Information overload is a recent phenomenon caused by a regular use of social media platforms among millions of users. Websites such as Twitter seem to be getting increasingly popular, providing a perfect platform for sharing information which can help in the process of modelling users and recommender system research. This research studies information overload and uses twitter user modelling through making use of explicit relationships amongst various users. This paper presents a novel personal profile mechanism that helps in the provision of more accurate recommendations by filtering overloaded information as it gathered from Twitter data. The presented method takes advantage of user explicit relationships on Twitter based on influence rule in order to gain information which is vital in the building of the personal profile of the user. In order to validate this proposed method's usefulness a simple tweet recommendation service was implemented by using content-based recommender system. This has also been evaluated using an offline evaluation process. Our proposed user profiles are compared against other profiles such as the baseline in order to have the proposed method's effectiveness checked. The experiment is implemented based on an experimental number of users.
Alshammari, Abdullah; Polatidis, Nikolaos; Kapetanakis, Stelios; Evans, Roger; and Alshammari, Gharbi, "Personalized Recommendations on Twitter based on Explicit User Relationship Modelling" (2019). UK Academy for Information Systems Conference Proceedings 2019. 10.