Previous MSc Theses

"Developing Recommendation Systems for Movies Using Graph Database Clustering." D. Moorhead. C. Kupke. Department of Computer and Information Sciences, University of Strathclyde. 2019. Download PDF (BibTeX) ACSBD

Abstract:
Thispapercoveredtheideaofdevelopingrecommendationsystemsfromclusteringgraphdatabases. Graph databases are a form of storing data where each piece of information is stored as a node. Each node is connected by an edge that is defined by some relationship in the data between the nodes. This paper investigated two methods of representing movie data in a graph database sing the TMDb 5000 Movie database. The first was by representing each node as either an actor, a movie, a director or a genre. The relationships were between each non-movie node and the director/genre/actor node associated with it. The second graph contained only movie nodes that were connected by shared actors, directors, genres and keywords. Clustering algorithms were tested on these graphs to see if the clusters found were suitable to be used for a recommendation system where a user is associated a cluster that contains movies they have enjoyed. Other movies in this cluster would then be recommended to the user by the system. The clustering algorithms investigate where Connected Components, Strongly Connected Components, Louvain Modularity and Label Propagation The graph databases were successfully constructed, however none of the algorithms were able to produce clusters that could be used for a recommendation system.