NQuest – Natural Language Query Exploration System
At a glance
- Project leader : Prof. Dr. Kurt Stockinger
- Deputy of project leader : João Pedro Monteiro
- Project team : Prof. Dr. Abraham Bernstein, Yasamin Eslahi, Till Haug, Stefan Holdener, Claude Lehmann, Marcos Monteiro, Carlo Saladin, Ana-Claudia Sima
- Project status : ongoing
- Funding partner : Innosuisse (Innovationsprojekt / Projekt Nr. 34223.1 IP-ICT)
- Project partner : Veezoo AG, Universität Zürich
- Contact person : Kurt Stockinger
Description
There is a huge amount of valuable information hidden in a
company's database which is not easily accessible to business
people. To query these databases, end users need to know the
technical query language SQL as well as the database structure.
However, typical end users do not have enough SQL skills to
formulate complex queries. Even more so, higher-level analytics,
e.g. "trend analysis over last month" or "detect outliers in the
price fluctuation of product X over the last year" are hard to
formulate even for SQL experts. Hence, the majority of nonexpert
users are basically not able to explore the available knowledge of
their company.
Veezoo currently provides a system that can answer natural language
queries against databases, with the goal of empowering all users
inside a company to become data-driven and benefit from the
available information. However, feedback from existing users shows
that a wide range of customers completely lack familiarity with
their own company's databases. In practice, this leads to a
severely limited adoption of systems that provide a natural
language interface for databases, given that most users are not
aware apriori which questions to ask or on which regions of data to
focus, in order to get the most added value from the large amounts
of knowledge made available to them. Therefore, in the lack of
proactive suggestions, recommended insights, as well as data
exploration guidance, only translating natural language questions
to equivalent database queries is simply not enough.
In this project we tackle this important open issue to make natural
language interfaces to databases more suitable for widespread
adoption by designing novel algorithms on top of the current Veezoo
system, through a service that proactively guides users in
exploring the data and augmenting the company's knowledge base. The
service, called NQuest, will provide analytics mechanisms that
empower a wide range of users to discover new insights in existing
databases.