Understanding User-generated Content on Social Media

Опубликовано 17 августа 2016, 20:55

In this talk I will cover some of my dissertation research that is devoted to the study of textual user-generated content on social media platforms. My work investigates the role of contextual information in discovering various types of semantic and experiential metadata from informal user-generated content toward the end goals of adding structure to unstructured utterances. I will present my investigations along three aspects of understanding user-generated content: (a) named entity recognition in informal text, with a focus on identifying cultural entities that allows us to trend and track their online mentions. (b) understanding user intentions behind expressed content, with a focus on identifying transactional and information seeking intents that are typically exploited by question answering and targeted content-delivery systems. (c) studying how users self-present on social media by observing the words they employ. An understanding of such micro-level variables of content allows us to comprehend something new about the data and dynamics of social media. In this talk, I will also present my work on two deployed social Web applications that tap into crowd-sourced intelligence.