Research On The Personalized Information Recommendation Method Of College Libraries Based On Data Mining

All along, college library plays an important role in university teaching andscientific research, which is the place to store all kinds of book resources, periodicals,magazines and audio-visual data, etc, so that through the library, teachers andstudents can not only increase knowledge, seek answers to questions, but alsodevelop reading interests and habits. And along with the rapid development ofdatabase technology and the wide application of database management system in thecollege library, massive historical access data to book resources of readers accumulatein the library databases, and massive important information is hidden behind the data.How to mine valuable information from library’s massive data using data miningtechnology, provide personalized information recommendation service to readers,improve the readers’ satisfaction, not only is one of the urgent problems in the field oflibrary, but also one of the hot spot in the field of research in data mining.Firstly, the article introduces the development of China’s college libraries, andanalyzes the present problems the college libraries are facing, puts forward the idea ofapplying data mining technology to develop personalized informationrecommendation service, at the same time introduced the basic theory and theresearch status of data mining. Then it stated the relative knowledge of theclassification of the books, synthetically uses the index types of books in ChineseLibrary Classification to build a book index distribution tree, calculating similaritybetween books, further to obtain the distance between the readers which reflect thereaders interest preference approximation degree, and in combination with a K-Meansclustering algorithm based on the Anti-Kruskal to make effective group classificationof readers of library, in order to realize association rules analysis according to thedifferent preferences、reading habits of readers. Thirdly, analyzed one commonalgorithm for assoiation rules mining–Apriori algorithm, discussed the basic ideas,mining steps, advantages and disadvantages of this algorithm, and proposed animproved Apriori algorithm based on the hash table (Apriori-Hash Algorithm). This algorithm uses hash table to record the different width of the transaction id in order torealize the transaction fast positioning, and using a hash function to quickly generatefrequent itemsets2directly, at the same time using the optimized pruning andconnection strategy to improving the efficiency of the algorithm. Finally, combinedwith practical design implements a model of personalized informationrecommendation service system based on data mining, this model is mainlycomposed of data preprocessing module, visualization module, the miningrecommendation module, such as several functional modules. First, extract the sourcedata needs from the university library, after carries on cleaning, conversion,integration of the pretreatment work, can get readers borrowing data set which meatdata format can be implement mining operation, then use a K-Means clusteringalgorithm based on the Anti-Kruskal to make effective group classification of readersof library, induction summed up the reader’s interest preference, reading habits andother features, and then using the improved Apriori algorithm based on Hash(Apriori-Hash algorithm) for mining association rules on borrowing data of eachreader group, find out the relationship between books resources readersborrowed, formed books recommended model for readers who have different lendingpreference, reading habits and so on characteristics, implemented as every reader tocarry out the personalized information recommendation service.