A Model for Representing and Retrieving Heterogeneous Structured Documents Based on Evidential Reasoning

Documents often display an internal structure; they are composed of components. For example, a journal contains several articles, which themselves contain paragraphs, tables, etc. With structured documents, the retrievable units should be the document components as well as the whole document. The components of a structured document can be of different types: various media, located in a number of sites, or written in several languages. An information retrieval model for heterogeneous structured documents must take into account this disparity among document components. We present a model for representing and retrieving heterogeneous structured documents, that is multimedia, distributed and multilingual documents. The model is based on evidential reasoning, a formal theory that allows for the representation and the combination of knowledge. Here, knowledge is the content of document components. We show that the model provides for an appropriate representation and retrieval of heterogeneous structured documents.