Preview

PDF, English Download (5MB) | Terms of use

Abstract

From the recent proliferation of social media channels to the immense amount of user-generated content, an increasing interest in social media mining is currently being witnessed. Messages continuously posted via these channels report a broad range of topics from daily life to global and local events. As a consequence, this has opened new opportunities for mining event information crucial in many application domains, especially in increasing the situational awareness in critical scenarios. Interestingly, many of these messages are enriched with location information, due to the wide- spread of mobile devices and the recent advancements of today’s location acquisition techniques. This enables location-aware event mining, i.e., the detection and tracking of localized events.

In this thesis, we propose novel frameworks and models that digest social media content for localized event detection, tracking, and recommendation. We first develop KeyPicker, a framework to extract and score event-related keywords in an online fashion, accounting for high levels of noise, temporal heterogeneity and outliers in the data. Then, LocEvent is proposed to incrementally detect and track events using a 4-stage procedure. That is, LocEvent receives the keywords extracted by KeyPicker, identifies local keywords, spatially clusters them, and finally scores the generated clusters. For each detected event, a set of descriptive keywords, a location, and a time interval are estimated at a fine-grained resolution. In addition to the sparsity of geo-tagged messages, people sometimes post about events far away from an event’s location. Such spatial problems are handled by novel spatial regularization techniques, namely, graph- and gazetteer-based regularization. To ensure scalability, we utilize a hierarchical spatial index in addition to a multi-stage filtering procedure that gradually suppresses noisy words and considers only event-related ones for complex spatial computations.

As for recommendation applications, we propose an event recommender system built upon model-based collaborative filtering. Our model is able to suggest events to users, taking into account a number of contextual features including the social links between users, the topical similarities of events, and the spatio-temporal proximity between users and events. To realize this model, we employ and adapt matrix factorization, which allows for uncovering latent user-event patterns. Our proposed features contribute to directing the learning process towards recommendations that better suit the taste of users, in particular when new users have very sparse (or even no) event attendance history.

To evaluate the effectiveness and efficiency of our proposed approaches, extensive comparative experiments are conducted using datasets collected from social media channels. Our analysis of the experimental results reveals the superiority and advantages of our frameworks over existing methods in terms of the relevancy and precision of the obtained results.