A Research On Event Recommendation Models Based On Internet Data Of Sports Events

With the rapid economy booming and the change of society ideology,the current enthusiasm of people to participate in sports events is increasing.With marathon events as an example,there have been more than a thousand competitions in the country last year.However,the complicated classification standard and the uneven supply of events information also increase the difficulty and time cost for runners to select events.How to make better use of a large amount of existing events network data and resolve the overload of events information circulation has become a potential application problem to promote the digital development of sports information.Based on this problem orientation,this paper chooses to research the common characteristics of current Internet data of sports events and corresponding applicable event recommendation technologies,and build a feasible event recommendation model as the main solution.This article explores the characteristics of current events and compares recommended technologies suitable for Internet event data.Taking the marathon data of Iranshao website as an example,the algorithm selection and model construction of the corresponding event recommendation model are discussed,and it is determined that the content-based recommendation model and its three key algorithms are determined to achieve the goal of constructing the event recommendation model.On the basis of the above,further research the construction framework of the event recommendation model,analyzes the steps of event vectorization and matrix of the similarity of the event,and take experimental comparison on the collected marathon event data set.from the results,the constructed event recommendation model has better performance in the recommendation of marathon events,which verifies the feasibility of the content-based recommendation algorithm in recommending event information.This model can effectively meet the needs of existing people for marathon recommendation,and also provides technical support and theoretical basis for the research on building an effective Internet data processing mechanism and event recommendation model for sports events.The following is a summary of the research content of this article:(1)This paper research the current Internet data of sports events,which are characterized by huge amounts,diverse types,fragmentation,and low relevance.And build a general event data collection framework and method.Taking popular marathon events among the people as an example,the similarities and unique characteristics of network data are explored and analyzed to provide basic data support for the construction of event recommendation models.(2)The classification and evaluation criteria of current commonly used recommendation technologies are studied,and a content-based recommendation method is selected according to the target and feasibility of the event recommendation.Based on the analysis of Internet data characteristics of sports events,the LDA theme model,TF-IDF text weight model,and Word2 vec sequence model are selected as the key algorithms for the event recommendation model.(3)Research on the basic event recommendation model building framework,focusing on the completion of the event vectorization and event similarity calculation under the LDA theme model,TF-IDF text weight model,Word2 vec sequence model,based on the advantages and disadvantages of the algorithms,propose and build a comprehensive event recommendation model based on TF-IDF model and Word2 vec sequence model.This article believes that the best recommendation results can only be achieved by selecting a suitable event recommendation model based on the model input data set and application scenarios.The comprehensive event recommendation model can obtain the best recommendation performance on multiple event datasets,while the TF-IDF-based recommendation model and the Word2 vec sequence-based event recommendation model can obtain better recommendations performance on a single input data set.(4)Four event recommendation models were constructed and applied to the collected marathon events data for experiments.The experimental comparison verified the validity of the content-based recommendation algorithm technology on the event network data set,and obtained suitable algorithm model for marathon events.Experiments show that the comprehensive event recommendation model based on TF-IDF text weights and Race2 vec entry sequence has the best recommendation performance on the marathon event dataset;according to the event's recommendation goals and events data type characteristics,a single or comprehensive recommendation algorithm can be selected Build a model to implement correct recommendations.In a single algorithm model,the event recommendation model based on the Word2 vec sequence and the event recommendation model based on the TF-IDF text weight can achieve better recommendation results on the respective user entry dataset and the event details dataset.This result validates the performance evaluation assumptions of the recommended models for each event constructed in this paper.