Research On Personalized Recommendation Based On Frequent Item Set

With the rapid growth of science and technology, today's society is in the rapid development of information era,and information has become an important resource in today's society.our life and on the internet are filled with all kinds of information,there are useful information we needed, also garbage information that have nothing to do with us, such as the various spam and spam mail, etc.Traditional way of information recommendation is mostly based on the user's clicks or based on transaction records,This recommendation algorithm can call a recommendation algorithm based on experience,It has its own advantages.However, in this era of information explosion, traditional recommendation algorithm based on experience have been unable to meet the needs of the users for information.The problem that how to understand reader's interest and requirements fully and accurately and recommend the information to the user timely and effectively has become an urgent need to solve.This paper mainly completed the following work:First, first analyses the characteristics of news information and the technology used for common personalized news recommendation,content-based recommendation and collaborative filtering recommendation ,including user-based,item-based and Slope-one recommendation,at the same time,we study the ways to compute similarity ,including introducing the cosine similarity.Second, design the model as the core of the content-based recommendation and collaborative filtering recommendation.using the technology of content-based recommendation to process the data and remove duplicate content,using collaborative filtering to generate the result of recommendation.using the the concept of frequent itemsets to narrow the scope of the calculated data, improve the accuracy and the efficiency of the recommendation,Using similarity algorithm is to find the most suitable news are recommended.Third, we realize the personalized news recommendation system based on frequent itemsets, and using the collected information as Experimental data for testing, and analysis the precision and recall rate of the test result.and validate the effect of recommended recommendation system model,we take the application into some news recommended in the project, it worked well and realize the personalized news recommendation...