Knowledge Discovery in Heterogeneous Information Networks
School of Science and Technology, MSc in Information & Communication Technology Systems
This dissertation was written as a part of the MSc in ICT Systems at the International Hellenic University. Graph mining techniques were utilized for modelling and discovering knowledge from various Information Networks.
Most Information Networks are usually assumed to be homogeneous where nodes are objects of the same entity and links represent relationships between them. This work deals with heterogeneous networks which seem to represent more ideally the real world networks where nodes and relations are consisted of different typed entities.
During the implementation, the first goal is the collection and selection of appropriate data and their modelling as a multi-type network. Next, the graph structure which was created in the first step will be mined for producing knowledge. Mining goals will focus on: a) Standard graph metrics (centrality, closeness, etc.) and b) utilization of hybrid algorithms for mining tasks such as ranking and clustering.
OrientDB is used as a DBMS for achieving our goals, which offers a native Java API, implementing algorithms to perform the modelling and analysis tasks for reaching our experimental analysis results.