Based On Active Collaborative Filtering Recommendation System Key Technology Research

With the coming of the information society, the information we need rely on the Internet more and more. But the rapid growth of online information has far exceeded our ability to be able to accurately identify the scope. If we want to find some information that we need on the Internet, the most information we find through the traditional retrieval system or search engine is not often that we need or simply not be in vain. That is the phenomenon called "Information Overload" and "Information Labyrinth". In order to overcome such difficulties to obtain information, recommendation system came into being. It helped us come out from "Information Labyrinth" and find the most valuable parts of large amounts of information for us.Collaborative filtering recommendation and content-based filtering recommendation are the two kinds of the recommendations that are the most common and also the most commonly used. The former considers the users who have the similar interests to some information and then determine whether or not to recommend this information to the users. The latter considers whether the information itself is correlated to the users'interest. Then work out the similarity of the two and based on it to determine whether or not to recommend. Although these two filtering technologies have been successful not only in theory but also in practice and have got effective applications in information retrieval and e-commerce, they still have insufficient of themselves. As for collaborative filtering recommendation, sparse, credibility and the geometric growth of the computation complexity with the growth of the scale are the main facing problems. As for content-based filtering, its main disadvantages are the limited capacity of feature extraction, unable to recommend more updated information resources and needs too much users'feedback.This paper will discuss the recommendation of instruments based on Scientific Instrument Shared Service Platform of YunNan and talk about something about user rights management. Through some study and comparation, here decides to use collaborative filtering technology and RBAC-based user rights management. These can not only complete parts of the platform design but also improve some problems in collaborative filtering. It overcomes some adverse effects of the sparsity in a way and at the same time, it makes the credibility of the recommendation increase.