Comparative Study Of Recommendation Algorithms Based On Collaborative Filtering

With the continuous development of science and technology,we have entered the era of "big data",All kinds of data and information emerging in an endless stream.In the information overload Society,what people really need and are interested in is a tiny fraction of the ocean.Under these circumstances,we have to use search engines(Bai Du?Google etc.)to find out what we really want.However,there is an obvious restrictions in search engine,we must know the exact information of the object to be searched in advance.In real life,people always can't know exactly the information they want to search or the product information.At this time,we need a professional professor offering us some useful information and the professor is actually a recommendation system.On the one hand,it help users find things they might like without having a clear search object,on the other hand,it is able to help the provider of goods to push the goods to a more accurate target population.Firstly,this paper introduces the research and development of recommend system at home and abroad as well as its comprehensive application in different fields,and this paper also introduces the classification of recommendation system from the Apriori algorithm.At present,collaborative filtering recommendation system is widely used,and content-based recommendation and hybrid recommendation are more commonly used.Moreover,in order to evaluate a recommendation system,several commonly used evaluation indexes are also introduced such asprecision rate and the recall rate.For the two most common algorithms in collaborative filtering——Item CF and User CF,we also put up the improvement methods and make a comparison.Besides,the paper introduces the other recommend system——LFM and give a new algorithm——K-Means+LFM which improves the result of recommendation.Lastly,This paper discusses some existing problems in the recommendation system,which will be the future research direction.