A framework for implementing customer-service chatbots: The case for LUZ shopping centre in Jerez, Spain

Sammanfattning

With the advancement of artificial intelligence, customers and users in general have been more used to having a virtual assistance method for service-related tasks. However, in the shopping centre industry, the self-help solutions mostly follow the traditional pattern of having a conciérge or a totem where you can look for the location of a particular store. This thesis aims to look at this with a different approach, by proposing a chatbot that can be integrated in the shopping centres and prepared to recommend stores based on a product search. To create the required database for product search, web-scraping techniques were applied, combined with natural language processing to make the inputs machine-readable. The chatbot was created based on a set of rules and conversational scenarios common to a self-help desk solution. The evaluation of the proposed method was based on expert interviews for feedback. The main qualities of the model are its straightforward use, and clear communication style. Regarding the limitations of the model, the most commonly mentioned were the robotic response style, and the absence of the capabilities of providing a store map after the recommendation. Potential improvements that could be adopted are a rank of the suggested stores, based on experts’ inputs, and link the outputs of the conversation with discounts and promotions. However, even with the limited current setup, all experts saw value in the implementation of the model, especially in the shopping centre website. Expected positive effects include more visitation to the facility and more visibility for smaller stores.