kd- $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ? ? ? ?
? ? ? ? 7? ;? E? F? a? e? o? p? ?? ?? ?? ?? ?? ?? Ì Č ?? ? ? ? ?? ?? Í ͍ ? ?? ?? ? ?? ?? ?? ?? ?? ?? ?? ?? ?? ͒ Ւ ֒ ? ?? ??
? ? ? -? 3? 4? O? W? X? v? |? }? ?? ?? ?? ?? ?? ?? ? ? ?? ?? ?? ?? ?? ? ? ? ? #? ?????????????????????????????????????????????????????????????????????????????????? h?TB 5?\? h?K? h?TB 6?]? h?TB h?qH 5?\? hAXj h?qH 5?h?qH h?qH CJ aJ R?? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ? ? ? kd=. $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ? ?
? ? ? ? ? ? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ? ? ? kd?. $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ? ? 7? ;? >? ?? @? E? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If E? F? ? kd?/ $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH F? J? a? e? h? i? j? o? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If o? p? ? kdw0 $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH p? s? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If ?? ?? ? kd51 $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ?? ?? ?? ?? ?? Ì ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj $If Ì Č ? kd?1 $$If ? ? ֞ ?? ?i&k ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH Č ?? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? ? $$If a$ ?? gd?TB ?? ?? ? kd?2 $$If ? ? ֞ ??z???L ? ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ɒ ͒ ђ Ӓ Ԓ Ւ ? ? ? ? ? ? ? $$If a$ $If Ւ ֒ ? kdo3 $$If ? ? ֞ ??z???L ? ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ֒ ْ ? ? ? ? ? ?? ? ? ? ? ? ? ? $$If a$ $If ?? ?? ? kd-4 $$If ? ? ֞ ??z???L ? ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ?
? ? ? ? ? ? ? ? ? ? ? ? $$If a$ $If ? ? ? kd?4 $$If ? ? ֞ ??z???L ? ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ? ? )? -? 0? 1? 2? 3? ? ? ? ? ? ? ? $$If a$ $If 3? 4? ? kd?5 $$If ? ? ֞ ??z???L ? ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH 4? 8? K? O? S? U? V? W? ? ? ? ? ? ? ? $$If a$ $If W? X? ? kdg6 $$If ? ? ֞ ??z???L ? ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH X? [? r? v? y? z? {? |? ? ? ? ? ? ? ? $$If a$ $If |? }? ? kd%7 $$If ? ? ֞ ??z???L ? ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH }? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $If ?? ?? ? kd?7 $$If ? ? ֞ ??z???L ? ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? $$If a$ $If ?? ?? ? kd?8 $$If ? ? ֞ ??z???L ? ?!? ? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b? yt?qH ?? ?? ? ? .? 4? 9? =? B? ? ? ? ? ? ? ? ? $$If a$ ?? gd?TB #? ? B? C? l? p? v? w? ?? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? ? +? /? 3? 4? h? l? p? q? ?? ?? ?? ?? ? ? ?? ?? ? ? ? ? M? Q? U? V? f? j? s? t? ?? ?? ?? ?? ?? Ś ɚ ʚ ? ?? ?? ?? ? $? (? )? O? S? X? Y? t? x? }? ~? ?? ?? ?? ?? ?? ś ț ɛ Λ ϛ ? ? ? ? ????????????????????????????????????????????????????????????????????????????????????????? hAXj h?qH 5?h?qH h?qH CJ aJ h?qH 5?\? h?TB YB? C? F? l? + % % $If ? kd_9 $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH l? p? t? u? v? ? ? ? ? $$If a$ $$If a$gdAXj v? w? z? ?? + % % $If ? kd$: $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ?? ?? ?? ?? ?? ? ? ? ? $$If a$ $$If a$gdAXj ?? ?? ?? ?? + % % $If ? kd?: $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ?? ?? ?? ?? ?? ? ? ? ? $$If a$ $$If a$gdAXj ?? ?? ?? ? + % % $If ? kd?; $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ? ? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj ? ? ?? ? + % % $If ? kds< $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ? ?
