regression.ntcir8-zh.yaml Maven / Gradle / Ivy

Go to download
---
corpus: ntcir8-zh
corpus_path: collections/newswire/clir/ntcir.zh/ntcir8-zh/

index_path: indexes/lucene-index.ntcir8-zh/
collection_class: CleanTrecCollection
generator_class: DefaultLuceneDocumentGenerator
index_threads: 16
index_options: -storePositions -storeDocvectors -storeRaw -language zh -uniqueDocid -optimize
# we need -optimize because without, "documents (non-empty)" would be non-deterministic, failing regression.
index_stats:
  documents: 308832
  documents (non-empty): 308832
  total terms: 134618634  # on Mac, I seem to get this value instead: 134618027 (no idea why)

metrics:
  - metric: MAP
    command: tools/eval/trec_eval.9.0.4/trec_eval
    params: -m map
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: true
  - metric: P20
    command: tools/eval/trec_eval.9.0.4/trec_eval
    params: -m P.20
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: true
  - metric: nDCG@20
    command: tools/eval/trec_eval.9.0.4/trec_eval
    params: -m ndcg_cut.20
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: true

topic_reader: TsvString
topic_root: src/main/resources/topics-and-qrels/
qrels_root: src/main/resources/topics-and-qrels/
topics:
  - name: "[NTCIR-8 ACLIA (IR4QA subtask, Monolingual Chinese)](../src/main/resources/topics-and-qrels/topics.ntcir8zh.eval.txt)"
    id: ntcir8
    path: topics.ntcir8zh.eval.txt
    qrel: qrels.ntcir8.eval.txt

models:
  - name: bm25
    display: BM25
    params: -bm25 -language zh
    results:
      MAP:
        - 0.4014
      P20:
        - 0.3849
      nDCG@20:
        - 0.4757