Artificial Intelligence Course 4th Project: Implementing Bigram and Unigram models for filtering comments.
In this group project we (pouyanhessabi and omidmahyar) implemented Bigram and Unigram models to filter comments.

We trained these models on these positive and negative datasets. We also used smoothing in both models (you can change coefficients). For preprocessing first we removed punctuation marks and we also have a cut_down parameter which specifies that words with equal or less number of repetition to this parameter should be removed. Also there is a cut_above parameter that specifies that how many of most repeated words should be removed.

A sample run:

Check full description here

Project report (in persian): here
tried different coefficients and tried the models with and without cut_down and cut_above and checked the results

Check our other AI Course projects:

• Project 1: Searching Algorithms (IDS, BBFS, A*)
• Project 2: Genetic Algorithm