Indexing and searching for information in hypertext systems

There is an increasing interest in hypertext systems. Nevertheless, only a few experiments deal with large-sized hypertexts. Due to the huge number of nodes and links, browsing cannot be viewed as an adequate searching process even with the introduction of a table of contents, indexes, local or global maps. Nevertheless, in such cases, an effective means of accessing information is needed. Based on two well-known test-collections, namely the CACM (3,204 documents, 50 queries) and the CISI corpus (1,460 documents 35 queries), this study demonstrates the usefulness of interdocument relationships to significantly enhance the performance of a retrieval model dedicated to hypertext systems.