No abstract

SUMMARYThere is increasing demand for a filtering system for scientific papers that can be utilized by researchers in business organizations. In conventional paper filtering systems, relevant papers are selected without sufficiently representing the researcher's interest structure, and thus the accuracy of filtering is not satisfactory. The purpose of this paper is to solve this problem and to propose a method of providing a structured representation of the researcher's interests by clustering papers of interest to the researcher. The similarity of newly appearing papers to the structured representation is calculated and the result of filtering is presented with rankings, resulting in both accuracy and convenience of use. With three researchers as the participants, an experimental evaluation of the filtering was performed by an N-fold cross-validation test, using 2 years of papers from proceedings journals in information science. An accuracy of 44% to 69% was achieved, 9.6% to 17.9% higher than the results of a filtering system without clustering used as the baseline.

A new method for automatic keyword extracting and "role" setting is proposed based on the Japanese sentence structure analysis. The analysis takes into account the following features of Japanese sentences, i.e., the structure of a sentence is determined by the noun-predicate verb dependency, and the case indicating words (kaku-joshi) play an important role in deep case structure. By utilizing the meaning of a noun as it depends on each predicate verb, restricted semantic processing becomes possible. An automatic indexing system, equipped with a man-machine interactive error-correcting function, has been developed. The evaluation of the system is performed by applying it in news information retrieval. The results of this evaluation show that the system can be put to practical use.

No abstract

No abstract