Development of a Computational System for Integrating Usage into Document Indexing.
Loading...
Date
2014
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Obafemi Awolowo University
Abstract
The study formulated a model that augments document with usage, designed, implemented and evaluated a system based on the model. This is with the view of enhancing the quality and quantity of useful documents that are returned during document search operation.
Attribute Value Pair technique of data abstraction in document annotation and vector model technique of Information Retrieval were used to formulate the document usage model. Unifying Modelling Language (UML 2.0) was used to design the Competitive Intelligence based Document Usage Creation and Exploration (CIDUCE) system. The prototype was implemented with the use of PHP and MySQL technology. Data on document usage was collected through questionnaire administration and guided interview from 20 selected postgraduate students (M.Sc. and Ph.D.) in various departments in the Faculty of Technology. Ninety-nine (99) documents and twenty (20) decision problems were extracted from the questionnaire and used to populate the database of the system. Document recall rate, a function of the similarity measure between identified relevant documents by the respondents and their decision problems (i.e. research problems) was used to evaluate the system.
The results showed that the usage-based document index consistently produce high recall rate, that is, identified high number of relevant documents at different retrieval thresholds than the keyterm-based index. For example, at the retrieval thresholds of 0.20, 0.30, 0.40, 0.50, 0.60, 0.70 and 0.80, the keyterm-based index has 47.47, 27.27, 14.14, 9.09, 2.02, 1.01 and 0.00% recall rates, respectively as compare with the usage-based index with recall rate of 100.00, 100.00, 100.00, 100.00, 100.00, 91.92 and 61.62%, respectively. These recall rates at different thresholds translated to 47, 27, 14, 9, 2, 1 and 0 documents, respectively in the keyterm-based index and 99, 99, 99, 99, 99, 92 and 62 documents, respectively in the usage-based index.
The study concluded that in an information seeking process, there are usually documents in the document collection space whose index may not contain terms in the users query but which are very relevant to users’ need.
Description
xvii,235 Pages
Keywords
Computational System, Document Indexing, Usage model, Competitive Intelligence based Document Usage Creation and Exploration, PHP and MySQL technology
Citation
Akanbi,L.A.(2014).Development of a computational system for integrating usage into document indexing.Obafemi Awolowo University.