Instructions

Goal: build your own TF-IDF/cosine based search_engine.

1. have a look at the cisi files (see below)
2. read the provided code in search_engine.py
3. complete the code (wherever there is '???') in search_engine.py, from top to bottom. Test at each step: python3 search_engine.py -c cisi_collection.txt -q cisi_queries.txt -o run1
4. run the evaluation program (see below)

CISI collection

cisi_collection.txt : The CISI collection, i.e. one file in which documents are concatenated

<- Balise de début de document / starting tag of a document doc_42 <- identifiant du document / id of the document

<title>18 Editions of the Dewey Decimal Classifications</title> <- Titre / Title The present study is a history of the DEWEY Decimal <- Résumé / Abstract Classification. The first edition of the DDC was published in 1876, the eighteenth edition in 1971, and future editions will continue to appear as needed. In spite of the DDC's long and healthy life, however, its full story has never been told. There have been biographies of Dewey that briefly describe his system, but this is the first attempt to provide a detailed history of the work that more than any other has spurred the growth of librarianship in this country and abroad.

The query file has an almost similar format.

The cisi.rel file is the groun-truth used for the evaluation.

Stop_word list

stop_words.en.txt

evaluation.pl

evaluation programme for TREC formatted input (relevance file and result file)

Once downloaded, put execution rights chmod u+x evaluation.pl

To use it:

evaluation.pl -results my_result_file.res -relevance cisi.rel

Result file format:

num_query 'Q0' id_document rank score 'Exp'

For example: 1 Q0 doc_38 1 23.5217647552 Exp
1 Q0 doc_52 2 23.1733417511 Exp
1 Q0 doc_53 3 23.036365509 Exp
1 Q0 doc_76 4 22.7806720734 Exp
...