Return to author index

Volume: 5 Issue: 1

Domain specific Information retrieval system with a correspondence graph applied to car diagnosis

Azarian Armin, Siadat Ali

Abstract:
This paper relates our experience in the design of an information retrieval module with the expression of request in natural language. The experimental platform is the car diagnosis system called SIDIS Enterprise developed by Siemens AG. This paper analyses the difficulties of automatic treatment of natural language and reports on the experience resulting from the implementation of usual research algorithms based on vector modeling of documents. Furthermore, we have developed a new approach based on an artificial neuronal network, which emulates cognitive relevance judgments. Finally, several other aspects such as storage requirements, algorithm complexity, implementation difficulty and processing time, were also considered during the tests.

Keywords:
Information retrieval, correspondence graph, neural networks, car diagnosis, perceived symptoms, natural language processing

doi:10.5019/j.ijcir.2004.169

Full Text PDF
















^ TOP