Situational Awareness Oriented Intent Research And Recommender System Design

With the rapid development of computer technology,the design of a human-computer interaction interface for complex information systems has received significant attention.The merits of the interface design are directly related to the effectiveness of the operator’s decision-making.In a situational awareness environment,the human-computer interface often contains multi-dimensional information such as target identification,intelligence analysis,threat judgment,and situational assessment.However,current complex information systems are often not fully integrated with human factors theory in their design,and operators are very prone to leakage,misinformation,and information overload.At the same time,along with the rapid development of artificial intelligence,big data,and other information technology,the ability to obtain and share information in the situational awareness environment has been dramatically enhanced.The excessive situational information presented in the interface within a short period can seriously interfere with the user’s retrieval and effective use of essential information,increasing the risk of cognitive overload.Taking the battlefield situational environment as an example,this paper designs a situational recommendation model with high accuracy,diversity,and completeness to address the cognitive overload problem when users interact with the system and summarizes the corresponding design strategies to enable users to focus on high-value situational information,shorten the cognitive and decision-making time and reduce the cognitive load when facing the vast amount of situational data.At the same time,the intelligent recommendation system designed in the study can obtain the "implicit intent" of users that is difficult to get by traditional methods based on the user’s historical interaction records and the intrinsic correlation between gesture information and can enhance the naturalness of human-computer interaction using intelligent push.The main research content and innovation points of the paper are as follows.1.To address the problem of "intent recognition," we conducted a study on situational awareness requirements and defined the commander’s intent in the battlefield situational awareness environment as explicit and implicit intent.By designing a recommendation system based on situational knowledge mapping,the identification and mining of the above two kinds of the commanders’ intent are realized.2.Taking battlefield situational recommendation as an example,504 battlefield situational elements and inter-element entity relationships are identified,and the historical interaction data and situational knowledge map of battlefield commanders are designed in this way.The task of intention identification and situational recommendation for commanders is completed.At the same time,based on the standard evaluation algorithms of recommendation systems and the actual situation recommendation environment,we determine the situation relevance,situation completeness,and situation diversity as the evaluation indexes of the model to evaluate the merits of the model and start the optimization experiments of the model.The advantages of graph neural networks in capturing node relationships in graph-structured data are exploited to perform the gestalt recommendation better.The recommendation system is designed based on three classical graph neural network recommendation models: Pin Sage,GCMC,and NGCF.Two different knowledge graph enhancement recommendation methods make the system more diverse,accurate,and interpretable.We also analyze and discuss the structure of the model through experiments,optimize the over-smoothing problem and small sample problem encountered in the experiments,and finally propose a gestalt recommendation design strategy that includes model design and recommendation process design.