This readme only includes altered or new folders and files in the ReQue. To run this method, clone the ReQue and execute the ReQue based on its readme document. After installing ReQue follow these steps:
Add following folders to the ReQue:
Source Folder:
ds/trec2009/source: this folder Contains TREC2009 Million Query data set, the relevance judgment and query classes (query types).
ds/trec2009/lucene-index-trec2009mq: this folder includes clueweb indexes.
+---ds
| +---trec2009mq
| | +---lucene-index-trec2009mq
| | +---source
| | | | 09.mq.topics.20001-60000
| | | | prels.20001-60000
| | | | queryclasses
Codebase:
Execute following files respectively:
- preprocess_trec2009/preprocess_trec2009.py
- qe/main.py
- stat/stat.py
preprocess_trec2009/: Running preprocess_trec2009.py would results in preprocessed input data set and relevance judgment. The out put would be placed in ds/trec2009/output.
+---preprocess_trec2009
| | preprocess_trec2009.py
qe/cmn/param.py: to run the ReQue for the Trec2009mq, change the folder address for antique in param.py files to the following address:
'index': '../ds/trec2009mq/lucene-index-trec2009mq',
'topics': '../ds/trec2009mq/output/topics_trec',
'qrels':'../ds/trec2009mq/output/qrels_trec',
stat/stat.py: by running this files matrix consist of expanders and query type and also heat map of the matrix would be built.
+---stat
| | stat.py
Target Folders: The target folders are the outputs for the preprocessed and stat and ReQue expanders.
ds/trec2009/output: the results of preprocessing of input queries and the relevance judgment would be in this folder after running preprocess_trec2009.py.
+---ds
| +---trec2009mq
| | +---output
| | | | topics_trec
| | | | qrels_trec
qe/output/trec2009mq: this folder includes ReQue results in txt files and also the stat results in png format.
+---qe
| +---output
| | +---trec2009mq
| | | | topics.antique.glove.topn3.bm25.txt
| | | | topics.antique.bm25.map.dataset.csv
| | | | .....
| | +---stat
| | | | heatmap.png
| | | | matrix_result.csv