User Requirement Elicitation Method Based Multiple Rounds Dialogue

Currently,with the sweeping progress of artificial intelligence,software service bot,which has the human way of thinking,has become the future trend.As a research hotspot in the field of service computing,software service bot has been developed in some fields.It is an important goal of software service bot to perceive and automatically identify explicit and invisible requirements of users,and to actively,efficiently,accurately and safely provide service solutions to meet customer needs.However,as an important issue in the field of service computing,the elicitation and cognition of users' uncertain requirements bring great challenges to the realization of this goal.In the aspect of requirement elicitation,the software service bots still need to further understand the users' uncertain requirement expression accurately,such as getting rid of the template and formalized knowledge structure and enhancing the reasoning ability of the service.Therefore,the core work of this paper is to actively perceive the multi-source heterogeneous rich semantic requirement knowledge base to support the reasoning analysis of the user demand expression in the interaction process,and then identify the user's complete requirement proposition,so that the software service bots could actively provide accurate and efficient services for the user.Around this work,this paper would mainly study the following aspects:(1)In order to improve the efficiency of interaction between the software service bot and the user,this paper purpose a knowledge processing method based on the theory of granular computing,and a method of automatic determination of the next round of inquiry content guided by the software service bot in multiple rounds of dialogue after knowledge processing in the off-line state and before the dialogue begins.Multiple rounds of dialogues used to elicit user requirements are guided by knowledge-based reasoning,but not based on predefined domain rules.(2)In the face of different application scenarios with different requirements characteristics,this paper proposes a set of framework in which the software service bot uses different multi-round dialogue strategies.In the framework,the software service bots use the rich semantic knowledge base of different structures to capture user requirements.(3)Since the personalized expression of user requirements is an important factor for the software service bot to accurately understand the user requirements,this paper attempts to add the emotion recognition results of multi-modal fusion to influence the interactive dialogue strategies with users.The experimental results show that,compared with existing methods,this method could reduce the restrictions of the input and successfully combines groups requirement modeling of abstract expression and user personalization requirement expression,significantly enhance the accuracy in user requirements elicitation.Meanwhile,the method could let the user would not have poor experience due to too many dialogue rounds in the process of multi-round dialogue with the software service bot,and improve the efficiency and performance of service recommendation effectively.