Research On The Recommendation Algorithm For Improving The Similarity Of Collaborative Filtering Based On Genetics Algorithm

In With the continuous development of Internet technology,the way of interaction between people and information is also constantly changing.From the initial Internet classified information website to search engine technology,people are looking for information on their own initiative.However,the information presentation index Level of growth,people are also more and more difficult to access and filter information.The recommendation system effectively reduces the difficulty of obtaining information.It provides tailored information to the user from the user's preference,which not only facilitates the acquisition of user information,but also helps the user to automatically filter out the useless information.Currently recommended system has become a very important direction for Internet applications.As one of the most classic algorithms in the domain of recommendation system,collaborative filtering algorithm is still playing an important role in industrial applications.Related research articles are also emerging in an endless stream,which fully expands the research field.However,there are still many unresolved problems in the collaborative filtering algorithm,such as the data sparsity,similarity calculation and cold start of the user-item scoring matrix.Context-aware recommendation systems are also a focus of current research.Scholars have put forward many research models,but in the face of the diversity of contextual information,there is still not a common solution.In this paper,some improvements are proposed for the model construction of similarity calculation and context-aware recommendation system in collaborative filtering algorithm.The main work is as follows:On the one hand,the genetic algorithm is added to the similarity calculation process,and a collaborative filtering recommendation algorithm based on genetic algorithm to improve the similarity is obtained.Through the selection,crossover and mutation process of genetic algorithm,the traditional similarity calculation method is optimized to obtain better similarity value.The algorithm can effectively reduce the accuracy of recommendation.At the same time,due to the adaptability of genetic algorithm and the non-derivative property,the scale of calculation is simplified.This experiment is validated by experiments.On the other hand,in the context-aware recommender system,due to the incorporation of contextual feature information in the recommendation process,the matrix is sparse in the matrix representation.With the use of the currently popular Factorization Machine(FM)At the same time,the concept of layering is introduced.Based on the hierarchical factorization machine,contextual features can be effectively used and the recommendation accuracy is improved.In this paper,the algorithm is implemented on the basis of two improvement studies,and an experimental comparison is made.The experimental results show that the improved algorithm has a better recommendation effect,and compares and verifies the accuracy of each index.