ارائه مدلي براي طبقه بندي مشتريان بر اساس ادراك آنها از شخصيت برند در شبكه‌هاي اجتماعي

In the competitive environment of businesses where products and services are easily copied, branding is recognized as a differentiating element based on which companies can show themselves differently and create more stable relationships with customers. Brand personality is a part of brand identity that plays an important role in this field and if the brand personality is well established, it also affects consumer preference and support and creates stronger emotional relationships with them. Considering the importance of brand personality, it is necessary to eva‎luate customersʹ perception of brand personality so that a brand can recognize its weaknesses and strengths in transferring brand personality and improve it. Social networks have created a good opportunity for brands to be able to indirectly analyze the perception of customers based on their comments and also have a closer interaction with their customers. What is presented in this research is to create a specific structure to create a dataset and build artificial intelligence models to classify customers based on their perception of brand personality. Customer classes were created based on Akerʹs personality model and considering alignment, non-alignment and neutrality of their opinions, which includes 9 classes. The classification shows the degree of homogeneity of the customersʹ perception of the brandʹs personality, and on the other hand, in each class, by performing the correspondence analysis technique, it is possible to identify the brandʹs strengths and weaknesses in presenting the desired personality. The Twitter platform has been used to collect data, these data are prepared for training deep learning models after pre-processing and using vectorization techniques. Two flat and hierarchical approaches have been considered for classification, and after eva‎luation, the models based on the architecture of converters showed better performance in the hierarchical approach. At the end, the proposed model was applied on the Snowa dataset including comments related to this brand on the Twitter social network. The results showed that Snowaʹs brand personality is close to the dimensions of honesty and competence, and the features related to the product played an important role in the customersʹ perception of the brandʹs personality, and on the other hand, effective advertising was not done by this brand to influence the customers.

6