rules.latin.xml Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of webservices Show documentation
Webservice API for Tēzaurs.lv and other ailab.lv Latvian computational linguistic tools
There is a newer version: 2.5.7
<?xml version="1.0" encoding="UTF-8"?>
<rules>
	<exact>
		<!-- 1. "ee" transliteration near the begin of the token: -->
		<!-- 1.1. prefix "ie" combined with stem starting with "ie" -->
		<!-- 1.1.1. in combination with prefix "vis" -->
		<r target="visneeeee">
			<replace position="begin">visneieie</replace>
		</r>
		<r target="visjāeeee">
			<replace position="begin">visjāieie</replace>
		</r>
		<r target="visneee">
			<replace position="begin">visneie</replace>
		</r>
		<r target="visjāee">
			<replace position="begin">visjāie</replace>
		</r>
		<r target="viseeee">
			<replace position="begin">visieie</replace>
		</r>
		<r target="visee">
			<replace position="begin">visie</replace>
		</r>
		
		<!-- 1.1.2. combinations without prefix "vis" -->
		<r target="neeeee">
			<replace position="begin">neieie</replace>
		</r>
		<r target="jāeeee">
			<replace position="begin">jāieie</replace>
		</r>
		<r target="neee">
			<replace position="begin">neie</replace>
		</r>
		<r target="jāee">
			<replace position="begin">jāie</replace>
		</r>
		<r target="eeee">
			<replace position="begin">ieie</replace>
		</r>
		<r target="ee">
			<replace position="begin">ie</replace>
		</r>
		
		<!-- 1.2. other prefixes combined with prefix "ie" -->
		<!-- 1.2.1. in combination with prefix "vis" -->
		<r target="visaizee">
			<replace position="begin">visaizie</replace>
		</r>
		<r target="visapee">
			<replace position="begin">visapie</replace>
		</r>
		<r target="visatee">
			<replace position="begin">visatie</replace>
		</r>
		<r target="visbezee">
			<replace position="begin">visbezie</replace>
		</r>
		<r target="viseksee">
			<replace position="begin">viseksie</replace>
		</r>
		<r target="visizee">
			<replace position="begin">visizie</replace>
		</r>
		<r target="visnoee">
			<replace position="begin">visnoie</replace>
		</r>
		<r target="vispaee">
			<replace position="begin">vispaie</replace>
		</r>
		<r target="vispāree">
			<replace position="begin">vispārie</replace>
		</r>
		<r target="visparee">
			<replace position="begin">visparie</replace>
			<!-- Why this was comented out in periodika I? -->
		</r>
		<r target="vispeeee">
			<replace position="begin">vispieie</replace>
		</r>
		<r target="vissaee">
			<replace position="begin">vissaie</replace>
		</r>
		<r target="visuzee">
			<replace position="begin">visuzie</replace>
		</r>
		
		<!-- 1.2.2. combinations without prefix "vis" -->
		<r target="aizee">
			<replace position="begin">aizie</replace>
		</r>
		<r target="apee">
			<replace position="begin">apie</replace>
		</r>
		<r target="atee">
			<replace position="begin">atie</replace>
		</r>
		<r target="bezee">
			<replace position="begin">bezie</replace>
		</r>
		<r target="eksee">
			<replace position="begin">eksie</replace>
		</r>
		<r target="izee">
			<replace position="begin">izie</replace>
		</r>
		<r target="noee">
			<replace position="begin">noie</replace>
		</r>
		<r target="paee">
			<replace position="begin">paie</replace>
		</r>
		<r target="pāree">
			<replace position="begin">pārie</replace>
		</r>
		<r target="paree">
			<replace position="begin">parie</replace>
			<!-- Why this was comented out in periodika I? -->
		</r>
		<r target="peeee">
			<replace position="begin">pieie</replace>
		</r>
		<r target="saee">
			<replace position="begin">saie</replace>
		</r>
		<r target="uzee">
			<replace position="begin">uzie</replace>
		</r>
		
		<!-- 1.3. other prefixes containing "ie" -->
		<r target="vispee">
			<replace position="begin">vispie</replace>
		</r>
		<r target="pee">
			<replace position="begin">pie</replace>
		</r>
		
		<!-- 2. Transliteration for the prefixes "iz", "uz", "aiz" in front of s: -->
		<!-- 2.1. in combination with prefix "vis" -->
		<r target="visneiss">
			<replace position="begin">visneizs</replace>
		</r>
		<r target="visjāiss">
			<replace position="begin">visjāizs</replace>
		</r>
		<r target="visiss">
			<replace position="begin">visizs</replace>
		</r>
		
