How to use indexes

From ATIRE

Jump to: navigation, search

Contents

How to use indexes

Assumed:

As an example, most of these commands will use parameter input specific to the TREC 1 collection.

TREC 1 collection example

  • 741,856 documents
  • Queries/Topics: 51-100

Generate index

 ~/atire/bin$ ./index -N10000 -rtrec ~/collections/TREC_1_COLLECTION

This will produce an index.aspt in your current working directory.

Explanation:

  • -N10000 the indexer will print time and memory information every N documents (0 -- never)
  • -rtrec <file> specifies a single file TREC collection

You can check this explanation by running the help command:

 ~/atire/bin$ ./index -?

Basic search

If the index and doclist are in your current directory:

 ~/atire/bin$ ./atire -QN:t -q ~/topics/topics.51-100 -a ~/queries/qrels.51-100 -l0

Note: the last parameter (-l0) is "ell zero".

</p> Otherwise:

 ~/atire/bin$ ./atire -findex ~/indexes/index.aspt -QN:t -q ~/topics/topics.51-100 -a ~/queries/qrels.51-100 -l0

</p>

Explanation:

  • -l0 "ell-zero" is an option which suppresses the results list: i.e, "less output"
  • -findex <file> specifies an index file that is not in the working directory
  • -QN: specifies that we are using a TREC query file, to be parsed by ATIRE. The value given is any combination of
    • t: titles
    • d: description
    • n: narrative
  • -q <file> specifies the query file (think "questions")
  • -a <file> specifies the relevance-judgements file (think "answers")

Remember, you can run ATIRE with the help command at any time to see a list of all options:

 ~/atire/bin$ ./atire -?