Research On Product Identification And Its Correlation In Online Foreign Trade Community

The rapid development of cross-border e-commerce has attracted widespread attention from all walks of life.For cross-border e-commerce companies,they are faced with two main decision-making problems,one is ”product decision”,that is,how to choose the right product for sale,and the other is ”timing decision”,that is,what products can be sold in different periods.Get better income.Due to the limitation of resource acquisition of small and medium-sized enterprises,they cannot catch the dynamic changes of the market in time,thus missing many business opportunities.In this case,considering that the online community is an important platform for many users to exchange knowledge and obtain information,as well as the interrelationship between various product information and products that users will involve in the topic discussion of major social events,this paper proposes a research framework based on online community data is proposed to help enterprises solve the above problems.Aiming at the enterprise product decision-making problem,this paper introduces natural language processing technology to identify the products that users care about from the review text data.This paper first crawls the posting and comment data of relevant users from the online community,and performs a series of preprocessing tasks such as data cleaning and data representation on the collected raw data.After the expert labeling,use the named entity recognition technology to identify the product of the comment data,and get the product collection,a total of 94 products,indicating that the products that the forum users are concerned about are not focused on one or a few categories,but diversified.These products are also candidate collections for companies to select.Finally,the correlation between product popularity and product customs export data is studied,and the results show that there is a certain correlation between product popularity and customs export data,which shows that the discussion popularity of products can reflect market demand to a certain extent,and also provide a certain guarantee for foreign trade enterprises to select products.Aiming at the timing decision-making problem of enterprises,this paper constructs product topic frequency series based on the frequency change pattern of each product being mentioned in the topic,uses the vector autoregressive model to model the product frequency sequence,and then combines these product frequency sequences with Granger test and impulse response function analysis.The results show that there is a Granger influence relationship between products,that is,there is a certain spillover effect between products,and the vector autoregressive model is used to predict the short-term popularity trend of products,so as to provide suggestions for corporate timing decisions.The research of this paper has the following contributions.First,this paper proposes to mine effective information from the massive comment data in online communities and identify the product collections that users are concerned about train of thought.Second,this paper uses popular products as signals to study the impact relationship between popular products and other product sequences,and provides suggestions for the operation strategies of related products.Third,the product identification and product sequence relationship research framework proposed in this paper is universal.When other markets are impacted,the framework of this paper can be used to conduct research and analysis on the affected product sequences.