Library Log Information-based Data Mining Research

In order to obtain information from the mass data sets, the technology of data mining emerges. This technology is widely used in most fields all over the world, but it is seldom used in the field of library.In our country, every library confronts the problem that how books-purchasing funds should be used rationally except the National Library. According to the incomplete statistics, over 200,000 kinds of books are published every year in our country, the amount of which is increasing with the speed of 7 percent every year, but the speed of the increased allocated funds cannot make ends meet. Therefore, the most effective way of resolving the problem above is to use limited funds to buy books of higher demand. And the daily record information is the best source. It can not only show the number of books borrowed and the group of the reader, but also predict the demand of types of unreserved books through the data mining technology.Through building the angled model of daily record information of library, this thesis attempts to carry out data mining by clustering, and combines it with web mining of bibliotheca searches' information, so that the demand for books could be mastered completely. The application of data mining technology will provide scientific reference to acquisition, so as to avoid declination and wrong estimate to real demand and low efficiency in use of the resource arising from subjective assume.