Data Analysis Project: Semi-Supervised Discovery of Named Entities and Relations from the Web

This project studies semi-supervised discovery of named entities, relational entities and prepositional phrase attachments within a read-the-web framework. Meanings of an entity can be improvised and updated faster in the internet world than printed references. The main idea of this project is to study the feasibility of characterizing entities by web content directly. The approach is that contextual words around an entity on web pages are first extracted and converted into a Bag-Of-Word (BOW) representation. We then apply several supervised and semi-supervised learning methods on top of these contextual words for several well known research problems: Named Entities Recognition, Relation Extraction and Prepositional Phrase Attachment.