Skip to content

Latest commit

 

History

History
181 lines (152 loc) · 14.1 KB

regressions-hc4-neuclir22-fa-en.md

File metadata and controls

181 lines (152 loc) · 14.1 KB

Anserini Regressions: HC4 (v1.0) on translated NeuCLIR22 — Persian

This page documents BM25 regression experiments for HC4 (v1.0) Persian topics on the NeuCLIR22 translated Persian corpus. The HC4 qrels have been filtered down to include only those in the intersection of the HC4 and NeuCLIR22 corpora. To be clear, the queries are in English and the corpus is in English (automatically translated by the organizers using Sockeye).

The exact configurations for these regressions are stored in this YAML file. Note that this page is automatically generated from this template as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead.

From one of our Waterloo servers (e.g., orca), the following command will perform the complete regression, end to end:

python src/main/python/run_regression.py --index --verify --search --regression hc4-neuclir22-fa-en

Corpus Download

The HC4 corpus can be downloaded following the instructions here.

After download, verify that all and only specified documents have been downloaded by running the code provided here.

With the corpus downloaded, unpack into collections/ and run the following command to perform the remaining steps below:

python src/main/python/run_regression.py --index --verify --search --regression hc4-neuclir22-fa-en \
  --corpus-path collections/neuclir22-fa-en

Indexing

Typical indexing command:

bin/run.sh io.anserini.index.IndexCollection \
  -threads 8 \
  -collection NeuClirCollection \
  -input /path/to/neuclir22-fa-en \
  -generator DefaultLuceneDocumentGenerator \
  -index indexes/lucene-index.neuclir22-fa-en \
  -storePositions -storeDocvectors -storeRaw \
  >& logs/log.neuclir22-fa-en &

See this page for more details about the HC4 corpus. For additional details, see explanation of common indexing options.

Retrieval

After indexing has completed, you should be able to perform retrieval as follows:

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.neuclir22-fa-en \
  -topics tools/topics-and-qrels/topics.hc4-v1.0-fa.en.test.title.tsv \
  -topicReader TsvInt \
  -output runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.title.txt \
  -bm25 &
bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.neuclir22-fa-en \
  -topics tools/topics-and-qrels/topics.hc4-v1.0-fa.en.test.desc.tsv \
  -topicReader TsvInt \
  -output runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.txt \
  -bm25 &
bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.neuclir22-fa-en \
  -topics tools/topics-and-qrels/topics.hc4-v1.0-fa.en.test.desc.title.tsv \
  -topicReader TsvInt \
  -output runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.title.txt \
  -bm25 &

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.neuclir22-fa-en \
  -topics tools/topics-and-qrels/topics.hc4-v1.0-fa.en.test.title.tsv \
  -topicReader TsvInt \
  -output runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.title.txt \
  -bm25 -rm3 &
bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.neuclir22-fa-en \
  -topics tools/topics-and-qrels/topics.hc4-v1.0-fa.en.test.desc.tsv \
  -topicReader TsvInt \
  -output runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.txt \
  -bm25 -rm3 &
bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.neuclir22-fa-en \
  -topics tools/topics-and-qrels/topics.hc4-v1.0-fa.en.test.desc.title.tsv \
  -topicReader TsvInt \
  -output runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.title.txt \
  -bm25 -rm3 &

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.neuclir22-fa-en \
  -topics tools/topics-and-qrels/topics.hc4-v1.0-fa.en.test.title.tsv \
  -topicReader TsvInt \
  -output runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.title.txt \
  -bm25 -rocchio &
bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.neuclir22-fa-en \
  -topics tools/topics-and-qrels/topics.hc4-v1.0-fa.en.test.desc.tsv \
  -topicReader TsvInt \
  -output runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.txt \
  -bm25 -rocchio &
bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.neuclir22-fa-en \
  -topics tools/topics-and-qrels/topics.hc4-v1.0-fa.en.test.desc.title.tsv \
  -topicReader TsvInt \
  -output runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.title.txt \
  -bm25 -rocchio &

Evaluation can be performed using trec_eval:

bin/trec_eval -c -m ndcg_cut.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.title.txt
python -m pyserini.eval.trec_eval -c -m judged.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.title.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.title.txt
bin/trec_eval -c -m map tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.title.txt
bin/trec_eval -c -m ndcg_cut.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.txt
python -m pyserini.eval.trec_eval -c -m judged.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.txt
bin/trec_eval -c -m map tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.txt
bin/trec_eval -c -m ndcg_cut.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.title.txt
python -m pyserini.eval.trec_eval -c -m judged.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.title.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.title.txt
bin/trec_eval -c -m map tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default.topics.hc4-v1.0-fa.en.test.desc.title.txt

bin/trec_eval -c -m ndcg_cut.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.title.txt
python -m pyserini.eval.trec_eval -c -m judged.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.title.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.title.txt
bin/trec_eval -c -m map tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.title.txt
bin/trec_eval -c -m ndcg_cut.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.txt
python -m pyserini.eval.trec_eval -c -m judged.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.txt
bin/trec_eval -c -m map tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.txt
bin/trec_eval -c -m ndcg_cut.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.title.txt
python -m pyserini.eval.trec_eval -c -m judged.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.title.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.title.txt
bin/trec_eval -c -m map tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rm3.topics.hc4-v1.0-fa.en.test.desc.title.txt

bin/trec_eval -c -m ndcg_cut.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.title.txt
python -m pyserini.eval.trec_eval -c -m judged.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.title.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.title.txt
bin/trec_eval -c -m map tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.title.txt
bin/trec_eval -c -m ndcg_cut.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.txt
python -m pyserini.eval.trec_eval -c -m judged.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.txt
bin/trec_eval -c -m map tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.txt
bin/trec_eval -c -m ndcg_cut.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.title.txt
python -m pyserini.eval.trec_eval -c -m judged.20 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.title.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.title.txt
bin/trec_eval -c -m map tools/topics-and-qrels/qrels.hc4-neuclir22-fa.test.txt runs/run.neuclir22-fa-en.bm25-default+rocchio.topics.hc4-v1.0-fa.en.test.desc.title.txt

Effectiveness

With the above commands, you should be able to reproduce the following results:

MAP BM25 (default) +RM3 +Rocchio
HC4 (Persian): test-topic title 0.0933 0.1052 0.1110
HC4 (Persian): test-topic description 0.0973 0.0968 0.1064
HC4 (Persian): test-topic description+title 0.1062 0.1082 0.1163
nDCG@20 BM25 (default) +RM3 +Rocchio
HC4 (Persian): test-topic title 0.1607 0.1594 0.1665
HC4 (Persian): test-topic description 0.1538 0.1459 0.1640
HC4 (Persian): test-topic description+title 0.1746 0.1688 0.1770
J@20 BM25 (default) +RM3 +Rocchio
HC4 (Persian): test-topic title 0.1350 0.1550 0.1510
HC4 (Persian): test-topic description 0.1110 0.1250 0.1270
HC4 (Persian): test-topic description+title 0.1370 0.1470 0.1470
Recall@1000 BM25 (default) +RM3 +Rocchio
HC4 (Persian): test-topic title 0.7671 0.7978 0.7979
HC4 (Persian): test-topic description 0.7087 0.7635 0.7638
HC4 (Persian): test-topic description+title 0.7616 0.8180 0.8140

The above results reproduce the BM25 title queries run in Table 2 of this paper.

Reproduction Log*

To add to this reproduction log, modify this template and run bin/build.sh to rebuild the documentation.