Research On The Method Of Query Intent Mining For E-Commerce Platform

E-commerce platform is a network platform integrating search query,online transaction,advertising and publicity.It permeates all aspects of people's life.In order to improve its customer satisfaction and enhance the user experience,the functions of ecommerce platforms,such as the identification of query intention,the guarantee of network environment security,and the optimization of user credibility evaluation,have become particularly critical problems.Query intent refers to the user's need for information.Rapid identification of query intention and rapid display of user satisfaction results according to the query intention are particularly important aspects in the function optimization of e-commerce platform.Therefore,the analysis,mining and identification of user query intention based on e-commerce platform has become one of the hot topics in data mining technology.The current classification of query intent into information,transaction,and navigation classes is applied by the academic.In recent years,some researches have been focused on the extraction of query intention according to the hot spots in the search log,the automatic identification of user query intention by machine learning,the multidimensional analysis of search data and the extraction of query intention.Due to the query intent is based on semantic information,and semantic information is fuzzy,so the research on query intent recognition is often also related with the study of fuzzy sets.Classification according to query intention can accelerate the mining and identification of query intention,but there are also multiple attribute decision problems.In the study of multi-attribute decision making problems,multiple operators of intuitionistic fuzzy sets in fuzzy sets can be used to analyze decision problems with different attribute weights.These research results promote the development and application of query intention mining.However,they cannot obtain sufficient information of query intention,such as the correlation between weight and query intention when query intention is a multi-attribute decision problem.Loss of query intention when semantic information is converted into fuzzy information;for the e-commerce platform,whether the user intention mining function is perfect or not.Therefore,it has great theoretical research value and practical application meaning to the extension,application and promotion of user intention mining based on fuzzy sets.A study of the following aspects is developed in this dissertation:(1)The research on query intention mainly focuses on the feature recognition of query intention and the classification of query intention.The current research has a problem with the integration of these two aspects,it has a great effect on the integrity of the query intent classification system.To solve this problem,this paper proposed a classification method of query intention based on hierarchical strategy.The classification method first preprocesses the query string proposed by the user,then the query intention is classified by hierarchical strategy: At the first level,according to the relationship between the query intention in the query log and the query subjects,this paper proposed a theme-oriented framework to cluster the query subject hidden in the query log through this framework.The second level quantified the query feature text.At the third level,this paper proposed a model of classification of query intention based on support vector machine(SVM)to query log.The experimental results show that the classification effect is remarkable.The classification accuracy,the classification speed and the fault tolerance of the model are all improved.(2)User's query intention is a typical fuzzy uncertainty problem.Since the intuitionistic fuzzy set needs to consider three factors,such as membership,non-membership and hesitation,so using fuzzy sets and the extension of intuitionistic fuzzy sets to deal with the fuzziness and uncertainty problems has more flexibility and effectiveness.In the processes of fuzzy multi-attribute decision making,even though Hamacher fuzzy integrator has a clear field,the parameters don't affect the characteristics of the results,but in the parameters of the parameters,it's easy to see the value of the extreme,which is not suitable for the real world.To solve this problem,this paper introduced the concept of fuzzy measure with Choquet integral to avoid this defect.In this paper,the integral of the Choquet integrator and the Hamacher integrated operator,based on the algorithm of the intuitive fuzzy set operator,this paper proposed an intuitive and fuzzy Hamacher Choquet mean operator(IFHCA operator),and illustrated the good properties of the operator.Finally,based on the intuitionistic fuzzy set(IFHCA)operator,a user query intention topic recognition method based on multi-attribute decision making is constructed.(3)Based on the identification of the user's intent,if the user submits a sensitive query,it needs to be judged on the user's credibility.Due to network user behavior has the characteristics of unpredictability,the credibility of e-commerce platform for the user or the user currently operating hierarchies are not clear,to some extent this obscures the potential danger.This potential risk factor is greatly reduced if the reliability evaluationof each user's numerical type is obtained based on the user's past or present operational behavior.In order to solve this problem,this paper first defined eight attributes to evaluate user credibility,and then used binary semantic information to evaluate the credibility of user activities in combination with Bonferroni mean operator to fully consider the advantages of the relationship between the two input attributes.In this paper,the binary semantic generalized Bonferroni mean(2TLGBM)operator and the binary semantically weighted Bonferroni mean(2TLGWBM)operator are defined.Based on these two operators,a method to evaluate and make decision by using binary semantics is proposed.Finally,a numerical example is used to show the user credibility.The experimental results show that the method is simple,effective and easy to calculate.(4)This paper proposed a method for functional evaluation of e-commerce website query intention mining.A variation of analytic hierarchy process is proposed in which fuzzy logic is used as formal semantics.The triangular fuzzy number is the criterion.The hierarchical model is constructed according to the formal scheme of the evaluation problem.This paper also presents a fuzzy tomographic analysis algorithm for calculating weights(for attributes and candidates)and for making decisions(sorting alternatives).Finally,the evaluation of query intention mining function based on fuzzy hierarchical analysis is realized.The experimental results verify the validity of the proposed solution and model.