Project Details


Advanced Collection Systems (ACS) specialises in debt recovery and cash-flow services. Traditional telephony processes are still very much the primary method of engagement for the majority of debt collections operations. Strategically, ACS cannot scale up their business to a £10m turnover operation through manual collection alone. Moreover, ACS will need to evolve to make effective use of constantly increasing amounts of real time data. An intelligent Adaptive Virtual Agent (AVA) software system, will help increase throughput and create new revenue streams in supplying their wider sector.

The vision of ACS is to intruding a new digital service which will lead in the developing Digital Market. To help ACS to achieve that, this project aims to create an adaptive, self-learning, negotiation and decision tool by applying Machine Learning (ML) and Data Mining techniques to a range of complex human/computer interaction patterns. Automating the negotiation process for straightforward transactions will allow the Agents to focus on more complex negotiations using the new adaptive tools to assist the resolution decision making. In addition, analysing human negotiation paths and replicating in computer programs that will continually improve as new data is added.

There is no known capability in the sector for a Machine Learning approach to automated negotiation and resolution. The new adaptive process for managing resolution pathways will be unique in the industry and the solution will require an evolution journey by accessing a combination of research in Data Mining, Machine Learning and the psychology of cognitive processes. The output, a new expert system, will be capable of absorbing, manipulating and analysing very large and complex data sets (Big Data), providing analysis and resolution scenarios. The new capability will enable ACS to take on more business from their existing client base which is currently restricted by labour and office facilities.
Effective start/end date1/03/1729/02/20


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.