Study On Knowledge Servicce And K-Medoids Algorithm Improving In Big Data Environment

Traditional knowledge service mainly through technology, data resources, service strategy to provide users with reliable and detailed information and services. Along with the network technology, computer technology and the development of information technology, digital information resources is growing, constantly hitting the traditional mode of knowledge service, some of the traditional technology and methods, service means cannot meet the users’ information demand under the large data sets. This article through to the adopted technology of the era of big data university library knowledge service means, service mode, combined with the core of the large data processing technology, the key to knowledge clustering algorithm and data processing platform, puts forward the big data under the background of the knowledge service model, improved the set of large number of K-medoids knowledge clustering algorithm, the corresponding retrieval system is designed.In this paper, the main work has the following several aspects:1. Starting from the present situation of big data under the background of knowledge service, combined with the database system, query system, service system, build the knowledge search model under the background of big data, for the large data background knowledge search service, provides a theoretical reference.2. In view of the K-medoids clustering algorithm to select the sensitivity of the clustering center and the processing of large amounts of data gathered in the process of the class the disadvantage of poor performance, proposed to improve the knowledge of K-medoids clustering algorithm, and use the relevant test documents for testing. Test results show that the improved algorithm with the combination of knowledge extraction algorithm, effectively improve the clustering process of documents in recall, precision and F1 value, reduces the document segmentation, raise the efficiency of clustering.3. According to the proposed model of knowledge service, combined with improved K medoids clustering methods, a clustering search system was implemented. System operation results show that the model based on the proposed knowledge service and the system improved clustering algorithm can overcome the traditional search engine when processing the data classification problem such as poor effect and time consuming.