Harnessing Social Media Streams for Local Information Needs
Dyaa Albakour
School of Computing Science, Glasgow University
Local search is increasingly attracting more demand, whereby the users are interested to find out about places or events in their local vicinity. In this talk, we address a number of emerging Information Retrieval (IR) tasks that involve using social media streams to serve local information needs of the citizens. These include (i) local event retrieval, (ii) social media filtering and (iii) contextual suggestion.
First, we present a novel event retrieval framework that identifies and ranks events within a city from social media posts [1]. The framework exploits the content of the posts and the volume of the activity and has been evaluated on local tweets and events within London. Second, we consider how the occurrence of such events may impose the challenge of topic drift on the task of real-time tweet filtering [2]. To tackle topic drift, we develop a model that balances between the importance of the short-term interests of the user, i.e. emerging events, and their long-term interests. For that purpose, we investigate an event detection method from Twitter and newswire streams to predict times at which drift may happen. The last part of the talk addresses yet another local information retrieval task, namely contextual suggestions of places to visit in the city. Using rich contextual data about venues in location-based social networks, such as Foursquare, we develop novel models for contextual suggestions [3]. For all the above models, we summarise the evaluation results and the insights we draw.
References:
[1] M-Dyaa Albakour, Craig Macdonald, Iadh Ounis. Identifying Local Events by Using Microblogs as Social Sensors. In proceedings of OAIR 2013, Lisbon, Portugal.
[2] M-Dyaa Albakour, Craig Macdonald, Iadh Ounis. On Sparsity and Drift for Effective Real-time Filtering in Microblogs. In proceedings of CIKM 2013, San Francisco, CA, USA.
[3] The SMART FP7 Consortium. D5.3.2 Query Scoring and Anticipation Subsystem, 2014.