Master Thesis: Conference Recommendations Based on Social Recommender Systems
Conference Recommendations Based on Social Recommender Systems
Regarding academic literature, there is a vast amount of information readily available, e.g. about authors, titles, content of papers, keywords, etc. Furthermore, there is available data for relations between different elements, e.g. co-authorship, citation, conference attendance, etc. Information about past conferences and published papers is available through portals like ACM Digital Library or IEEE Explorer. Websites like ResearchGate or Mendeley create services based on such data. In the case of ResearchGate, the focus lies on the connection with other researchers in an online social network fashion. With Mendeley, users can organize their references and PDF files. Recommendations for papers are made based on the data a user enters. What is missing so far is recommendations for upcoming conferences, or, expressed differently, a prediction who will attend an upcoming conference. WikiCfP offers information about upcoming conferences.
In this thesis, Natural Language Processing tools can be employed to extract relevant data from online libraries like ACM DL, IEEE Explorer, or WikiCfP in order to create an ontology. With this ontology, different social relations between users can be expressed, e.g. based on the colleagues, co-authorships or citations. The result is the construction of different social graphs. On this graphs should a social recommendation algorithm be applied, which means that the information of people closer in the social graph are weighted higher. The result of this thesis is an evaluation, which social graph best predicts the conference attendance of a researcher.
The goal of this thesis is to create a graph of the academic community in order for a system to estimate the relevance of an upcoming conference for a user.
This comprises the following tasks:
- Provide an overview on the state of the art of social recommender systems
- Conduct a detailed requirement analysis for conference recommendation and select an appropriate social recommendation approach
- Implement a prototype that imports conference information from several sources and create multiple social graphs based on different relationships
- Apply the social recommendation algorithm on all graphs
- Evaluate the results
by doing a comparison of the outcomes
Prerequisites: - Programming experience
- Preferably: experience with semantic web technologies and recommender systems
Supervisors: Felix Beierle, Kai Grunert 
Type: Master Thesis
Duration: 6 months
10587 Berlin, Germany
Phone: +49 30 8353 58811
Fax: +49 30 8353 58409
e-mail query