Research On User Access Pattern Mining Based On Web Log In Big Data Background

With the rapid development of the Internet,the scale of network users continues to expand,the proportion of network users continues to increase,and the behavior of network users becomes more and more complicated.Although many scholars have conducted in-depth research on Web mining technology and user behavior analysis,there are still some problems.Traditional data mining algorithms are not satisfactory in terms of big data processing efficiency.At the same time,in the field of ecommerce,with the increasing popularity of online shopping,the increasing number of Internet users and the emergence of new products,a large number of goods and users do not interact with each other,the system is more inclined to recommend popular products to users,resulting in most There are repeated recommendations in the e-commerce platform,and it is impossible to provide more accurate personalized services,as well as the discovery of long-tailed items and the weight adjustment of items.These problems urgently need to be solved.Therefore,this paper chooses the field of e-commerce,based on the data characteristics of e-commerce background log,focuses on analyzing and researching e-commerce user behavior,and constructs the basic mode of user behavior analysis.The research content of this paper mainly includes the following points:Firstly,based on the deep understanding of Web user behavior theory,the user's behavior is classified based on the content of interaction.At the same time,based on the big data background,based on the introduction of some traditional data mining algorithms,the paper further expands to In line with the reality of commercialization,in the case of Web logs,under the influence of big data environment and various user behaviors,the collection methods and processing methods become more complicated.Based on this,this paper deeply studies and understands Web users.Behavioral characteristics and data from Web logs represent and summarize the characteristics of Internet user behavior.Secondly,based on the above research results,for the application scenarios in the era of big data,parallelization processing is carried out on the basis of the improvement of traditional algorithms,which greatly improves the operational efficiency of the algorithm,and adopts a distributed file storage structure to improve the system data processing.Fault tolerance.At the same time,the advantages and disadvantages of the collaborative filtering recommendation algorithm are deeply studied.Collaborative filtering currently has a very wide range of applications.On the basis of not changing the collaborative filtering algorithm,it introduces migration learning and proposes a cross-domain migration based on tag sharing and user interest.The experimental results show that compared with the existing collaborative filtering recommendation algorithm,the accuracy of the recommendation model based on migration learning is improved to a certain extent,and it also helps to solve the cold start problem caused by data sparsity.At the same time,the existing recommendation algorithm in the original system is very invasive to the algorithm itself.Finally,based on the above research contents and results,the Web user behavior mining system constructed in this paper can carry out multi-dimensional and highefficiency mining.Through accurate marketing and accurate advice,it helps ecommerce merchants,content providers,etc.understand their users and achieve better business value,and complete the upgrade of data-driven services.