| 摘要 |
PROBLEM TO BE SOLVED: To prevent the accuracy of a co-occurrence dictionary from significantly deteriorating when suppressing bloating of a co-occurrence matrix thereof.SOLUTION: A device for creating a co-occurrence dictionary comprises: a sorting section for receiving a co-occurrence matrix and sorting rows and columns of the co-occurrence matrix such that words having similar meanings are adjacent to each other in row and column directions; an image creating section for creating an image that has the values recorded at the point of intersections of the rows and columns of the sorted co-occurrence matrix as the luminance values of pixels at the point of intersection of corresponding rows and columns of the image; an image reducing section for eliminating high frequency components from DCT coefficients obtained by performing Discrete Cosine Transform (DCT) on the image, and for performing Inverse Discrete Cosine Transform (IDCT) on the residual DCT coefficients to create a reduced image from the image; and a dictionary creating section for creating a co-occurrence dictionary composed of a reduced co-occurrence matrix which has rows and columns corresponding to those of the reduced image and to which the luminance values of pixels corresponding to those of the reduced image are recorded as frequencies at points of intersection of the rows and columns, and relational information representing the correspondence between a word and an identification number assigned to each of the rows and columns of the reduced co-occurrence matrix. |
