Enhanced customer support through effective identification of PMRs at high risk of escalations

点击翻译按钮获取中文摘要

In software engineering, customers' input is critical to gathering, analyzing and prioritizing product requirements, as well as in ensuring continuous customers' satisfaction through recording problem reports after deployment. These reports might come in the form of suggestions for product enhancements (useful for features in the next release) or software defect reports that need more urgent attention from the software vendor. For large-scale software products effectively responding to customer problem reporting is, however, a complex process, as it involves assessing the reported problem, considering tradeoffs and making non-trivial judgments about possible solutions and resources required to tackle the problem. The significant problem faced by software vendors is that, without effective resolution, some of these reports 'escalate' into critical situations, seriously impacting the customers' businesses as well the software vendor's loss of reputation and satisfaction. This project addresses this very specific problem at our industrial partner IBM: customer reported problems not always receiving adequate attention, leading to escalations that hurt customer relationships. Once a customer reports a problem, the customer support, management and developers lack the appropriate techniques and tools to analyze the information needed to assess and resolve it, resulting in significant delays in finding the relevant information and addressing the problem for effective response to the customer. In collaboration with the IBM Victoria lab, this project develops techniques and tools that implement machine learning and IBM Watson's cognitive capabilities to provide high levels of service to any support issues raised. More specifically, it provides for more effective identification of PMRs (customer reported problems) that have a high risk of escalating into critical customer situations, as well as managing the information and communication related to PMRs. It provides them with the ability to forecast and plan resources more optimally, and to focus at the right time to 'de-escalate' situations. Our research results will reduce the time and cost spent on escalations, and help maintain productive relationship with the IBM customers