Access the docs here: home.condor-ir.co
Access to roadmap here: roadmap.
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
-U parameter will upgrade the package to the latest version, a very
recommendable step for a unstable package.
The language support requires the enchant engine as well as some dictionaries, for that you can install using your package manager or external tool:
# Arch sudo pacman -S enchant \ aspell-es aspell-en aspell-fr \ aspell-it aspell-pt aspell-de
# Ubuntu sudo apt-get install enchant \ aspell-es aspell-en aspell-fr \ aspell-it aspell-pt aspell-de
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
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 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