摘要 |
<P>PROBLEM TO BE SOLVED: To enable performing updating of a document index in an efficient manner. <P>SOLUTION: A data processing device 30, performs recalculation processing of word weight values, at a given update timing. At this timing, it judges whether a change rate Np of a total number of documents N exceeds a predetermined threshold level or not (step S11), and in case the change rate Np exceeds the threshold level, concerning all of selected words, word weight values are collectively recalculated (step S12). On the other hand, if the change rate Np is lower than the threshold level (NO at step S11), collective recalculation of word weight values is not actioned. In this situation, it judges, on basis of each individual word, whether recalculation of word weight values is necessary or not (step S13). If the change rate of a document frequency df of a certain word exceeds a predetermined threshold level, recalculation of weight values is actioned for this word. Whereas, recalculation of weight values is not actioned for this word, if the change rate of a document frequency df of a certain word is lower than the predetermined threshold level (step S14). <P>COPYRIGHT: (C)2012,JPO&INPIT |