regression.hc4-v1.0-ru.yaml Maven / Gradle / Ivy

Go to download
---
corpus: hc4-v1.0-ru
corpus_path: collections/multilingual/hc4-v1.0-ru/

index_path: indexes/lucene-index.hc4-v1.0-ru
collection_class: NeuClirCollection
generator_class: DefaultLuceneDocumentGenerator
index_threads: 8
index_options: -storePositions -storeDocvectors -storeRaw -language ru
index_stats:
  documents: 4721064
  documents (non-empty): 4721055

metrics:
  - metric: MAP
    command: tools/eval/trec_eval.9.0.4/trec_eval
    params:  -c -m map
    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: -c -m ndcg_cut.20
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: J@20
    command: python -m pyserini.eval.trec_eval
    params: -c -m judged.20
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: Recall@1000
    command: tools/eval/trec_eval.9.0.4/trec_eval
    params: -c -m recall.1000
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false

topic_reader: TsvInt
topic_root: src/main/resources/topics-and-qrels/
qrels_root: src/main/resources/topics-and-qrels/
topics:
  - name: "[HC4 (Russian): dev-topic title](https://github.com/hltcoe/HC4)"
    id: dev_title
    path: topics.hc4-v1.0-ru.dev.title.tsv
    qrel: qrels.hc4-v1.0-ru.dev.txt
  - name: "[HC4 (Russian): dev-topic description](https://github.com/hltcoe/HC4)"
    id: dev_description
    path: topics.hc4-v1.0-ru.dev.desc.tsv
    qrel: qrels.hc4-v1.0-ru.dev.txt
  - name: "[HC4 (Russian): dev-topic description+title](https://github.com/hltcoe/HC4)"
    id: dev_description_title
    path: topics.hc4-v1.0-ru.dev.desc.title.tsv
    qrel: qrels.hc4-v1.0-ru.dev.txt
  - name: "[HC4 (Russian): test-topic title](https://github.com/hltcoe/HC4)"
    id: test_title
    path: topics.hc4-v1.0-ru.test.title.tsv
    qrel: qrels.hc4-v1.0-ru.test.txt
  - name: "[HC4 (Russian): test-topic description](https://github.com/hltcoe/HC4)"
    id: test_description
    path: topics.hc4-v1.0-ru.test.desc.tsv
    qrel: qrels.hc4-v1.0-ru.test.txt
  - name: "[HC4 (Russian): test-topic description+title](https://github.com/hltcoe/HC4)"
    id: test_description_title
    path: topics.hc4-v1.0-ru.test.desc.title.tsv
    qrel: qrels.hc4-v1.0-ru.test.txt


models:
  - name: bm25-default
    display: BM25 (default)
    params: -bm25 -language ru
    results:
      MAP:
        - 0.2937
        - 0.2373
        - 0.3186
        - 0.2186
        - 0.1883
        - 0.2265
      nDCG@20:
        - 0.3942
        - 0.2580
        - 0.3972
        - 0.2944
        - 0.2456
        - 0.2989
      J@20:
        - 0.4375
        - 0.5125
        - 0.5000
        - 0.3470
        - 0.3180
        - 0.3670
      Recall@1000:
        - 0.8432
        - 0.5942
        - 0.7619
        - 0.7182
        - 0.7355
        - 0.7721
  - name: bm25-default+rm3
    display: +RM3
    params: -bm25 -rm3 -language ru
    results:
      MAP:
        - 0.2389
        - 0.0842
        - 0.2163
        - 0.2375
        - 0.1896
        - 0.2317
      nDCG@20:
        - 0.3376
        - 0.1838
        - 0.3410
        - 0.3209
        - 0.2461
        - 0.3007
      J@20:
        - 0.4500
        - 0.3625
        - 0.4625
        - 0.3630
        - 0.2950
        - 0.3530
      Recall@1000:
        - 0.7598
        - 0.3707
        - 0.6428
        - 0.7208
        - 0.6530
        - 0.7335
  - name: bm25-default+rocchio
    display: +Rocchio
    params: -bm25 -rocchio -language ru
    results:
      MAP:
        - 0.3995
        - 0.2817
        - 0.3564
        - 0.2641
        - 0.2250
        - 0.2732
      nDCG@20:
        - 0.4719
        - 0.3168
        - 0.4400
        - 0.3163
        - 0.2767
        - 0.3265
      J@20:
        - 0.5125
        - 0.5500
        - 0.5875
        - 0.3930
        - 0.3510
        - 0.3990
      Recall@1000:
        - 0.8710
        - 0.6171
        - 0.7639
        - 0.7728
        - 0.7680
        - 0.8271