A Network Model for Adaptive Information Retrieval

Crestani, Fabio A (1992) A Network Model for Adaptive Information Retrieval. MSc(R) thesis, University of Glasgow.

Abstract

This thesis presents a network model which can be used to represent Associative Information Retrieval applications at a conceptual level. The model presents interesting characteristics of adaptability and it has been used to model both traditional and knowledge based Information Retrieval applications. Moreover, three different processing frameworks which can be used to implement the conceptual model are presented. They provide three different ways of using domain knowledge to adapt the user formulated query to the characteristics of a specific application domain using the domain knowledge stored in a sub-network. The advantages and drawbacks of these three adaptive retrieval strategies are pointed out and discussed. The thesis also reports the results of an experimental investigation into the effectiveness of the adaptive retrieval given by a processing framework based on Neural Networks. This processing framework makes use of the learning and generalisation capabilities of the Backpropagation learning procedure for Neural Networks to build up and use application domain knowledge in the form of a sub-symbolic knowledge representation. The knowledge is acquired from examples of queries and relevant documents of the collection in use. In the tests reported in this thesis the Cranfield document collection has been used. Three different learning strategies are introduced and analysed. Their results in terms of learning and generalisation of the application domain knowledge are studied from an Information Retrieval point of view. Their retrieval results are studied and compared with those obtained by a traditional retrieval approach. The thesis concludes with a critical analysis of the results obtained in the experimental investigation and with a critical view of the operational effectiveness of such an approach.

• Item Type: Thesis (MSc(R))
• Qualification Level: Masters
• Keywords: Computer science
• Date of Award: 1992
• Depositing User: Enlighten Team
• Unique ID: glathesis:1992-78371
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 28 Feb 2020 12:09
• Last Modified: 28 Feb 2020 12:09
• URI: https://theses.gla.ac.uk/id/eprint/78371