Method for constructing question comment pairs based on word vector representation

Abstract

The invention discloses a method for constructing question comment pairs based on word vector representation. The method comprises the following steps: obtaining a question and answer corpus and a comment corpus, and splicing questions and answers of the questions; representing the spliced question and answer statements and comments by using a word2vec tool and a bag-of-words model based on word vectors to obtain sentence vectors of the spliced question and answer statements and sentence vectors of comment statements; calculating the similarity between the spliced question and answer sentencesand comment sentences by utilizing the obtained sentence vectors; for each question, sorting the comments according to the obtained similarity, and taking the n comments with the highest similarity as candidate comments of the question; and marking the candidate comments of each question and marking comments capable of answering the questions and comments incapable of answering the questions by the expert, so that question and comment pairs can be obtained. According to the method, the information search cost of consumers during online shopping can be effectively reduced.

Description

technical field [0001] The invention relates to the technical field of big data mining, in particular to a method for constructing question comment pairs based on word vector representation. Background technique [0002] At present, e-commerce is developing rapidly, and more and more people choose to shop online. In the process of consumer purchase decision-making, in order to judge whether the product meets expectations, it is necessary to collect product-related information. In the process of traditional offline shopping, consumers will try some products to judge whether the products meet expectations. However, when shopping online, consumers cannot obtain first-hand product information because they cannot touch the actual product. Generally, the information provided by merchants is attribute information such as product parameters and models, and rarely has some experience-related information. At the same time, in order to sell more products, very few merchants provide ...

AI Technical Summary

Problems solved by technology

Although questions to merchants can sometimes be answered quickly, merchants hardly provide negative information about their products

If consumers give up obtaining further information due to the high cost of information acquisition, and purchase products without obtaining sufficient information, it is likely that the products do not meet expectations.

[0005] From the perspective of consumers, the large difference between the expected attributes of the product and the real attributes will bring a poor shopping experience and waste consumers' time and money

From the merchant’s point of view, consumers will consume a lot of manpower and material resources when applying for after-sales services. At the same time, negative comments will reduce the reputation of the merchant among consumers, which is not conducive to long-term operation.

From the perspective of the platform, mediating disputes between consumers and merchants will also consume relatively large costs

Images