Eirini Spyropoulou

A visual representation of some MCCS patterns. (Click to enlarge).

(N-)RMiner is the result of the research I did during my PhD, which was part of this project.


RMiner is our proposed algorithm for mining interesting patterns from multi-relational data. Our proposed pattern syntax is called Maximal Complete Connected Subsets (MCCSs) and is a generalisation of itemsets.

The pictures on the right hand side show a visualisation of two MCCS patterns that resulted from running RMiner on the MovieLens dataset.

Code:

You can download the code of RMiner here. The code is available under the NPOSL license.

The related publication is:

  • E. Spyropoulou, T. De Bie and M. Boley. Interesting Pattern Mining in Multi-Relational Data. Data Mining & Knowledge Discovery, 28(3), pp.808-849, 2014.


N-RMiner is a generalisation of RMiner for multi-relational data containing relationships of any arity (instead of just binary).

Code:

You can download the code of N-RMiner here. The code is available under the NPOSL license.

The related publication is:

  • E. Spyropoulou, T. De Bie and M. Boley. Mining Interesting Patterns in Multi-Relational Data with N-ary Relationships. Proceedings of the International Conference on Discovery Science, pp. 217-232, 2013.