Entity-Centric Search: Querying by Entities and for Entities

Опубликовано 21 июня 2016, 19:25

The immense scale of the Web has rendered itself as a huge repository storing information about various types of entities (e.g., persons, companies). Much of information retrieval operations on the Web nowadays are about entities, i.e., entity-centric. When the concept of entity is involved in a retrieval operation, people are usually interested finding more information about some known entities, i.e., querying by entities, or exploring unknown entities that satisfy certain information needs, i.e., querying for entities. In this talk, I will introduce the learning algorithms and prototype systems I developed for supporting these two types of entity-centric search operations. Specifically, I will focus on the querying by entities problem, and present a novel learning to rank framework which uses Restricted Boltzmann Machine to rank entity-relevant documents by keyword importance adaptation.