		<r target="visneuss">
			<replace position="begin">visneuzs</replace>
		</r>
		<r target="visjāuss">
			<replace position="begin">visjāuzs</replace>
		</r>
		<r target="visuss">
			<replace position="begin">visuzs</replace>
		</r>
		
		<r target="visneaiss">
			<replace position="begin">visneaizs</replace>
		</r>
		<r target="visjāaiss">
			<replace position="begin">visjāaizs</replace>
		</r>
		<r target="visaiss">
			<replace position="begin">visaizs</replace>
		</r>
		
		<!-- 2.2. combinations without prefix "vis" -->
		<r target="neiss">
			<replace position="begin">neizs</replace>
		</r>
		<r target="jāiss">
			<replace position="begin">jāizs</replace>
		</r>
		<r target="iss">
			<replace position="begin">izs</replace>
			<!-- In periodika I restricted to words longer than 4 characters. -->
		</r>
		
		<r target="neuss">
			<replace position="begin">neuzs</replace>
		</r>
		<r target="jāuss">
			<replace position="begin">jāuzs</replace>
		</r>
		<r target="uss">
			<replace position="begin">uzs</replace>
			<!-- In periodika I restricted to words longer than 4 characters. -->
		</r>

		<r target="neaiss">
			<replace position="begin">neaizs</replace>
		</r>
		<r target="jāaiss">
			<replace position="begin">jāaizs</replace>
		</r>
		<r target="aiss">
			<replace position="begin">aizs</replace>
		</r>

		<!-- 3. "ee" transliteration near the end of the token:-->
		<r target="ee">
			<replace position="end">ie</replace>
		</r>
		<r target="ees">
			<replace position="end">ies</replace>
		</r>
		<r target="eem">
			<replace position="end">iem</replace>
		</r>
		
		<!--4. other endings: -->
		
		<!-- 5. transliterations resulting in 1 letter: -->
		<!-- 5.1. cosonants -->
		<r target="ch">h</r>
		<r target="w">v</r>
		<r target="ŗ">r</r>
		<r target="ş">s</r>
		<r target="q">g</r>
		
		
		<!-- 5.2. vowels -->
		<r target="à">ā</r>
		<r target="á">ā</r>
		<r target="â">ā</r>
		<r target="ã">ā</r>
		<r target="ä">ā</r>

		<r target="è">ē</r>
		<r target="é">ē</r>
		<r target="ê">ē</r>
		<r target="ë">ē</r>
		
		<r target="ì">ī</r>
		<r target="í">ī</r>
		<r target="î">ī</r>
		<r target="ï">ī</r>
		
		<r target="ō">o</r>
		<r target="ò">o</r>
		<r target="ó">o</r>
		<r target="ô">o</r>
		<r target="õ">o</r>

		<r target="ù">ū</r>
		<r target="ú">ū</r>
		<r target="û">ū</r>
		<r target="ü">ū</r>
		
		<!-- 6. rules for specific words. -->
		<r target="vaj">
			<replace position="exact">vai</replace>
		</r>
		<r target="nau">
			<replace position="exact">nav</replace>
		</r>
		<r target="jav">
			<replace position="exact">jau</replace>
		</r>
		<r target="vel">
			<replace position="exact">vēl</replace>
		</r>
		
		<!-- Rules covering common OCR mistakes. -->
		<r target="X" sensitive="1">K</r>
		
	</exact>
	<fuzzy>
		<!-- Rules covering orthography differences. -->
		<r target="ee">ie</r>
		<r target="i">ī</r>
		<r target="e">ē</r>
		<r target="a">ā</r>
		<r target="u">ū</r>
		
		<r target="g">ģ</r>
		<r target="s">š</r>
		<r target="š">s</r>
		<r target="z">ž</r>

		<r target="j" position="end">i</r>

		<r target="use" position="end">usi</r>
		
		<!-- Rules covering common OCR mistakes. -->
		<r target="c">č</r>
		<r target="k">ķ</r>
		<r target="ķ">k</r>
		<r target="l">ļ</r>
		<r target="n">ņ</r>

		<r target="c">e</r>
		<r target="ce">ie</r>
		<r target="ec">ie</r>
		<r target="cc">ie</r>
		
		<r target="Gr" sensitive="1">G</r>
		
		<!-- Rules for specific words. -->
		
	</fuzzy>
</rules>