Skip to content

Latest commit

 

History

History
123 lines (92 loc) · 6.66 KB

regressions-core18.md

File metadata and controls

123 lines (92 loc) · 6.66 KB
Raw

Anserini Regressions: TREC 2018 Common Core Track

Models: various bag-of-words approaches

This page describes regressions for the TREC 2018 Common Core Track, which uses the TREC Washington Post Corpus. 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 core18

Indexing

Typical indexing command:

bin/run.sh io.anserini.index.IndexCollection \
  -threads 1 \
  -collection WashingtonPostCollection \
  -input /path/to/wapo.v2 \
  -generator WashingtonPostGenerator \
  -index indexes/lucene-index.wapo.v2/ \
  -storePositions -storeDocvectors -storeRaw \
  >& logs/log.wapo.v2 &

The directory /path/to/core18/ should be the root directory of the TREC Washington Post Corpus, i.e., ls /path/to/core18/ should bring up a single JSON file.

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. They are downloaded from NIST:

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

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.wapo.v2/ \
  -topics tools/topics-and-qrels/topics.core18.txt \
  -topicReader Trec \
  -output runs/run.wapo.v2.bm25.topics.core18.txt \
  -bm25 &

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.wapo.v2/ \
  -topics tools/topics-and-qrels/topics.core18.txt \
  -topicReader Trec \
  -output runs/run.wapo.v2.bm25+rm3.topics.core18.txt \
  -bm25 -rm3 &

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.wapo.v2/ \
  -topics tools/topics-and-qrels/topics.core18.txt \
  -topicReader Trec \
  -output runs/run.wapo.v2.bm25+ax.topics.core18.txt \
  -bm25 -axiom -axiom.deterministic -rerankCutoff 20 &

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.wapo.v2/ \
  -topics tools/topics-and-qrels/topics.core18.txt \
  -topicReader Trec \
  -output runs/run.wapo.v2.ql.topics.core18.txt \
  -qld &

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.wapo.v2/ \
  -topics tools/topics-and-qrels/topics.core18.txt \
  -topicReader Trec \
  -output runs/run.wapo.v2.ql+rm3.topics.core18.txt \
  -qld -rm3 &

bin/run.sh io.anserini.search.SearchCollection \
  -index indexes/lucene-index.wapo.v2/ \
  -topics tools/topics-and-qrels/topics.core18.txt \
  -topicReader Trec \
  -output runs/run.wapo.v2.ql+ax.topics.core18.txt \
  -qld -axiom -axiom.deterministic -rerankCutoff 20 &

Evaluation can be performed using trec_eval:

bin/trec_eval -m map -m P.30 tools/topics-and-qrels/qrels.core18.txt runs/run.wapo.v2.bm25.topics.core18.txt

bin/trec_eval -m map -m P.30 tools/topics-and-qrels/qrels.core18.txt runs/run.wapo.v2.bm25+rm3.topics.core18.txt

bin/trec_eval -m map -m P.30 tools/topics-and-qrels/qrels.core18.txt runs/run.wapo.v2.bm25+ax.topics.core18.txt

bin/trec_eval -m map -m P.30 tools/topics-and-qrels/qrels.core18.txt runs/run.wapo.v2.ql.topics.core18.txt

bin/trec_eval -m map -m P.30 tools/topics-and-qrels/qrels.core18.txt runs/run.wapo.v2.ql+rm3.topics.core18.txt

bin/trec_eval -m map -m P.30 tools/topics-and-qrels/qrels.core18.txt runs/run.wapo.v2.ql+ax.topics.core18.txt

Effectiveness

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

MAP BM25 +RM3 +Ax QL +RM3 +Ax
TREC 2018 Common Core Track Topics 0.2496 0.3129 0.2840 0.2527 0.3077 0.2920
P30 BM25 +RM3 +Ax QL +RM3 +Ax
TREC 2018 Common Core Track Topics 0.3573 0.4167 0.3947 0.3653 0.4007 0.4013

Reproduction Log*