Text Mining Based On Consumer Demand For Used Electronic Equipment In Taobao

As second-hand market prospects,due to the chaos in second-hand market trading and information asymmetry,consumers still have doubts about the second-hand goods which have absolute price advantage.According to a survey of Chinese Consumers Association,more than half of the users in China have two or more mobile phones,which indicates that the replacement rate must be in a high level and the rising problem of how to deal with second-hand mobile phones.Secondary sales on mobile phone are the preferred approach to the public.Trading information on secondary platform can reflect concerns and needs of consumer on second-hand mobile phone business.This paper analyzes text mining on text data of secondary trading from taobao by using natural language processing technologies,explores the consumer’s demand status on second-hand electronic equipment from taobao,and gives multi-dimensional user portrait of the secondary market consumer groups.The main contents of this paper are as follows:Firstly,the local hash sensitive algorithm is used to identify the similarity text matching and calculate the similarity coefficient of Jaccard,retain only one message for each group which text matching with a similarity score is more than 0.1,and filter the rest.Meanwhile,extract the proper noun dictionary of the electricity supplier,computer and mobile phone from the cell thesaurus,then load the word segmentation tools to carry out high quality word segmentation.Secondly,the topic model is used in this paper to capture the information that the consumer is more concerned about and the topic is divided into three categories,which are phone parameter settings,mobile phone status and the state of the transaction.The key words of mobile phone parameters set are dual card dual standby,system type,battery capacity,etc.The key words of mobile phone status are the purchase of the time,the use of time,warranty time,whether there is an receipt,etc.Key words in the transaction state are face to face trade,whether the package mail and telephone contact.Thirdly,the Hownet emotion dictionary and the comments in the dictionary are combined as the emotional words,while adding a part of the custom entries to do emotional tagging set at the same time.From the results of emotional tagging point of view,positive emotion text scores more than negative one.The emotional polarity is about 0.66.Overall market’s sentiment is weakly positive.In this paper,the deep learning algorithm is used to split second-hand trading text into vector for words.Fourthly,this paper constructs the user value evaluation system of the secondary market based on the heat of discussion,customer loyalty,topic density,emotional intensity and use value.The maximum expectation algorithm is used to divide the crowd into the wait and see user,the reserve user and the value user.Finally the three types of people are discussed from the prospects of location,emotion,topic focus,heat,loyalty and so on.