Design And Impl Amementation Of Automatic Su-Bsystem Of Customer Service Emotion Diction-Ary Based On Word2vec

With the development of artificial intelligence,more and more enterprises have launched intelligent quality inspection projects.Robots are used to automatically detect the emotional changes of customers and customer service personnel through the chat records of customer service,providing great help for the development of enterprises.But robot intelligent customer service involves emotional analysis and other issues,so it is urgent to build an emotional dictionary to provide basis for the company's emotional analysis project.The traditional manual construction of emotion dictionary is not only time-consuming and laborious,but also too one-sided to meet the needs of emotion analysis.The conventional automatic dictionary building method is too single,the quality of emotion dictionary building is poor,and the operation of building multi domain dictionary is too complex,so it can not quickly cover all the service areas of the company.Aiming at the above key problems,this paper will design and implement an emotion dictionary construction subsystem based on word2vec,which can assist users to build their own emotion dictionary by mining useful information in user chat corpus through a variety of algorithm combinations and mutual cooperation.The intelligent customer system designed and implemented in this paper,users can upload the chat corpus of different industries,through this system can get the domain emotion Dictionary of different industries very conveniently and quickly.The emotion dictionary construction subsystem designed and implemented in this paper mainly includes word vector construction,positive and negative two meaning emotion dictionary construction,multi emotion dictionary construction,auxiliary construction of basic positive and negative two meaning emotion dictionary and auxiliary construction of basic multi emotion dictionary.Word vector construction is the basis of all functions of the system,which mainly maps words to high-dimensional space.Based on word2vec,three algorithms are used to mine the positive and negative biases of emotion words from the user's chat corpus.The construction of polysemous dictionaries is also based on word2vec to mine the preference and weight of emotional words in the corpus.Auxiliary construction of basic positive and negative emotion dictionaries and auxiliary construction of basic multi emotion dictionaries are both to assist users in building a user-defined basic emotion dictionary.The function of basic dictionaries is to assist users in building a complete positive and negative emotion dictionary and multi emotion dictionary.Finally,the system is developed and tested.The test results show that all the functions of the system can operate correctly.According to the user's expectation of chatting,the positive and negative two meaning dictionaries and the multi feeling dictionaries are generated,which can also assist the user to build a custom basic emotion dictionary.Finally,the generated emotion dictionary is applied to emotion analysis,and the test results achieve the expected results.It shows that the design and implementation of the emotion dictionary building subsystem meets the expected requirements.At present,the emotion dictionary building subsystem implemented in this paper has been successfully launched,which can provide help for users to build emotion dictionary.