Research On User Experience Analysis From Multiple Data

With the progress of science and technology,the software is changing people's life remarkably,and software companies pay more attentions to Experience of software.Software companies deeply dig user information to improve their software,improve User Experience and making their software more competitive.Analyzing User Experience accurately can improve User Experience and extend the life of software.Analyzing User Experience involves evaluating User Experience and analyzing the evaluation.Assessing User Experience accurately can get a real User Experience and provide reliability for analyzing User Experience.At the same time,the user group can be found by acquiring the user characteristics related to User Experience,and User Experience and software benefits can be improved by acquiring and inproving the software factors related to User Experience.In addition,since the research samples in the field of User Experience are scarce,the results analyzed on scarce samples are not accurate enough.Therefore,the purpose of this paper is to evaluate User Experience accurately,analyze the relationships between User Experience and related factors fully,and propose an accurate and interpretable nodel that can be used in scarce data of User Experience.Due to subjective methods or objective methods can only evaluate User Experience through one side.This paper improves evaluation method based on a combination of subjective and objective lethods and evaluates User Experience by this improved method.At first,this paper designs a questionnaire that containing User experience and user characteristics to obtain subjective experience,and then obtains the relationships between User Experience and related factors.Finally,the Heart Rate Variability is used to verify the above relationships and the subjective experience.Existing methods for analyzing the relationships between User Experience and related factors are statistical learning methods and Machine Learning nethods.However,existing research using statistical methods and machine learning models only obtain partial User Experience relationships and the accuracy of existing Machine Learning models for analyzing scarce samples in User Experience field is not accurate enough.In order to fully explain User Experience,this paper chooses the interpretable RIPPER algorithm.In addition,this paper improves the formula of FOIL formula in rule learning for addressing the problem of lacking specific value samples and learns the relationships among User Experience and user characteristics and software factors by the improved RIPPER.The existing methods for analyzing scarce samples are Machine Learning methods analyzed on scarce samples and Transfer Learning that use auxiliary data to improve the knowledge of scarce samples.This paper introduces TrAdaBoost algorithm for the first time into the field of User Experience to improve the accuracy for analyzing scarce sample.In order to improve the accuracy of analyzing scarce samples as well as explain the relationships among User Experience and user characteristics and software factors,this paper proposes Transfer in Cart(TrCart)algorithm based on interpretable Transfer in Decision Tree.In addition,This paper combines instance-based TrAdaBoost with model-based TrCart for the first time and obtains the interpretable Transfer AdaBoost in Cart(TrAdaboostCart)algorithm with high accuracy in scarce data of User Experience.