COMP4133 Information Retrieval Project

A simple search engine written in C++.

How to build

Build on Linux

Install gcc and g++. Clone the project from Github. make

##Build on Windows Install MinGW (mingw32-base, mingw32-gcc-g++, msys-base). Clone the project from Github. Run msys. Move to the project directory. make

Testing

This program has been tested on Fedora 24.

Run

Place post1.txt into data/data/ directory.

Usage: comp4133 [-b | -n | -v] [-tf <tf function>] 
                [-idf <idf function>] [-nf <nf function> <slope>]
                queryfile
Options:

Select retrieval model
   -b        Boolean model
   -n        Boolean model(NLP) - Require file produced by NLP model
   -v        Vector space model(VSM)

Select tf function (for VSM)
   -tf 1     Binary
   -tf 2     Raw frequency
   -tf 3     Log normalization
   -tf 4     Double normalization 0.5
   -tf 5     Double normalization 0.4 0.6 (default)

Select idf function (for VSM)
   -idf 1     Unary
   -idf 2     Inverse document frequency
   -idf 3     Inverse document frequency smooth
   -idf 4     Inverse document frequency max
   -idf 5     probabilistic inverse document frequency

Select normalization function (for VSM)
   -nf 1            Cosine Similarity - vector lengths (default)
   -nf 2 <slope>    Pivoted cosine normalization
   -nf 3 <slope>    Pivoted unique normalization