Customer Chatbot
1.0.0
Chinese intelligent customer chatbot Demo, including the gossip and the professional Q&A(FAQ), support for custom components!
1. This project consists of two parts, one is the recall model based on tf-idf retrieval , and the other is the fine ranking model based on CNN . This project integrates the two to build a recall + ranking customer service chat robot. The system supports chat mode and FAQ question and answer mode , and the data used are the Xiaohuangji chat data set and the FAQ question and answer data set in vertical fields. This version is the first version, and other performance such as speed needs to be improved. These tasks will be uploaded in the future. According to the current feedback, the difficulty of the system is to build a high-precision and short-time-consuming rerank model. If it is to be used in industry, it needs to be significantly modified. If you want to be familiar with the entire process of the problem system, this project can 100% meet the needs. .
2. Only the system in the recall stage can view:
Question and answer robot based on tf-idf
This project is improved on the basis of the First version, and some rules have been added. The current advantages of this system are:
1. The two modules of recall + sorting do not interfere with each other, which is convenient for custom modification and maintenance. 2. The system adopts optimization of sorting rules, which improves the retrieval speed. 3. A simple inverted index is added to optimize the retrieval process.
This project relies on the route function to convert question and answer tasks, which are divided into chat mode and faq mode. The main purpose of this is that the system can set up different situational dialogues according to different tasks. At the same time, the system manages the two corpora separately to avoid search time. increase. The current effect is that if you do not enter end to terminate the conversation, you can freely convert between chat mode and faq mode during the conversation, as you like!
Cite
If you use xiaotian-chatbot1.0 in your research, please cite it in the following format:
@software{xiaotian-chatbot1.0,
author = {ZhengWen Xie},
title = {xiaotian-chatbot1.0: A Customer-Chatbot System},
year = {2019},
url = {https://github.com/WenRichard/Customer-Chatbot},
}
