On the Use of the Dempster Shafer Model in Information Indexing and Retrieval Applications
The Dempster Shafer theory of evidence concerns the elicitation and manipulationof degrees of belief rendered by multiple sources of evidence to a commonset of propositions. Information indexing and retrieval applications use a varietyof quantitative means - both probabilistic and quasi-probabilistic - to representand manipulate relevance numbers and index vectors. Recently, severalproposals were made to use the Dempster Shafes model as a relevance calculusin such applications. The paper provides a critical review of these proposals,pointing at several theoretical caveats and suggesting ways to resolve them.The methodology is based on expounding a canonical indexing model whoserelevance measures and combination mechanisms are shown to be isomorphicto Shafer's belief functions and to Dempster's rule, respectively. Hence, thepaper has two objectives: (i) to describe and resolve some caveats in the waythe Dempster Shafer theory is applied to information indexing and retrieval,and (ii) to provide an intuitive interpretation of the Dempster Shafer theory, asit unfolds in the simple context of a canonical indexing model