? ? ? ? ? ? ? $$If a$ $$If a$gdAXj ? ? ? +? + % % $If ? kd8= $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH +? /? 1? 2? 3? ? ? ? ? $$If a$ $$If a$gdAXj 3? 4? :? h? + % % $If ? kd?= $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH h? l? n? o? p? ? ? ? ? $$If a$ $$If a$gdAXj p? q? y? ?? + % % $If ? kd?> $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ?? ?? ?? ?? ?? ? ? ? ? $$If a$ $$If a$gdAXj ?? ?? ?? ? + % % $If ? kd?? $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ? ? ?? ?? ?? ? ? ? ? $$If a$ $$If a$gdAXj ?? ?? ?? ? + % % $If ? kdL@ $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ? ? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj ? ? $? M? + % % $If ? kdA $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH M? Q? S? T? U? ? ? ? ? $$If a$ $$If a$gdAXj U? V? Y? f? + % % $If ? kd?A $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH f? j? m? n? s? ? ? ? ? $$If a$ $$If a$gdAXj s? t? z? ?? + % % $If ? kd?B $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ?? ?? ?? ?? ?? ? ? ? ? $$If a$ $$If a$gdAXj ?? ?? ?? ?? + % % $If ? kd`C $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ?? Ś ǚ Ț ɚ ? ? ? ? $$If a$ $$If a$gdAXj ɚ ʚ К ? + % % $If ? kd%D $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ? ?? ?? ?? ?? ? ? ? ? $$If a$ $$If a$gdAXj ?? ?? ? ? + % % $If ? kd?D $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ? $? &? '? (? ? ? ? ? $$If a$ $$If a$gdAXj (? )? /? O? + % % $If ? kd?E $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH O? S? V? W? X? ? ? ? ? $$If a$ $$If a$gdAXj X? Y? \? t? + % % $If ? kdtF $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH t? x? z? |? }? ? ? ? ? $$If a$ $$If a$gdAXj }? ~? ?? ?? + % % $If ? kd9G $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ?? ?? ?? ?? ?? ? ? ? ? $$If a$ $$If a$gdAXj ?? ?? ?? ś + % % $If ? kd?G $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ś ɛ ̛ ͛ Λ ? ? ? ? $$If a$ $$If a$gdAXj Λ ϛ ֛ ? + % % $If ? kd?H $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ? ? ? ? ? ? ? ? ? $$If a$ $$If a$gdAXj ?
? ? #? + % % $If ? kd?I $$If ? ? ֈ ???5 ? ?!?? ? ? ? ? ? ?0 ? ? ? ? ? ? ??6? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?3? 4? a? b?yt?qH ?
? #? &? '? +? ,? E? H? I? N? O? j? m? n? w? x? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ? ?
? ? ?? ?? ̝ ԝ ?? ?? T? W? ?? ?? &? *? x? y? z? ~? ? ?? k? ?? ?? ?? ?? ?? Ǣ Ȣ Ԣ ע آ ? ? ? ? ? ? $? %? ? ?? ?? ? a Y M
& F ?d [$gd?
& F gd? M
& F ?d Eƀ ?+?f . [$gd? P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB ?? X? ?? ? ? ? ` M
& F ?d Eƀ ?+?f ?? \$gd? gd?TB M
& F ?d Eƀ ?+?f . \$gd? ? U? u? ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB u? ?? K? ?? ? ? ` M
& F ?d Eƀ ?+?f ?? \$gd? gd?TB M
& F ?d Eƀ ?+?f ?? [$gd? ?? ?? ?? ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ?? ?? ? ? ] P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB ? ?? ?? ?? ?? ?? ?? ? ] X S N I gd?TB gd? gdJjT gd? P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB P
& F ?d ?d Eƀ ?+?f ?? [$\$gd?TB #? e? f? ? ?? ?? ?? ?? ?? ?? ?? `? a? r? s? ?? ?? ?? ?? ? ? N? O? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ? ? )? *? :? ;? P? h? ??????????????????????????????????????????ϚϚ??ρ h?? he"? 6?CJ aJ h?? he"? 0J CJ aJ j h?? he"? CJ UaJ h x? 0J ]?mH nH u h?? he"? 0J ]? j h?? he"? 0J U]? h?? he"? CJ aJ he"? h?` j h?` Uhz h?TB j h?TB U
h?TB 0J 2?? h? ?? ? ] P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB ?? ?? ? N? ? ] X gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB P
& F ?d ?d Eƀ ?+?f . [$\$gd?TB N? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ? @ A B C D E F G H I J K L M N O P ????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????? s 2 ? ? ? ? & 6 F V f v ? ? ? ? ? ? 2 ( ? ? & 6 F V f v ? ? ? ? ? ? & 6 F V f v ? ? ? ? ? ? & 6 F V f v ? ? ? ? ? ? & 6 F V f v ? ? ? ? ? ? & 6 F V f v ? ? ? ? ? ? & 6 F V f v ? ? 8 X ? V ~ ? ? ? ? ? ? ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ ? ? 0 @ _HmH nH sH tH @ `?? @ N o r m a l CJ _HaJ mH sH tH Z @ Z gE H e a d i n g 1 $?? ?< @& 5?CJ KH OJ QJ \?^J aJ N @ " N ?TB H e a d i n g 2 ?d ?d @&[$\$ 5?CJ$ \?aJ$ N @ 2 N ?TB H e a d i n g 3 ?d ?d @&[$\$ 5?CJ \?aJ D A`??? D D e f a u l t P a r a g r a p h F o n t R i@??? R T a b l e N o r m a l ? 4?
