| 摘要 |
PURPOSE:To extract an important part of good quality from a document by simple processing without assuming a specific document format or context. CONSTITUTION:A document analysis part 5 decomposes the document which is inputted from a document input part 1 into paragraph, sentences, and words. A context vector generation part 6 generates context vectors of the sentences, paragraphs, and document by using a word dictionary 8. A context vector comparison part 7 compares the comments with sentences by paragraphs, the document with the respective paragraphs, the paragraphs with respective sentences in the paragraphs, and the document with context vectors of the respective sentences to calculate distances between the context vectors. A document processing part 4 generates two kind of summaries of the paragraph which is the closest to the document and plural sentences which are the closest to the document and two kind of summaries of sentences by paragraphs which are the closest to the document and sentences which are the closest to the respective paragraphs by referring to the distances between the respective context vectors. Thus, the input document is analyzed by using the context vectors to extract the important part of good quality by the simple processing without assuming the specific document format or context. |
