Abstract: In the era of Web2.0, social media is not only the access for users to get information, but also the platform for them to post and build a relationship. User participation, user-led and user creation have become a powerful tool for social media to attract users, enhance competitiveness and improve enterprise value, generating a large number of User Generated Content (referred to as "UGC") of great business value. As a unique form of UGC, Online Negative Word-of-Mouth (referred to as "ONWOM") is a kind of negative view or comment which the consumers express on the Internet. If the company does not respond to it in time, it will lead to a large area of dissatisfaction, resulting in damage to corporate reputation, brand value and other serious consequences. However, due to the large number, the fast speed of transmission and other characteristics of ONWOM, the company will face the dilemma of high costs, slow responses and resource shortages if just relies on the traditional manual processing method. In handling negative word-of-mouth, excessive participation of the company may made itself be suspected, resulting in the loss of credibility. Therefore, taking a small number of experts with more knowledge in UGC as a resource for ONWOM processing, a new way to deal with ONWOM for companies can be provided.
Most studies in the field of expert recognition only focus on the knowledge capability of experts. However, it cannot meet the value demands of each subject in the processing of ONWOM. From the perspective of resources, a resource mapping framework of expert recognition is established, and the explicit and implicit resource mapping process based on it is analyzed. In addition to considering the ability which directly reflects the professional level of the experts, the emotional comfort ability when experts deal with ONWOM is comsidered, and emotional infection mechanism to obtain the quantitative indicators of emotional ability is used. Moreover, as a prerequisite for resource integration and emotional infection, the interaction degree is also introduced into the user's capacity characteristic space to construct the artificial neural network model for the expert recognition of ONWOM processing. By comparing the experimental results, it is found that the model proposed in this paper can significantly improve the recognition performance in the expert recognition of ONWOM processing. In addition, a new method for expert recognition is provided by the resource mapping framework proposed in this paper, and a new idea for understanding the ability structure of the expert of ONWOM processing is provided.

Key words: online negative word-of-mouth, expert identification, artificial neural network