Hila
Weisman-Zohar

Poster – Agent Behavioral Analytics using CNN & Linguistics

NICE

Hila Weisman

Hila
Weisman-Zohar

Poster – Agent Behavioral Analytics using CNN & Linguistics

NICE

Hila Weisman

Bio

Hi, my name is Hila and I love thinking of and implementing algorithms that deal with natural language. I finished my masters in NLP 6 years ago and have been working at NICE ever since.

 

At NICE I build systems that discover insights from call center conversations, for example discovering prominent topics, analyzing agent’s conversational behavior and much more. I’m married to Yotam and a mother to Noga and Tamar 🙂

Bio

Hi, my name is Hila and I love thinking of and implementing algorithms that deal with natural language. I finished my masters in NLP 6 years ago and have been working at NICE ever since.

 

At NICE I build systems that discover insights from call center conversations, for example discovering prominent topics, analyzing agent’s conversational behavior and much more. I’m married to Yotam and a mother to Noga and Tamar 🙂

Abstract

In this work, I’ll show how we’re creating a system able to predict call center agent’s behavioral characteristics (e.g., knowledgeable, friendly etc.) using a Deep Learning based solution which also utilizes novel linguistic-based features.

 

These features try to model key aspects in the agent’s conversations, for example: the agent’s manner of speaking (e.g., fluent/dis-fluent), complexity of language (long sentences, unique words), level of adaptability (to change in customer’s sentiment) etc. We will show preliminary results using different settings and features and discuss future directions for research.

Abstract

In this work, I’ll show how we’re creating a system able to predict call center agent’s behavioral characteristics (e.g., knowledgeable, friendly etc.) using a Deep Learning based solution which also utilizes novel linguistic-based features.

 

These features try to model key aspects in the agent’s conversations, for example: the agent’s manner of speaking (e.g., fluent/dis-fluent), complexity of language (long sentences, unique words), level of adaptability (to change in customer’s sentiment) etc. We will show preliminary results using different settings and features and discuss future directions for research.