nefelitav/recommender-systems

Small programs performing tasks related to recommendation systems, pagerank and binary classification.

Recommendation Systems

python recommendation.py

We tried all prediction algorithms offered by the Surprise library, (Basic, k-NN, Matrix Factorization) and then we compared different configurations for KNNBaseline and SVD.

Pagerank

python pagerank.py

We created a graph of Pokemon and computed the Topic-Specific Pagerank for a different topic each time, using the Networkx library. Also, we proved that this procedure builds teams not simply by aggregating teams generated from individual nodes.

Communities

python communities.py

Using the same Pokemon graph, we found the local communities around them, using Personalized Pagerank.

Embeddings

python embeddings.py

Get word embeddings from text , train and test two models and then classify claims as REFUTES or SUPPORTS.