Automated scientific document retrieval

Internet has revolutionized knowledge acquisition. Scientific knowledge in books, journals, and conference proceedings are available in digital libraries online. Current digital library end users are not adequately equipped with the technological know-how on information retrieval. While these systems normally retrieve documents based on keywords, users normally have a more abstract perception what information they require. Semantic gap, which is the disparity between user´s request and query results, has been identified as a challenging issue. In this paper, we are interested in scientific document indexing for retrieval. Knowing the structure of such a document is helpful to the retrieval process, as the document is decomposed into different elements and specific indexing methods are applied to the different element types. Also, as the document is viewed as a composition of many elements, query responses may return some of the elements instead of the entire document. Therefore the aim of this paper is to present a novel approach to information retrieval from scientific documents, combining text mining with image mining.