Sentiment Analysis and Anomally Detection for Call Centers
Can we identify the causes of staff inefficiencies?
Many call centers use rule-based systems to answer customer questions or carry on campaings. Though data is logged, it might be hard to extract meaningfull patterns that can identify problematic pathways or divergence from normal behavior.
Raw data in this case is the interaction between two people and conventional methodologies may not yield significant results. Different sentimental and behavioral approached should be provided.
Both in-campaign and cross-campaign our algorithms were able to detect causes of staff inefficiencies and divergence from normal behavior. We can identify patterns and suggest solutions to increase overall performance. Our algorithms were also able to detect the problematic paths of rule-based system helping to prune the call-center trees.