Augmenting Customer Support with an NLP-based Receptionist

TL;DR

This paper presents a Portuguese BERT-based chatbot integrated with structured data to predict client contact motivation, demonstrating human-level performance and business benefits over traditional NLP methods.

Contribution

It introduces a novel combination of Portuguese BERT with structured data for a finite state machine chatbot in real estate, achieving high accuracy and practical advantages.

Findings

Achieves human-level results on unbalanced dataset

Outperforms classical NLP methods in business impact

Demonstrates effective integration of BERT with structured data

Abstract

In this paper, we show how a Portuguese BERT model can be combined with structured data in order to deploy a chatbot based on a finite state machine to create a conversational AI system that helps a real-estate company to predict its client's contact motivation. The model achieves human level results in a dataset that contains 235 unbalanced labels. Then, we also show its benefits considering the business impact comparing it against classical NLP methods.