In order to be able to search quickly in a data base containing natural language documents, it is usual to create an inverted file directory. In an effort to avoid the bothersome creation of such an inverted file, experiments have been done in which automatically links are made between documents. These links are based on similarities in the text of documents. The degree of similarity is basically detected by the co-occurrence of 3 letter fragments. Retrieval from a data base, using this theory, will be tested and reported on in future.
Samenvatting