Hape: optimizing customer relation by automatic task distribution using constrained optimization and natural language processing

Since 37 million customers use Enedis services daily, the company must manage massive requests. To affect these requests as fast as possible to the agent, the company needs several advanced planning models. Therefore, the problem is to distribute tasks to the agents automatically while respecting the time available to them to carry out the tasks and their competence. To this end, we present the different models embedded in HAPe, an application that allows us to address the problem. This paper proposes a constrained linear optimization model for automatic distribution, a text classification model to help predict agent skills, and an automatic entity recognition model that extracts essential domain-specific information. These two NLP models are based on our language model, which we implemented. The results obtained from our different studies are promising and allow us to increase the agents' efficiency. This study is the first study on this subject for DSO activities.