| To invigorate the country through science and education is a basic strategy of our country. A large number of human resources, material resources and financial resources are put into different kinds of research and development in our country. Scientific, economic and social value coming from scientific and technological project has direct and important effect on the development of S&T and society. With more and more outlay put into the S&T projects all over our country and expansion of the input, the scientific and technological productivity have been improved fast and therefore we encounter many problems in management.The evaluation and examination of S&T projects is an important way to distribute national resources scientifically. S&T activities in different fields have different characteristics. Classifying the S&T projects with scientific and reasonable method can provide basis for financial and risk evaluation in different fields. All kinds of models will become more pertinent, scientific and accurate to the S&T projects evaluation.The scientific and technological project classifying model based on hierarchical clustering firstly calculate the similarity of the projects through making comparison between their eigenvectors composed of the keywords frequency statistical information .Secondly it use the least-square method to fit the similarity distribution curve and get the inflection points of this curve. Lastly it builds a hierarchical tree through the hierarchical clustering method and realizes project classification with different granularity.When searching similar projects, searching the hierarchical tree get from clustering by using breadth-first search algorithm can improve the speed and efficiency greatly. It also can find out exactly the similar projects whose similar threshold are higher than the guard threshold and prevent the similar projects being established repeatedly, which makes a waste on resource distribution. |
PDF Full Text Request