ReAdvisor: Intelligent Customer Feedback Management
Online recommendation portals such as TripAdvisor or Google Maps collect and provide user-generated content like reviews provided by tourists, local guides, as well as hotel or restaurant guests. Given the growing impact of those portals on customers’ and guests’ decisions, businesses increasingly depend on effective management of the customer feedback: an adequate reaction to complaints and suggestions forms the core of the online customer care. However, many small business lack resources and skills concerning customer feedback management and prefer to focus on their core activity: hospitality, experience, and craftsmanship. Consequently, they delegate the management of their online presence in the recommendation portals.
Despite the growing demand, companies offering customer feedback management services face specific challenges as well. It is increasingly harder to find talented authors who can generate the responses. Simultaneously their clients and business partners expect highly individualised, well-crafted responses composed in line with the experience they offer to their customers or guests. As a consequence, customer feedback management service providers experience increasing workload.
Together with the Dept. of Computational Linguistics, re:spondelligent and welante, IMRG explores the potentials of artificial intelligence (AI) to streamline and support the processes of composing adequate responses to reviews in online recommendation portals. Identification of an adequate response author, understanding the content and the implications of a particular review, composing a high-quality response, as well as quality and sanity checks — all those activities can be augmented with AI. For instance, response authors can benefit from automatically generated suggestions and correction recommendations, ultimately leading to a hybrid response generation involving collaboration between humans and AI. IMRG investigates ways to enable, structure, and coordinate effective collaboration between human and non-human agents in the process of response authoring. The project leads to the identification of new work processes, description of new roles which emerge in effective human-AI collaboration, as well as portrayal of the archetypes or metaphors governing the collaboration in the given context.