A neural network-based knowledge retrieval system with relevance feedback

Authors:
Sassan Sheedvash;Mahmood R. Azimi-Sadjadi
Affiliations:
Hewellet Packard, IPG-IT, San Diego, CA;Department of Electrical and Computer Engineering, Colorado State University, Fort Collins, CO
Venue:
ASMCSS'09 Proceedings of the 3rd International Conference on Applied Mathematics, Simulation, Modelling, Circuits, Systems and Signals
Year:
2009

Citing 3
Cited 0

Relevance feedback and other query modification techniques

Information retrieval
The nature of statistical learning theory

The nature of statistical learning theory
An Adaptable Connectionist Text-Retrieval System With Relevance Feedback

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents the results of a new neural network-based system for a large scale knowledge retrieval system. The system optimally map the original users' queries using relevance feedback from multiple users. The learning can be implmented in either regression or classification modes using a simple three layer linear network. The first layer is an adaptable layer that maps from the query domain to the document domain. The second and third layers perform document-to-term mapping, search and scoring tasks. The proposed learning algorithms are successfully tested on a large text collection encompassing a wide range of HP products, and for a large number of commonly-used single and multi-term queries. The system was successfully deployed in April 2005, and is currently used by over 12,000 call center agents to resolve customer issues in a real time environment. The production data indicates a drastic improvement in resolution rates.