Building Corporate Reputation on the Internet: Is the Claimed Sales-Ranking Trustworthy?

Abstract

This paper aims to develop corporate reputation on the Internet by collecting various online information related with a specific company. The online corporate reputation is then used to verify whether the sales-ranking claimed by this company is trustworthy. Firstly, a distributed and generic Web crawler is developed for automatically collecting public opinion data related with a specific company. Meanwhile, basic information of enterprise and open financial data are also collected for auxiliary analysis. Secondly, a filter method is proposed to filter the noisy documents from the amount of public opinion set with word embedding and KL-Divergence methods. Thirdly, a Maximal Marginal Relevance (MMR) based index model is presented for computing the confidence score of the sales-ranking claim. That is, a company with a great deal of positive comments posted by customers is more likely to announce a reliable claim. Unlike traditional media, many negative news about a company cannot be concealed by the user-generated contents, which indicates the objectivity of the corporate reputation on the Internet. The experimental results demonstrate that the corporate reputation built based on public opinion data has an implicit correspondence with the trustworthy of a company.