regression.neuclir22-fa-dt-splade.yaml Maven / Gradle / Ivy

Go to download
---
corpus: neuclir22-fa-en-splade
corpus_path: collections/multilingual/neuclir22-fa-en-splade/

index_path: indexes/lucene-index.neuclir22-fa-en-splade
collection_class: JsonVectorCollection
generator_class: DefaultLuceneDocumentGenerator
index_threads: 8
index_options: -impact -pretokenized -storeRaw
index_stats:
  documents: 2232016
  documents (non-empty): 2232016

metrics:
  - metric: MAP
    command: bin/trec_eval
    params: -c -m map
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: true
  - metric: nDCG@20
    command: bin/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: bin/trec_eval
    params: -c -m recall.1000
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false

topic_reader: TsvInt
topics:
  - name: "[NeuCLIR 2022 (Persian): title (original English queries)](https://neuclir.github.io/)"
    id: title
    path: topics.neuclir22-en.splade.original-title.txt.gz
    qrel: qrels.neuclir22-fa.txt
  - name: "[NeuCLIR 2022 (Persian): desc (original English queries)](https://neuclir.github.io/)"
    id: desc
    path: topics.neuclir22-en.splade.original-desc.txt.gz
    qrel: qrels.neuclir22-fa.txt
  - name: "[NeuCLIR 2022 (Persian): desc+title (original English queries)](https://neuclir.github.io/)"
    id: desc_title
    path: topics.neuclir22-en.splade.original-desc_title.txt.gz
    qrel: qrels.neuclir22-fa.txt

models:
  - name: splade
    display: SPLADE
    params: -impact -pretokenized
    results:
      MAP:
        - 0.2977
        - 0.3057
        - 0.3203
      nDCG@20:
        - 0.4627
        - 0.4618
        - 0.4802
      J@20:
        - 0.3768
        - 0.3882
        - 0.3917
      Recall@1000:
        - 0.8478
        - 0.8796
        - 0.8860
  - name: splade+rm3
    display: +RM3
    params: -impact -pretokenized -rm3 -collection JsonVectorCollection
    results:
      MAP:
        - 0.2800
        - 0.2956
        - 0.3002
      nDCG@20:
        - 0.4258
        - 0.4480
        - 0.4477
      J@20:
        - 0.3496
        - 0.3504
        - 0.3610
      Recall@1000:
        - 0.8018
        - 0.8061
        - 0.7948
  - name: splade+rocchio
    display: +Rocchio
    params: -impact -pretokenized -rocchio -collection JsonVectorCollection
    results:
      MAP:
        - 0.2835
        - 0.3275
        - 0.3267
      nDCG@20:
        - 0.4438
        - 0.4675
        - 0.4645
      J@20:
        - 0.3781
        - 0.3860
        - 0.3908
      Recall@1000:
        - 0.8592
        - 0.8735
        - 0.8703