Research On Development Platform Of Intelligent Dialogue Software Help System And Its Application In Design

With the rapid progress of artificial intelligence theory and technology,the attention of various circles on artificial intelligence continues to increase.At the same time,software technology is also driving progress in various fields including instrumentation,and software systems have become an indispensable part of most products.However,the gradual increase of software system functions has caused problems such as the difficulty of using software and low utilization rate.Therefore,the integration of artificial intelligence technology in the process of software use assistance can not only improve the level of software assistance,but also help the rapid development of the industry.After analyzing the research status of software help system and human-computer dialogue system,this thesis proposes a model of ellipsis ambiguity processing based on software real scene,and then builds the development platform of real scene-based intelligent dialogue software help system.It mainly includes the two processes of development and dialogue help based on software real scene.Firstly,for the development process,combined with HOOK technology,a method for obtaining software real scene is designed and implemented;a software real scene dictionary is generated,which includes establishing the a description form of software function button position relationships based on a tree structure,and defining the software real scene-based interaction generation template to automatically generate"question-answer pairs";a conceptual subordinate tree of related knowledge is constructed,and then a domain knowledge base is builded preliminarily,so as to accurately understand the intent during the dialogue help process.Secondly,for the dialogue help process,by combining the knowledge base and the BERT language model based on deep learning,a software real scene-based dialogue ellipsis ambiguity processing model is established,where the real scene includes the software's actual operating environment,dialogue context,and knowledge.In this process,feature importance,semantic mapping weights,and real scene mapping weight are defined;a text similarity calculation method suitable for this study is proposed to calculate the matching degree between the natural language input by the user and the corresponding feature combination in the software real scene dictionary;then,the methods of semantic parsing and sorting are proposed to determine the relatively optimal natural language understanding result,so as to provide corresponding help for the user.Finally,the development platform of real scene-based intelligent dialogue software help system is applied to the design of the virtual instrumentation,and a real scene-based intelligent dialogue prototype help system is developed for Lab VIEW software,which can provide corresponding assistance on software function button searching,knowing and operating intent on the basis of accurately understanding user's need,thereby improving user experience and satisfaction.