l 4? a? ( k ??? ( N o L i s t 4 U`??? 4 ?TB H y p e r l i n k >*ph ? D V ??D ?TB F o l l o w e d H y p e r l i n k >*ph ? B ^@ B ?TB N o r m a l ( W e b ) ?d ?d [$\$ N g`??!N ?TB H T M L T y p e w r i t e r CJ OJ PJ QJ ^J aJ : `@ 2: ?TB H T M L A d d r e s s 6?]?4 B4 ?TB H e a d e r
? ??! 4 @ R4 ?TB F o o t e r
? ??! j ? ? cj n?
T a b l e G r i d 7 :V ?0 ? ? ? ? ? ? . )@? q. n? P a g e N u m b e r H ? ?H ?]? B a l l o o n T e x t CJ OJ QJ ^J aJ R ?o???R ?K? H e a d i n g 3 C h a r 5CJ \_HaJ mH sH tH PK ! ???? [Content_Types].xml???N?0E?H???-J??@%?ǎǢ|?ș$?ز?U??L?TB? l,?3??;?r? ?Ø??J??B+$?G]??7O٭V??vn??B`2ǃ??,?!"?E3?p#9??G??? Qd????; H?
xuv? ? ?? ???? 0F[,?F᚜K sO?'3 w ?#?????vfS?Vbs?ؠy??X? ??p?5v??euw 1?z@? ?l,???i!b
I? ?jZ??2??|?9???????L$??Z??15?x??l.?(?z???m${?d??:\@?'?2?3?ln$?^-??@^?i????D&? ?|?#?td? !??6l?ġ??? B?"?&6?3?yy@??t?!?H??jp?U??*ye?Xry3?~{s:FXI
????O?5Y??[Y!}??S˪.??7b????d|n]???????6??71?.?
?t??n?/????w/?+???[?t??6???}??P?s???ںs?L?. J;?̊??i??N? ?$A??I??????)?t2? ??L?m????x:??(}\-?i* ??x????QC??Ju???Wl'?QyI@ھ
m2?D?B?? ?AR?4 ??w¢naQ??`Բ???Ɂ
W??=0#x?B?d??T/????.3?-?F?>bY?L?%˓?KK? ?6Hh?f?P?Q=??h??)G?B?ms]_?Ԡ'C???ZѨ????ys
v??@c])h? 7?Jه??i? ?c? FS? .???N???P$?
e?&\?Ӏ+?I? "'%QÕ???@c?!?[??pa??AV.9H??d< ??Ӯ??H????VX?*??%?A?{????Yr A??բ ??px?SL?9"?:3?U??5?????U
??NC??(???p%? u??@;[???d??`???4?)??]?t#9M4W=?P?5?*f̰lk?<_??X-C ??w?T??%??Ժ?}B?% ?Y?,]?
A???̠&oʰ??Ũ?;? \lc`?|,b?U?v? ?? PK !
ѐ?? ' theme/theme/_rels/themeManager.xml.rels??M
?0???wooӺ?&݈Э???5
6?$Q??
?,.?a??i????c2?1h?:?q??m??@RN??;d?`??o7?g?K(M&$R(.1?r'J??ЊT???8??V?"??AȻ?H?u}??|?$?b{? ?P????8?g/ ]?QAsم(????#??L?[???? ?? PK- ! ???? [Content_Types].xmlPK- ! ?֧?? 6 0 _rels/.relsPK- ! ky?? ? theme/theme/themeManager.xmlPK- ! ?R%?? ? ? theme/theme/theme1.xmlPK- !
ѐ?? ' ? theme/theme/_rels/themeManager.xml.relsPK ] ?
© 2015 - 2025 Weber Informatics LLC | Privacy Policy