SEMANTIC APPROACH TO AGENT ROUTING USING A HYBRID ATTRIBUTE-BASED RECOMMENDER SYSTEM

Anwitha Paruchuri

Keywords: Hybrid Recommendation, Customer Relationship Management, Semantic Embedding, Approximate Nearest Neighbor

Abstract: Traditionally contact centers route an issue to an agent based on ticket load or skill of the agent. When a ticket comes into the system, it is either manually analyzed and pushed to an agent or automatically routed to an agent based on some business rules. A Customer Relationship Management (CRM) system often has predefined categories that an issue could belong to. The agents are generally proficient in handling multiple categories, the categories in the CRM system are often related to each other, and a ticket typically contains content across multiple categories. This makes the traditional approach sub-optimal. We propose a Hybrid Recommendation based approach that recommends top N agents for a ticket by jointly modelling on the interactions between the agents and categories as well as on the semantic features of the categories and the agents.

One-sentence Summary: A Hybrid Recommendation based approach that recommends top N agents for a ticket by jointly modelling on the interactions between the agents and categories as well as on the semantic features of the categories and the agents

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Reviewed Version (pdf): https://openreview.net/references/pdf?id=ldLpeg6L9B0