Research And Implementation Of Personalized Incremental Slope One Algorithm

With the advent of the era of big data,in today's era,a large amount of data is generated every day.How to make more effective use of this information is a major problem.The current recommendation system arises at the historic moment.Based on past information collection and analysis,it can quickly find the needed information from the vast amounts of information,and make the personalized recommendation,so as to improve customers' loyalty continuously.Recommendation system as an information processing system can be roughly divided into three types: content-based recommendation,collaborative filtering recommendation and recommendation based on the model.In 2005,Dianl Lemire proposed a slope one collaborative filtering algorithm,which has the advantages of simple algorithm,easy implementation,high efficiency and accuracy,but it does not take into account the individual differences between users.We have to admit that everyone has their own personality,that is,the so-called similarity refers to the similarity in a certain aspect.Therefore when we use the overall similarity difference to predict the user's difference in some aspects,there will be a big error.Take the recommended movie as an example.The movie types that user A likes to watch are science fiction and martial arts movies,while user B likes to watch science fiction and love movies.When we predict the user B's score for science fiction movies,we calculate the difference between user B and user A.The difference between users A and B in the non sci-fi movies that jointly evaluate and the sci-fi movies is unknown,because we only know that these two customers like sci-fi movies,while the preference of other movies is unknown.Therefore,when using slope one algorithm to calculate the difference between two users,the non sci-fi movie score may cause errors.Therefore,this paper proposes a personalized slope one recommendation algorithm,which only calculates the difference between user A and user B in sci-fi movie when predicting user B's sci-fi movie,reduces the noise as much as possible,and makes the prediction more accurate and effective,At the same time,it also reduces the computation of the model,improves the running speed,and uses the dynamic update method to implement the change of recommendation information,so as to achieve the accuracy,effectiveness and real-time of the recommendation algorithm.