Entity Linking and Knowledge Discovery in Microblogs.

Social media platforms have become significantly popular and are widely used for various customer services and communication. As a result, they experience a real-time emergence of new entities, ranging from product launches to trending mentions of celebrities. On the other hand, a Knowledge Base (KB) is used to represent entities of interest/relevance for general public, however, unlikely to cover all entities appearing on social media. One of the key tasks towards bridging the gap between Web of Unstructured Data and Web of Data is identifying such entities from social media streams which are important and haven’t been yet represented in a KB. The main focus of this PhD work is discovery of new knowledge from social media streams in the form of new entities and/or new mentions of existing entities while enriching KBs as well as lexically extending them for existing entities. Based on the discovery of new entities or new mentions, structured data in the form of RDF (Resource Description Framework) can be extracted from the Web.