Application And Research Of Nerve Network In Component Classification And Fuzzy Retrieval

Software-reuse appears with the concept as early as in 1968 NATO conference Until now, Software-reuse uses development and research affected area already extremely widespread, will reuse more and more is also big with the degree, thus, applied in the different domain, the different function various software component emerges one after another incessantly. Facing the so multitudinous software component, how will we carry on the classification in the software development to the component, how will choose the component which we will need, has become an essential solution question which we first will face.Nowadays, had very many methods to the component classification and the retrieval, for example: Faceted, Index, Based on Knowledge Library and so on, but all has the certain flaw, specially has the very big problem in the component automatic sorting and the fuzzy retrieval. This article separately has carried on the discussion to these classifications and the retrieval method, and selected has studied the component with the artificial intelligence nerve network method the classification and the retrieval.This article first introduced the artificial nerve network theory, and proposed may use from the Self-Organization feature Map nerve network (the SOM net) non-teacher, the automatic study, the automatic sorting characteristic carries on the automatic sorting to the component, and realizes the component fuzzy retrieval function in this foundation.The fourth chapter in detail narrate with the example how to construct the organization nerve network to carry on the component classification retrieval, as well as how did the network automatically study, the automatic sorting, and through graph objective response classification result quality. Then realizes the component fuzzy retrieval in this foundation. The result indicated introduces the nerve network the component in the classified retrieval, truly can achieve the automatic sorting, the fuzzy retrieval goal, greatly reduced the manual classification time expenses, simultaneously also gave the component the retrieval to provide a brand-new retrieval method.