An Empirical Study Of GUI Development Based On Question And Answer Discussion

Graphical User Interface(GUI)has become the most effective human-computer communication technology today,and a good GUI is crucial to the success of software applications.However,the development and implementation of GUI is challenging.Developers not only need to understand the interaction of various components,but also must follow many design and implementation principles and constraints.Analyzing Q&As related to GUI development on the Q&A community can help understand the challenges these developers are facing.Stack Overflow is one of the most representative technical question-and-answer communities.The website is dedicated to providing developers with a platform to ask questions and answer questions,serving as a knowledge base for developers’ various needs.So far,there is no GUI development research based on the Q&A discussion community Stack Overflow.Therefore,this article analyzes 23,741 Q&A discussions related to GUI development on Stack Overflow.Through GUI empirical research,it helps to understand the main problems and challenges developers encounter in the development process.Through the identification research of GUI invalid answers,it helps developers to quickly identify invalid answers when browsing websites and improve development efficiency.The main contributions of this paper are as follows:(1)GUI empirical research based on LDA(Latent Dirichlet Allocation)model.This paper proposes a set of topic model analysis methods based on question-and-answer discussion,which mainly includes four steps: data acquisition,data preprocessing,topic modeling and data analysis.In terms of data acquisition,we first determine the coverage of GUI-related posts and collect relevant datasets from Stack Exchange Data Dump.In terms of data preprocessing,we remove stop words,build Bi-gram models,and restore lemmatization to the data.In terms of topic modeling,we use the LDA model to model the topic of GUI-related question posts and comment posts respectively.In terms of data analysis,we analyzed the topic taxonomy of GUIs,the popularity and difficulty of each topic,and the problems developers encountered when reusing answers.The research found that the topic "tool selection" was the most popular topic,and the topics "GUI automation" and "client/server" were more difficult,GUI developers had wrong answers when reusing Stack Overflow answers,answers did not match requirements,and the phenomenon of invalid answers.(2)Aiming at the invalid answers encountered by GUI developers when reusing answers,conduct research on the identification of GUI invalid answers.This article first builds an automated tool to identify invalid answers in GUI-related posts on Stack Overflow from two aspects of invalid APIs and invalid URLs,effectively helping developers quickly determine whether the answers used are invalid when browsing the web.Then we analyze the time flow of invalid GUI answers,and provide actionable suggestions to Stack Overflow and developers to improve the quality of answers in the Q&A community.