Lingke: A Fine-grained Multi-turn Chatbot for Customer Service

TL;DR

Lingke is a multi-turn customer service chatbot that leverages information retrieval and fine-grained processing to improve response quality based on product documents.

Contribution

The paper introduces Lingke, a novel chatbot architecture that combines information retrieval with fine-grained and context-aware response generation for multi-turn interactions.

Findings

Effective handling of multi-turn conversations.

Improved response accuracy based on product documents.

Robustness in customer service scenarios.

Abstract

Traditional chatbots usually need a mass of human dialogue data, especially when using supervised machine learning method. Though they can easily deal with single-turn question answering, for multi-turn the performance is usually unsatisfactory. In this paper, we present Lingke, an information retrieval augmented chatbot which is able to answer questions based on given product introduction document and deal with multi-turn conversations. We will introduce a fine-grained pipeline processing to distill responses based on unstructured documents, and attentive sequential context-response matching for multi-turn conversations.