Research And Implementation Of A User Attention Measurement Method On WEB Survey

In recent years,with the popularity of the Internet,the use of web survey is more and more frequent and extensive,and the control of the data quality of the questionnaire has always been an important subject for investigators.Web survey provides an interactive mode for network users.With the development of psychological science,research topics such as collecting users' mouse track and other behavioral information to analyze the data quality in web survey,identifying users' confused emotions,and helping surveyors improve questionnaire design have gradually become popular.Based on the existing research on the problems related to the web survey and mouse track,and combined with the machine learning method,this paper proposes and implements an analytical method for measuring the user attention of web survey.The user's attention in the web survey was modeled and analyzed,and the feature engineering and classification methods of machine learning were used to measure the user's attention in answering the questionnaire(i.e.,conscientiousness/unconscientiousness).The main work of this paper includes the following two aspects:(1)First network user behavior modeling in the web survey,detailed describes the behavior of the users in completing the survey may be,define the behavior of the collecting data and web survey measure problem users attention,formulated to describe the problem of input,processing,and output.Input is the user's behavior information data,including the speed,acceleration,distance,angle change related to the mouse track,mouse click,scroll and answer time interval,etc.The processing process is a binary model of machine learning.According to the user behavior information,the user's attention in answering the survey is identified and measured,and then the user's attention is divided into two categories(i.e.,conscientiousness/unconscientiousness)as the output of the question.(2)On the basis of the definition of user behavior modeling and problem,put forward in this paper,in view of the web survey user attention measuring model and method,through the design and implementation of web survey website and the mouse trajectory collection module,experimental design and implementation of web survey to collect users' data,using machine learning method to solve the problems of the defined,evaluation and validation.Firstly,the data are preprocessed,and then the processed data are classified into dimensionality reduction.Then,performance indexes such as precision,recall and F1 measurement are used to evaluate and compare the effects of different dimensionality reduction methods and classification methods,among which the method with better effect is selected as the method to solve the problem model.Through the design and realization of web survey collection module and user behavior,a total of 578 valid user data collected,the data pretreatment,dimension reduction,classification and evaluation,the experimental results show that the chi-squared and gradient of the decision tree effect is relatively good,the precision,recall and F1 measure were 78.1%,84.25% and 81.06%,respectively.At the same time,it is also compared with other methods to analyze user behavior data.The results show that the method in this paper is superior to the comparison method in all performance indexes,with an average performance improvement of 16.11%.This article from the network user interaction with the web survey this point of view,on the basis of literature study,thinking and analysis of a user in response to the web survey,based on user behavior such as the mouse trajectory information using machine learning method,to help investigators to improve the questionnaire design and control of data quality,has a certain practical significance and practical value.