Title: Learning to Map Natural Language to Executable Programs Over Databases
Advisor: Dragomir Radev
Other committee members:
Kathleen McKeown (Columbia University)
Luke Zettlemoyer (University of Washington)
Natural language is a fundamental form of information and communication and will be the next frontier in computer interfaces. Natural Language Interfaces (NLI) bridge the data and the user, significantly promoting the possibility and efficiency of information access for many users besides data experts. All consumer-facing software will one day have a dialogue interface, the next vital leap in the evolution of search engines. Such intelligent dialogue systems should be able to understand the meaning of language grounded in various contexts and generate effective language responses in different forms for information requests and human-computer communication. In particular, this dissertation focuses on answering the following three questions: 1) How can NLI ground human requests in natural language into formal programs (e.g., SQL)? 2) How can NLI converse with humans in a robust manner? 3) How can neural models understand compositionality in human language?