info: | A perfect repo for your college project on data mining - link prediction - friends recommender. |
---|
pip install networkx
To run and understand this code first reach ___EntryPoint___ in python code .
- 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
- 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
Following Scoring Methods are used to construct feature Set
- common_neighbors
- resource_allocation_index
- jaccard_coefficient
- adamic_adar_index
- preferential_attachment
SVM ANN and Logistic Regresssion is used for classificaion