Research And Implementation Of Science And Technology Project Declaration Aided Detection System

As the state to encourage scientific and technological work and a lot of investment incapital, scientific and technical workers upsurge of enthusiasm for scientific research and thenumber of projects declaration is gradually increasing. However, due to the popularizationand development of the network, the way to access information is also increasinglybroad,which is bound to give science and technology project vetting workers a great deal ofconfusion in the process of project formal examination: how to determine whether the projectrepeated declaration faced with a huge number of projects declaration. But the examinationonly through artificial form would be unrealistic, designing a text-aided detection system isvery necessary. According to the text feature of Jiangxi science and technology awarddeclaration and combining with semantic analysis, this paper computes the text semanticsimilarity and implements a science and technology project declaration text-aided detectionsystem that can provide automatic and fair decision support for science and technologyproject vetting workers. The main work of this paper is as follows:(1) From the perspective of text detection technology, this paper researched andsummarized the applicability and limitations of several existing mainstream detectiontechnology and prototype systems, then combined with the system common architecture,proposes the architecture of my system design.(2) According to researching the existing semantic analysis methods, analyzing andcomparing the characteristics of these methods, combined with the characteristics of Chineselanguage and the usage of existing systems development, this paper uses HowNet which is alanguage knowledge base, and utilizes words semantic similarity based-HowNet to computesemantic of text. To solve the problem that HowNet is unable to compute unknown wordsemantic similarity, this paper improves it and considers the unknown words semanticcalculation.(3) According to research several text similarity calculation methods and summarizingtheir characteristics and existing limitations, this paper proposes a text similaritycalculation Method combined text structure with semantic information. The main idea of thismethod is: the text represents a combination of different semantic parts, and each differentpart of the text uses a different semantic calculation method which considers the influence ofsemantics and word order in the calculation of sentence similarity. While comparing withother methods though experiments and analysis, demonstrates that this proposed methodimproves the precision and recall in detection of science and technology project declarations. Based on the above researches, this paper applies the new method into the system:designs and implements a text-aided detection system and describes in detail the mainfunction modules of implementing the system, which include Data storage module, textpreprocessing module, similarity calculation module and analysis module. By running thesystem, the results show that the system can effectively detect similar project declaration, candisplay detailed plagiarism cases, and has relatively strong practicability.