Seminar talk by Dr. DiptiKalyan Saha

Dr. Diptikalyan Saha visited IIIT-Delhi on September 9th, 2016. He gave a talk on his recent research work titled "ATHENA: An Ontology-Driven System for Natural Language Querying over Relational Data Stores" which was presented in VLDB 2016. Talk is about ATHENA, an ontology-driven system for natural language querying of complex relational databases. Natural language interfaces to databases enable users easy access to data, without the need to learn a complex query language, such as SQL. ATHENA uses domain specific ontologies, which describe the semantic entities, and their relationships in a domain. It is a unique two-stage approach, where the input natural language query (NLQ) is first translated into an intermediate query language over the ontology, called OQL, and subsequently translated into SQL. The approach allows to decouple the physical layout of the data in the relational store from the semantics of the query, providing physical independence. Moreover, ontologies provide richer semantic information, such as inheritance and membership relations, that are lost in a relational schema. By reasoning over the ontologies, the NLQ engine is able to accurately capture the user intent.