ON-LINE INFORMATION RETRIEVAL USING ASSOCIATIVE INDEXING

Abstract : Experiments were performed to determine the feasibility of using ALCAPP as one form of on-line dialogue. Assuming the ALCAPP (Automatic List Classification and Profile Production) system is in an on-line mode, investigations of those parameters which could affect its stability and reliability were conducted. Fifty-two full text documents were used to test how type of indexing, depth of indexing, the classification algorithm, the order of document presentation, and the homogeneity of the document collection would affect the hierarchical grouping programs of ALCAPP. Six hundred abstracts were used to study the effect on document clusters when more documents are added to the data base and the effect on the final cluster arrangement when the initial assignment of documents to clusters is arbitrary. Results reveal that the only time significant differences in the classification of documents does not occur is when the order of document presentation is varied. Final clusters are significantly affected by the initial assignment of documents to clusters. The number of documents added to a data base allows stability of clusters only to a cutoff point which is some percentage of the original number of documents in the data base.