Abstract: a kind of decision algorithm of R&D project termination based on improved rough set theory is proposed.The algorithm makes use of the vector space theory to improve the attributes reduction and the rules extraction in the rough set theory and to resolve the questions that calculation of reduction is huge and characteristic is not clear when there are few samples of R&D project to some extent.At the same time,considering R&D project experience that is aborted or successful in the past as samples for learning,it can discern the classification of R&D project with uncertain feature so as to make decision of termination or continued studying.Finally,the feasibility and validity of this algorithm are testified by the instance from a large-scaled enterprise in Chongqing and the comparison with the results by using supported vector machine.

Key words: R&D project, termination decision, rough set theory, vector space, support vector machine