| With the rapid development of online retailing,many e-commerce giants have entered the fresh food field,which diversified the fresh product sales channels and promoted the accelerated development of fresh food e-commerce greatly.At the same time,along with the improvement of consumption level,people put forward higher requirements for product quality,high-quality fresh products gradually become the pursuit of consumers,but fresh products have the characteristics of easy corruption and difficult to preserve,in such a broad market demand,if fresh e-commerce enterprises can’t grasp the product demand accurately,it is easy to cause product defects,corruption and other problems,thereby reducing customer satisfaction and increasing business costs additionally,therefore making accurate forecasts of product demand is an important way to solve such problems.At present,external environmental factors such as historical demand data,weather changes,and promotions were the main factors considered when predicting the demand for fresh products.However,with the development of social media,consumers often express their experience with the products on the online comment area of e-commerce platform after purchasing products.When potential consumers are ready to buy a product,browsing the online review content becomes an important reference for them to decide whether or not to execute their purchase decision,thus to a certain extent the final demand of the products is influenced.Therefore,how to mine the implicit customer perception information based on the content of online reviews,identify the product features that customers are concerned about,and use the factors to predict product demand has become a problem that fresh food e-commerce companies should pay attention to.With the goal of grasping the changing trend of fresh product demand and improving consumer satisfaction,conducted a study on the factors of product characteristics perceived by customers in online reviews and demand forecasting for fresh products.The main contents are as follows:(1)Emotion analysis.Based on the How Net’s sentiment dictionary,the sentiment analysis dictionary of fresh products was constructed by combining the characteristics of fresh products,and Python software was used to calculate the sentiment value of each online review content.(2)Factors extraction.The Word2 vec model and NLPIR Natural Language Processing System were used to extract the factors of product characteristics perceived by customers in online reviews and use them as input variables of the demand forecasting model.(3)Demand forecast.Based on the customer perception factors that affect customer satisfaction extracted from the online review content,a multivariate support vector regression model was introduced to construct a multivariate support vector regression prediction model so as to forecast the demand of fresh products.The perceptual factors that affect customer satisfaction implicit in online reviews has been extracted,and the factor system was used to build a multivariate SVR demand forecasting model for fresh products.Moreover compared with the traditional SVR forecasting model,the model constructed in the paper improved the forecasting accuracy of fresh product demand effectively,thus providing an important reference for fresh produce e-commerce enterprises to grasp the trend of product demand changes accurately. |
