Research And Application On Enterprise Knowledge Personalized Recommendation Methods

Under today's social environment,enterprises have accumulated lots of historical data.These data contain lots of experience and knowledge.A good company will put great emphasis on these data.These data indicate the development trend of the industry,which called enterprise knowledge.With the popularity of the Internet,knowledge spreads faster and faster.Enterprise knowledge is increasing quickly.Employees can't obtain the right knowledge on the right time.The effective of using knowledge base is decline.So far,there are two ways to solve the problem.One is information retrieval,such as search engine.Another is information filtering,such as recommendation system.The major study is the recommendation system.Currently,there are so many recommend algorithms to be proposed.Some algorithms are widely available,such as collaborative filtering.Although collaborative filtering has been applied to many commercial recommender systems successfully,but it still exists problems,such as data sparseness?cold start and it can't recommended on enterprise knowledge.With the popularity of the enterprise knowledge base,the recommendation of enterprise knowledge has become more popular.It becomes an important field for study that how to improve the recommended performance of enterprise knowledge.So we studied in the following aspect.Firstly,we focus on the study of Markov prediction model.The paper considered the schedule of seeking habits to propose an algorithm that named collaborative filtering recommendation algorithm based on MDP model.The experimental shows that the algorithm can improve the quality of recommendation.Secondly,we focus on the relevant study on rough set and enterprise knowledge.Recommend knowledge actively can improve the employees' efficiency of get knowledge.Meanwhile it can help to promote the innovation and application of enterprise knowledge.To describe the knowledge by three aspects: property,process and domain.To generate the knowledge recommend architecture based the classification.So the domain knowledge actively recommendation system based by process-driven and rough set was proposed.The architecture analysis the conception and data structure of employees?process?knowledge and domain.So we are using rough set to make the precise recommendation.