iamsinghrajat / friends-recommender-in-social-network

Friends Recommendation and Link Prediction in Social Netowork

Link Prediction in Social Netowork

info:	A perfect repo for your college project on data mining - link prediction - friends recommender.

How To Use

1. pip install networkx

2. To run and understand this code first reach ___EntryPoint___ in python code .

3. • facebook_combine.txt is dataset.
   • Facebook dataset is included download others from http://snap.stanford.edu/data/ . Use one with format like twitter_combine.txt or gplus_combine.txt .
   • At entry point you can choose which dataset to use you can even add your own dataset from http://snap.stanford.edu/data/ . Code will work for all dataset.
   • By default it is set to Facebook Data Set
4. • sample_positive.txt: all positive friend relation ie. people with connecting edge in graph
   • sample_negative.txt: all negative friend relation ie. people with no connecting edge in graph
   • training.txt: combination of positive and negative to get a dataset of relations
   • features_combined_2.txt: for each relation features like AdamicAdar are calculated and then used for training model

5. Following Scoring Methods are used to construct feature Set

   • common_neighbors
   • resource_allocation_index
   • jaccard_coefficient
   • adamic_adar_index
   • preferential_attachment

6. SVM ANN and Logistic Regresssion is used for classificaion