Detecting Egregious Conversations between Customers and Virtual Agents
This addresses the issue of poor customer service interactions for companies using virtual agents, though it is incremental as it builds on existing detection methods.
The paper tackled the problem of detecting egregious conversations between customers and virtual agents by using behavioral cues, agent response patterns, and interaction features, resulting in a 20% improvement in F1-score over textual features alone.
Virtual agents are becoming a prominent channel of interaction in customer service. Not all customer interactions are smooth, however, and some can become almost comically bad. In such instances, a human agent might need to step in and salvage the conversation. Detecting bad conversations is important since disappointing customer service may threaten customer loyalty and impact revenue. In this paper, we outline an approach to detecting such egregious conversations, using behavioral cues from the user, patterns in agent responses, and user-agent interaction. Using logs of two commercial systems, we show that using these features improves the detection F1-score by around 20% over using textual features alone. In addition, we show that those features are common across two quite different domains and, arguably, universal.