| As our country has given strong support to the scientific enterprise, the research grant offinance support has been highly risen. In order to prevent the repeated project application orthe behavior of plagiarizing that may lead to the waste of research grant or the standstill of thescience and technology enterprise during the application process, the similarity detectionalgorithm aimed at the application of science and technology project is regarded as theresearch object, then the intensive study on the similarity calculation method of Chinesephrases, sentences and texts are made respectively. After that, the research result can be usedin the application system of science and technology project.The traditional similarity calculation method of phrases, sentences and texts are dissected,then the existing defect and the method that can be improved are found. Secondly, based onthe insufficient of the common similarity calculation method of phrases and texts, two newmethod are proposed, one is a modified similarity calculation method based on the《HowNet》, another is a similarity calculation method that combines with the building oflanguage network and semantic analysis. Thus, the calculation accuracy is highly improved,but the complexity is also be increased. Finally, after the research result being used in theapplication process of science and technology project, a detecting system on the textsimilarity of science and technology project is designed and realized, so as to test thesimilarity of the application text of science and technology project. This paper does theresearch on the detecting algorithm of text similarity that aims at the specific area on theapplication of science and technology project, and the research result is used in the textdetecting system of science and technology project combining with the practice. It is valuablefor the theoretical research, and also be helpful for promoting the development of scientificand technology enterprise. |
PDF Full Text Request