Research And Design Of A Model Of Generative Chatbot For Different Personality Characteristics

In the context of intelligent chatbots replacing human online replies on major platforms,humans have entered the artificial intelligence’s era in a broad sense.The traditional template matching and retrieval approach to content generation in chatbots has been supplanted by the real-time generation method of deep learning model algorithms,particularly with the emergence of large-scale deep neural network learning models.The application’s ubiquity is augmenting its efficacy,and chatbots are gaining traction,thus making it a highly sought-after area of research for artificial intelligence.Based on Jung’s theory of psychological typology,psychologists proposed that human psychological theory research can be described by four dimensions: the direction of energy flow,the way of perception of obtaining information,the way of decision-making about processing information,and the way of contact with the surrounding world.Guided by this theory,they developed the MBTI(Myers-Briggs Type Indicator)personality type scale.At present,the generative chatbot technology based on deep learning and natural language processing technology is relatively mature,and the content generated based on the training corpus is relatively smooth and accurate.In the context that the reply content is relatively acceptable,researchers are more pursuing the anthropomorphization of chatbots,and pursuing more diversified and more matching content for generating replies.Based on the same theoretical and technical principles,combined with the relevant knowledge of MBTI personality type,the chatbot reply generation model and the re-analysis of the reply content after generating the reply content are analyzed and studied,and a chatbot model is designed on this basis,namely personality characteristics and Sequence to A generative chatbot model with personality characteristics obtained by fusing the Sequence model.The specific research work is as follows:(1)It is proposed to determine the personality characteristics of chat subjects through the text information used by chat subjects,and change the research and judgment of personal personality mainly in the form of questionnaires in the past.Firstly,the model training of the classification algorithm is carried out by the professional MBTI dataset and machine learning classification algorithm,and the text information of the chat subject is encoded and sent to the machine learning classification algorithm for personality prediction.(2)A personality feature vector is proposed.According to the MBTI dataset,the text information of the same personality is integrated,and the corresponding personality feature vector is generated by the Word2 Vec algorithm,which is used for the training of the subsequent chatbot reply generation model.(3)The background of personality characteristics that matches the personality of the chat subject for the chatbot is proposed.Just as neural networks refer to human brain nerves,we should consider the most important personality factors for humans to the background of chatbots to train chatbots’ response generation models.When the model is trained,according to the personality characteristics of the chat subject,the personality feature vector that matches the personality characteristics of the chat subject is selected to be embedded in the reply generation model of the chatbot,in order to generate the reply content with the background of the personality characteristics.and assessing whether the model has a background in personality traits.