Location

MIT Media Lab, Room TBA

Description

Many organizations are very concerned with understanding how groups of customers, employees, or the general public feel about their products and services. Electronic commerce, customer and employee forums, blogs, and citizen journalism provide new sources of feedback to "close the loop" on the effects of companies' operations. However, many of the sources are in free-form natural language rather than controlled surveys or focus groups, and thus large-scale analysis of these sources is usually done in simplistic and inaccurate ways. We will explore new techniques that provide deeper understanding of opinion from natural-language sources, using a large common-sense database, state-of-the-art natural-language processing, and new inference techniques that find previously unseen dimensions with which to categorize opinion.

Additional Featured Research By

(Unpublished) Software Agents