mingtop/drfaq

Demo for FAQ

Dr.Covid Telegrame Demo

• DrFAQ is a FAQ demo for Dr COVID (2020), full version will be released.
• Designed and implemented a NLP Question Answering architecture using spaCy, BERT language model, ElasticSearch, and Telegram Bot API.
• DOCTOR COVID applied to SingHealth at Singapore EXPO and D’Resort with more information from News of Dr.COVID.

Requirement

pip install python-telegram-bot
pip install 

• Prepare a FAQ csv corpus, we have some demo in /match

Demo

Telegrame Demo Screenshot

Spacy/BERT Comparison Results

Pipeline

When a question is asked, the following processes are performed:

1. FAQ Question Matching using spaCy's Similarity - /match
2. NLP Question Answering using huggingface's BERT - /nlp

To do list

1. Update BERT model and pipeline 2
2. Transformer-based Close domain search
3. Support WhatApp & Wechat
4. Answer Search using ElasticSearch

The BERT-based Pipeline

References

• explosion/spaCy - Industrial-strength Natural Language Processing (NLP) with Python and Cython
• huggingface/transformers - Transformers: State-of-the-art Natural Language Processing for TensorFlow 2.0 and Pytorch
• elastic/elasticsearch-py - Official Python low-level client for Elasticsearch
• python-telegram-bot/python-telegram-bot - Python Wrapper for Telegram Bots
• google-research/bert - TensorFlow code and pre-trained models for BERT
• BERT - Pre-training of Deep Bidirectional Transformers for Language Understanding
• DrQA Telegrame demo for FAQ