Research Title: Argumentation in the Social Web
Start date: October 2012
The theoretical foundations behind the functioning of modern search engines rely on a structuration of documents that is no longer the unique norm on the Web. Indeed, the shift from the documentary Web to the Web as a platform for conversation that has been witnessed in the past few years means that a massive quantity of untapped information can be found in user-contributed content. This type of content is highly subjective in nature and focuses on the experience of the users. In addition, it has the particularity of evolving over time, because of the faculty of any user to add their own content to the discussion at any point in time, which makes it difficult to index automatically from the point of view of web search engines.
The goal of this project is to leverage alternative theories, based no more on a statistical analysis of static documents but on the formal study of debate patterns and argumentation schemes. We investigate the use of machine learning techniques to discover argumentation patterns and the ways such knowledge could help us index social media data more effectively.