BroDyn '18: Workshop on Analysis of Broad Dynamic Topics over Social Media

In recent years, users developed a widespread perception of social media as news sources that they follow (almost all day long) to get updated on their topics of interest. Many of those topics attract long-standing user interest (i.e., span or stay active for long period of time), are very broad (i.e., cover many aspects or subtopics), and are dynamic (i.e., develop and change focus over time with subtopics becoming obsolete over time and new subtopics emerge). Such kind of broad and dynamic topics can run for few weeks, such as crisis events (e.g., “Hurricane Irma”), or up to years, such as “the Syrian conflict”. Other examples of broad dynamic topics include “Refugees in Europe”, “Qatar Crisis”, “Brexit”, “North Korea and US conflict” to name a few.

Effective exploitation of content posted on social media that is relevant to broad dynamic topics requires adaptive techniques to effectively capture the different and changing aspects of the topics. The techniques should also be scalable to cope with the large volume of posts and the diversity of the sub-topics, real-time to be responsive to the high velocity of the data, and reliable to perform effectively over the long duration of the topic. In addition to that, new evaluation frameworks for such domain are also needed with novel evaluation measures that capture the nature of topics (and thus systems) and new large reusable datasets that enable the researchers to run meaningful and representative experiments.

The workshop aims to bring together audience at all levels, including researchers from academia and industry as well as potential users, to create a forum for discussing recent advances in this emerging area. The theme of the workshop concerns, besides IR and social computing researchers, a wide spectrum of potential users of the needed technology, such as journalists, historians, politicians, social scientists and analysts.

The BroDyn workshop aims to achieve the following objectives:

1. Directing the attention of IR researchers to the new emerging type of information needs that require following broad and dynamic topics and events over social media streams.
2. Better understanding of the limitations of the current methods and inspiring research on new algorithms and techniques.
3. Encouraging the design of new evaluation measures, datasets, and test collections to support the research in that area.
4. Forming a new community that brings together IR and social computing researchers as well as potential users who are interested in the domain.