Skip to content

A simple lsa engine written in python3

License

Notifications You must be signed in to change notification settings

odarbelaeze/condor-ir

Repository files navigation

condor-ir

https://travis-ci.org/odarbelaeze/condor-ir.svg?branch=master

This is a program to work with examples of Latent Semantic Analysis search engines, a.k.a., LSA. The program is set up so that it understands froac xml documents on input as well as plain text records from isi web of knowledge.

You can find more information about froac repositories at http://froac.manizales.unal.edu.co/froac/ http://froac.manizales.unal.edu.co/froac/ and about isi web of knowledge text files at the thomson reuters website

Installing the condor-ir package

The second thing you will need is to download the program from its pypi repository,

pip install -U condor-ir

the -U parameter will upgrade the package to the latest version, a very recommendable step for a unstable package.

For specific databases support you can install their appropriate extra package:

pip install -U condor-ir[mysql]
pip install -U condor-ir[postgres]

Furthermore, we require a bit of the nltk data package for the stems and stop word removal to work.

python -m nltk.downloader snowball_data stopwords

Finally, in order to prepare the database or reset the database in preparation for a new version of condor-ir you can run the database preparation script,

condor utils preparedb

If you need to specify a database other than the default you can do so through environment variables:

export CONDOR_DB_URL="mysql://localhost/condor"
condor utils preparedb # will now work on mysql://localhost/condor

CLI Interface

After installing the program you will have three basic commands at your disposal, for handling bibliography sets, term document matrices and engines, the CLI interface gives you most CRUD operations in a hierachical manner.

condor triggers the main program and you can get top level help by running condor --help.

condor bibliography namespaces the bibliography set related commands, you can list and get help about those using condor bibliography --help.

condor model is a short cut that offers the condor model create sub command, that creates both a term document matrix and an lsa search engine, get help on models using condor model --help.

condor query <string...> this non crud command search a bibliography set using a previously created search engine, the search engine can be targeted figure out how using condor query --help.

Feel free to check detailed descriptions of these commands using their --help flag.