Data Mining: Text Mining, Visualization and Social Media

AAAI: Spatial and Temporal Reasoning

On Sunday, I attended the workshop on spatial and temporal reasoning. My intuition is that social media analysis requires technologies from this field of research both at the object and meta data levels. At the object level, much of our attention is give to discussing and reporting events in the real world. At the meta level, social media is about the relationships between individuals.

One of the most interesting challenges in social media analysis (for me) is figuring out not just what people are talking about, but where those people are. Is the iPhone getting more anticipatory buzz in Germany or France? A good number of individuals provide explicit information about their location in the profile. But often this information is missing or incomplete. However, we can fill in this information by discovering evidence in the content the individual provides and, importantly, by reasoning about these various pieces of evidence.

AAAI Programme

With 8 parallel sessions to choose from, AAAI is a pretty tricky conference to navigate. However, you could do worse than simply go to the excellent AI and the Web track that Tim Finin and Peter Norvig are chairing. This track includes sessions on Semantic Web, Wikipedia, User Interactions, Trust and Authority (key for social network analysis), Ontologies (key for the enemies of folksonomy), NLP, Extraction and Understanding, Reputation and Sentiment (another great social network analysis area) and Social Networks.

There is also a session on Web Mining/Retrieval which includes the paper: 'TableRank: A Ranking Algorithm for Table Search and Retrieval' by Ying Li, Kun Bai, Prasenjit Mitr and C. Lee Giles. This is of particular interest to me (regular readers will know of my ongoing interest in table understanding) and I'm sure to Peter Norvig who, at the recent Future of Search event hosted by Berkeley, stated that tables were of particular interest to the search community.