Technology Opportunities

Available Knowhow

Provides numerical information about all sorts of objects in reply to search engine queries

Professor Ari Rappoport., Natural Language Processing and Computational Cognitive Linguistics group, Rachel and Selim Benin School of Computer Science and Engineering

Highlights

• Many queries submitted by search engines and in question answering systems relate to numerical data – height, length, weight, cost – of various objects
• This unique system rapidly searches for an answer, and if the exact data is not available, provides a highly accurate approximation derived by comparison with available data for similar objects
• Currently it is possible to manually annotate objects with numerical properties, but this is a hard and labor intensive task and is impractical for dealing with the vast number of objects of interest.
• There is thus a need for automated semantic acquisition algorithms targeting numerical properties.

Our Innovation

Novel query-response framework that automatically extracts or approximates numerical attributes of all types of objects from the Web

Key Features

• Framework utilises relation defining patterns and WordNet similarity information.
• Framework provides a significant improvement when compared with answers given in Wikipedia and by leading search engines such as Google and Wolfram Alpha.

There are currently about 1.5 billion Web surfers worldwide and virtually all of them use search engines in order to answer questions. A substantial percentage of these questions involve numerical answers.