Study On Hierarchical Text Categorization Of Patent Data Based On Fuzzy Logistic

In the era of economic globalization, the patent technology has become a core competitiveness of a country or a region, at the same time; patents of intellectual property attract more and more attention from companies. However, the prevailing technology of patent analysis is inefficient and of long cycle. The surging of patent applications increased the demand for a more rapidly automatical patent analysis technology; on the other hand, it prepares adequate resources for the patent text mining methods based on data mining technology. Therefore, computer-aided patents analysis will be a trend of the times.Defects, such as inefficiency, and too many mistakes and so on, are widespread in the current Artificial classifications; in practical classification of patent text, a patent paper may belongs to different classes, that was different with general test classification; Most classifications of patent test now adopt the traditional text classification algorithms, without considering the special problems ,such as that patent data involves in cross-subject area, and the high similarity existing in the family patents, and so on.Based on the above considerations, this paper focused on the patent text mining in the automatic classification issues. Firstly, This paper firstly introduced information characteristics of the patent documents and general classifications of international IPC. Based on the characteristics of patent documents, this paper forwarded the feature extraction method of patent document. The introduction of the location weight in the method makes the Vector description of patents document more accurate. Secondly, the fourth chapter outlined the general text classification algorithm, and forwarded a patent automatically text classification algorithm. Considering the patent document subdivision on deep level in the patent analysis, and Cross-disciplinary patent belonging to more than one category on the family patent study, we advanced a patent data classification algorithm on fuzzy logic-level. The detailed algorithm is interpreted in chapter 5. Finally, we formulated a category Hierarchical model, and practice simulation on the 170 patent documents. The classifications in the 1st and 2nd level are ideal, but still need improvements in third level...