Skip to content

Latest commit

 

History

History
183 lines (139 loc) · 10.3 KB

regressions-msmarco-v1-doc-segmented.md

File metadata and controls

183 lines (139 loc) · 10.3 KB

Anserini Regressions: MS MARCO Document Ranking

Models: various bag-of-words approaches on segmented documents

This page documents regression experiments on the MS MARCO document ranking task, which is integrated into Anserini's regression testing framework. Note that there are four different bag-of-words regression conditions for this task, and this page describes the following:

  • Indexing Condition: each MS MARCO document is first segmented into passages, each passage is treated as a unit of indexing
  • Expansion Condition: none

All four conditions are described in detail here, in the context of doc2query-T5. In the passage (i.e., segment) indexing condition, we select the score of the highest-scoring passage from a document as the score for that document to produce a document ranking; this is known as the MaxP technique.

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.

Note that in November 2021 we discovered issues in our regression tests, documented here. As a result, we have had to rebuild all our regressions from the raw corpus. These new versions yield end-to-end scores that are slightly different, so if numbers reported in a paper do not exactly match the numbers here, this may be the reason.

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 msmarco-v1-doc-segmented

Indexing

Typical indexing command:

bin/run.sh io.anserini.index.IndexCollection \
  -threads 16 \
  -collection JsonCollection \
  -input /path/to/msmarco-doc-segmented \
  -generator DefaultLuceneDocumentGenerator \
  -index indexes/lucene-inverted.msmarco-v1-doc-segmented/ \
  -storePositions -storeDocvectors -storeRaw \
  >& logs/log.msmarco-doc-segmented &

The directory /path/to/msmarco-doc-segmented/ should be a directory containing the segmented corpus in Anserini's jsonl format. See this page for how to prepare the corpus.

For additional details, see explanation of common indexing options.

Retrieval

Topics and qrels are stored here, which is linked to the Anserini repo as a submodule. The regression experiments here evaluate on the 5193 dev set questions.

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

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-inverted.msmarco-v1-doc-segmented/ \
  -topics tools/topics-and-qrels/topics.msmarco-doc.dev.txt \
  -topicReader TsvInt \
  -output runs/run.msmarco-doc-segmented.bm25-default.topics.msmarco-doc.dev.txt \
  -bm25 -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000 &

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-inverted.msmarco-v1-doc-segmented/ \
  -topics tools/topics-and-qrels/topics.msmarco-doc.dev.txt \
  -topicReader TsvInt \
  -output runs/run.msmarco-doc-segmented.bm25-tuned.topics.msmarco-doc.dev.txt \
  -bm25 -bm25.k1 2.16 -bm25.b 0.61 -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000 &

Evaluation can be performed using trec_eval:

bin/trec_eval -c -m map tools/topics-and-qrels/qrels.msmarco-doc.dev.txt runs/run.msmarco-doc-segmented.bm25-default.topics.msmarco-doc.dev.txt
bin/trec_eval -c -M 100 -m recip_rank tools/topics-and-qrels/qrels.msmarco-doc.dev.txt runs/run.msmarco-doc-segmented.bm25-default.topics.msmarco-doc.dev.txt
bin/trec_eval -c -m recall.100 tools/topics-and-qrels/qrels.msmarco-doc.dev.txt runs/run.msmarco-doc-segmented.bm25-default.topics.msmarco-doc.dev.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.msmarco-doc.dev.txt runs/run.msmarco-doc-segmented.bm25-default.topics.msmarco-doc.dev.txt

bin/trec_eval -c -m map tools/topics-and-qrels/qrels.msmarco-doc.dev.txt runs/run.msmarco-doc-segmented.bm25-tuned.topics.msmarco-doc.dev.txt
bin/trec_eval -c -M 100 -m recip_rank tools/topics-and-qrels/qrels.msmarco-doc.dev.txt runs/run.msmarco-doc-segmented.bm25-tuned.topics.msmarco-doc.dev.txt
bin/trec_eval -c -m recall.100 tools/topics-and-qrels/qrels.msmarco-doc.dev.txt runs/run.msmarco-doc-segmented.bm25-tuned.topics.msmarco-doc.dev.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.msmarco-doc.dev.txt runs/run.msmarco-doc-segmented.bm25-tuned.topics.msmarco-doc.dev.txt

Effectiveness

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

AP@1000 BM25 (default) BM25 (tuned)
MS MARCO Doc: Dev 0.2690 0.2762
RR@100 BM25 (default) BM25 (tuned)
MS MARCO Doc: Dev 0.2684 0.2756
R@100 BM25 (default) BM25 (tuned)
MS MARCO Doc: Dev 0.7847 0.8013
R@1000 BM25 (default) BM25 (tuned)
MS MARCO Doc: Dev 0.9178 0.9311

Explanation of settings:

  • The setting "default" refers the default BM25 settings of k1=0.9, b=0.4.
  • The setting "tuned" refers to k1=2.16, b=0.61, tuned in 2020/12 using the MS MARCO document sparse judgments to optimize for recall@100 (i.e., for first-stage retrieval).

