A Dempster-Shafer indexing for structured document retrieval: implementation and experiments on a Web museum collection
Ekaterini Moutogianni · 1999
This study investigates the effectiveness of a model for the focussed retrieval of hierarchically structured documents based on the Dempster-Shafer theory of evidence advanced by Lalmas. The particularity of the model is that the representation of a document is defined as the aggregation of the representation of its own content and that of its inter-linked documents. In this study, we seek to apply the proposed model on the World Wide Web and evaluate its effectiveness. The model assumes a hierarchical structure, and concentrates on the following two aspects: (1) indexing: the Dempster combination rule is used to aggregate recursively from bottom to top the individual document representations as provided by the standard indexing procedure; (2) retrieval: belief functions are used to retrieve documents for given queries. Based on the resulting belief values and the ranking of documents, two different algorithms are used to implement the focussed retrieval. An experimental system was developed to investigate the effectiveness of the focussed retrieval, using the Dempster-Shafer indexing (the aggregation). (3 pages)