Skip to content

Latest commit

 

History

History
67 lines (44 loc) · 4.32 KB

regressions-backgroundlinking18.md

File metadata and controls

67 lines (44 loc) · 4.32 KB
Raw

Anserini: Regressions for TREC 2018 Background Linking

This page describes regressions for the background linking task in the TREC 2018 News Track. 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.

Indexing

Typical indexing command:

nohup sh target/appassembler/bin/IndexCollection -collection WashingtonPostCollection -input /path/to/backgroundlinking18 \
 -index lucene-index.backgroundlinking18.pos+docvectors+rawdocs -generator WashingtonPostGenerator -threads 1 \
 -storePositions -storeDocvectors -storeRaw >& log.backgroundlinking18.pos+docvectors+rawdocs &

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 in src/main/resources/topics-and-qrels/, downloaded from NIST:

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

nohup target/appassembler/bin/SearchCollection -index lucene-index.backgroundlinking18.pos+docvectors+rawdocs \
 -topicreader BackgroundLinking -topics src/main/resources/topics-and-qrels/topics.backgroundlinking18.txt \
 -backgroundlinking -backgroundlinking.k 100 -bm25 -hits 100 -output run.backgroundlinking18.bm25.topics.backgroundlinking18.txt &

nohup target/appassembler/bin/SearchCollection -index lucene-index.backgroundlinking18.pos+docvectors+rawdocs \
 -topicreader BackgroundLinking -topics src/main/resources/topics-and-qrels/topics.backgroundlinking18.txt \
 -backgroundlinking -backgroundlinking.k 100 -bm25 -rm3 -hits 100 -output run.backgroundlinking18.bm25+rm3.topics.backgroundlinking18.txt &

nohup target/appassembler/bin/SearchCollection -index lucene-index.backgroundlinking18.pos+docvectors+rawdocs \
 -topicreader BackgroundLinking -topics src/main/resources/topics-and-qrels/topics.backgroundlinking18.txt \
 -backgroundlinking -backgroundlinking.datefilter -backgroundlinking.k 100 -bm25 -rm3 -hits 100 -output run.backgroundlinking18.bm25+rm3+df.topics.backgroundlinking18.txt &

Evaluation can be performed using trec_eval:

eval/trec_eval.9.0.4/trec_eval -c -M1000 -m ndcg_cut.5 -c -M1000 -m map src/main/resources/topics-and-qrels/qrels.backgroundlinking18.txt run.backgroundlinking18.bm25.topics.backgroundlinking18.txt

eval/trec_eval.9.0.4/trec_eval -c -M1000 -m ndcg_cut.5 -c -M1000 -m map src/main/resources/topics-and-qrels/qrels.backgroundlinking18.txt run.backgroundlinking18.bm25+rm3.topics.backgroundlinking18.txt

eval/trec_eval.9.0.4/trec_eval -c -M1000 -m ndcg_cut.5 -c -M1000 -m map src/main/resources/topics-and-qrels/qrels.backgroundlinking18.txt run.backgroundlinking18.bm25+rm3+df.topics.backgroundlinking18.txt

Effectiveness

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

NCDG@5 BM25 +RM3 +RM3+DF
TREC 2018 Topics 0.3293 0.3526 0.4171
AP BM25 +RM3 +RM3+DF
TREC 2018 Topics 0.2490 0.2642 0.2692