Title

Natural Language Interfaces to Data

Abstract

Abstract

Enterprises are building domain specific complex knowledge graphs (KG) to use in their analytical applications. Exploring such domain specific KGs requires different querying capabilities: In addition to search, these systems also require very precise structured queries, as well as complex graph queries. Instead of learning and using many complex query languages, one natural way to query the data in these KGs is using natural language interfaces to explore the data. In this talk, we will describe a unique end-to-end ontology-based system for natural language querying over complex data sets. The system uses domain ontologies, which describe the data in terms of entities and their relationships, providing a semantic abstraction to capture user intent. We propose a unique two-stage approach: In the first stage, a natural language query (NLQ) is translated into an intermediate query language, called Ontology Query Language (OQL), over the domain ontology. In the second stage, each OQL query is translated into a backend query (e.g., SQL). The system has been used in multiple domains, and we will highlight a few use cases, including one from finance, and one from healthcare.

Conversational interfaces are the natural next step, which extends one-shot NLQ to a dialog between the system and the user, bringing the context into consideration, and allowing informed and better disambiguation. In this talk, we also discuss how to extend and adapt our ontology-based NLQ system for conversational services, and discuss the challenges involved. One important use case is the business intelligence (BI) applications where users explore the data using BI tools. We describe an end-to-end methodology to bootstrap a conversation service for a BI application by exploiting the well-defined set of access patterns.

Biography: Fatma Özcan is a Principal Research Staff Member and a senior manager at IBM Almaden Research Center. Her current research focuses on platforms and infra-structure for large-scale data analysis, knowledge graphs, democratizing analytics via NLQ and conversational interfaces to data, and query processing and optimization of semi-structured data. Dr Özcan got her PhD degree in computer science from University of Maryland, College Park, and her BSc in computer engineering from METU, Ankara. She has over 18 years of experience in industrial research, and has delivered core technologies into IBM products She is the co-author of the book "Heterogeneous Agent Systems", and co-author of several conference papers and patents. She serves on program committees of leading data conferences, and editorial boards of journals, as well as NSF panels. She is an elected member of the SIGMOD Executive Committee, and is on the board of trustees for the VLDB Endowment. She is an ACM Distinguished Member.

Supervisor(s)

Supervisor(s)

Dr. Fatma Ozcan

Date and Location

Date and Location

2019-12-27;11:00:00-A105

Category

Category

Seminar