In these runs, we are retrieving the top 1000 hits for each query and using trec_eval to evaluate all 1000 hits. Since we're in the passage condition, we fetch the 10000 passages and select the top 1000 documents using MaxP. This lets us measure R@100 and R@1000; the latter is particularly important when these runs are used as first-stage retrieval. Beware, an official MS MARCO document ranking task leaderboard submission comprises only 100 hits per query. See this page for details on Anserini baseline runs that were submitted to the official leaderboard.

The MaxP passage retrieval functionality is available in SearchCollection. To generate an MS MARCO submission with the BM25 default parameters, corresponding to "BM25 (default)" above:

$ target/appassembler/bin/SearchCollection -topicreader TsvString \
   -topics tools/topics-and-qrels/topics.msmarco-doc.dev.txt \
   -index indexes/lucene-index.msmarco-doc-segmented/ \
   -output runs/run.msmarco-doc-segmented.bm25-default.txt -format msmarco \
   -bm25 -bm25.k1 0.9 -bm25.b 0.4 -hits 1000 \
   -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 100

$ python tools/scripts/msmarco/msmarco_doc_eval.py \
   --judgments tools/topics-and-qrels/qrels.msmarco-doc.dev.txt \
   --run runs/run.msmarco-doc-segmented.bm25-default.txt

#####################
MRR @100: 0.2682349308946578
QueriesRanked: 5193
#####################

This run was not submitted to the MS MARCO document ranking leaderboard, but is reported in the Lin et al. (SIGIR 2021) Pyserini paper. Note that the above command uses -format msmarco to directly generate a run in the MS MARCO output format.

To generate an MS MARCO submission with the BM25 tuned parameters, corresponding to "BM25 (tuned)" above:

$ target/appassembler/bin/SearchCollection -topicreader TsvString \
   -topics tools/topics-and-qrels/topics.msmarco-doc.dev.txt \
   -index indexes/lucene-index.msmarco-doc-segmented/ \
   -output runs/run.msmarco-doc-segmented.bm25-tuned.txt -format msmarco \
   -bm25 -bm25.k1 2.16 -bm25.b 0.61 -hits 1000 \
   -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 100

$ python tools/scripts/msmarco/msmarco_doc_eval.py \
   --judgments tools/topics-and-qrels/qrels.msmarco-doc.dev.txt \
   --run runs/run.msmarco-doc-segmented.bm25-tuned.txt

#####################
MRR @100: 0.2751202109946902
QueriesRanked: 5193
#####################

This run corresponds to the MS MARCO document ranking leaderboard entry "Anserini's BM25 (per passage), parameters tuned for recall@100 (k1=2.16, b=0.61)" dated 2021/01/20, and is reported in the Lin et al. (SIGIR 2021) Pyserini paper. Again, note that the above command uses -format msmarco to directly generate a run in the MS MARCO output format.

As of February 2022, following resolution of #1721, BM25 runs for the MS MARCO leaderboard can be generated with the same commands as above. However, the effectiveness has changed slightly, since we corrected underlying issues with data preparation.

For default parameters (k1=0.9, b=0.4):

$ python tools/scripts/msmarco/msmarco_doc_eval.py \
   --judgments tools/topics-and-qrels/qrels.msmarco-doc.dev.txt \
   --run runs/run.msmarco-doc-segmented.bm25-default.txt

#####################
MRR @100: 0.26851990908986706
QueriesRanked: 5193
#####################

For tuned parameters (k1=2.16, b=0.61):

$ python tools/scripts/msmarco/msmarco_doc_eval.py \
   --judgments tools/topics-and-qrels/qrels.msmarco-doc.dev.txt \
   --run runs/run.msmarco-doc-segmented.bm25-tuned.txt

#####################
MRR @100: 0.27551963417683756
QueriesRanked: 5193
#####################