Optimizing Index of News Articles by Table Based Version of K Nearest Neighbors

In this research, we propose the index optimiza- tion for maximizing both the performance and the efficiency of information retrieval systems and the application of the proposed KNN to the task. The index optimization is mapped into a classification task within a domain, the task should be distinguished from the topic based word categorization, and the previous works show the better text categorization perfor- mances from encoding texts into tables than into numerical vectors. In the proposed system, a text which is tagged with its own domain is given as the input, the words which are indexed from the text are classified into expansion, inclusion, or removal, by the table based KNN version. We validated empirically that the proposed KNN works better than the tradition version in optimizing indexes of news articles. In future, we will connect the task with the text categorization, in order to process texts which are untagged with their domains.