Similarity Learning on Graph (SLOG) contains implementation of similarity learning methods
over relational data, where the relation between data points are given explicitly. 

The methods in the package divide into the following categories: 1) Link-based or
relational based only methods 2) Latent space model 3) Joint similarity learning on multi
relational data 4) Iterative ranking on relational data 5) Multi object similarity
learning 6) structured latent space models.

The usage of each function is shown in the SampleCode.m . 

Edges are represented by a matrix n*3 that n is number of Edges and each row is:
e1 e2 w

where e1 and e2 are the index of the first node and second node respectively. w is the
weight of the edge .

Nodes' features are shown by a sparse matrix X, m*d, in which m is number of objects and
d is number of features.


To read more please have a look at:

-Computing Text Semantic Relatedness using the Contents and Links of a Hypertext Encyclopedia, Majid Yazdani and Andrei Popescu-Belis, in: Artificial Intelligence Journal, 194:176–202, 2013
https://dx.doi.org/10.1016/j.artint.2012.06.004

-Learning to Rank on Network Data, Majid Yazdani, Ronan Collobert and Andrei Popescu-Belis, in: Mining and Learning with Graphs, 2013
https://snap.stanford.edu/mlg2013/submissions/mlg2013_submission_26.pdf

-Similarity Learning Over Large Collaborative Networks, Majid Yazdani, Thèse École polytechnique fédérale de Lausanne EPFL, n° 5696 (2013)
https://dx.doi.org/10.5075/epfl-thesis-5696

Please feel free to contact me at [email protected]