lucene-oxygen

Simple project in IR

Abstract

The focus of this project is the Non-factoid Question-Answering task. Within such a setting, a user may post questions in natural language and answers should be informative enough to satisfy the users information need

List of command-line options for this program

usage: java -jar yourJarName.jar [[option1] [option2]] 
 -i,--index              Reindex documents
 -l,--lambda             Create appropriate lambda for language model,
                         depends on median size of document in corpus
 -q,--create-questions   Create questions file from corpus json file

Reindexing option

-i,--index              Reindex documents

Without this option index will not be created

Automatic calculation of lambda for Jelinek-Mercer Smoothing

 -l,--lambda             Create appropriate lambda for language model,
                         depends on median size of document in corpus

Lambda parameter calculated by this formula:

Create questions option

 -q,--create-questions   Create questions file from corpus json file

This option needed for lambda parameter calculation, in case that there is no corpus of questions available. This requires regular .json file with corpus collection.

Source documentation

Can be found here https://louiscyphre.github.io/lucene-oxygen/

Additionally, it can be found here (same files) https://github.com/louiscyphre/lucene-oxygen/tree/master/docs