Chatbot - QA Component - Master's Thesis at Master of Engineering (MSE), Switzerland
This repository is used as a component of the Master's Thesis of Romain Claret.
This component is a question answering chatbot.
- Python 2 or 3
- Install the python requirements
conda/pip install spacy requests pybind11 hdt networkx- Install the spacy model:
python -m spacy download en_vectors_web_lg
- Get the submodules
git submodule init
git submodule update- Download the data
bash initialize.sh- If something goes wrong with the hdt library, check out this link https://pypi.org/project/hdt/, usually the fix is to install the Python Development headers.
- CONVEX fork is used for the subgraph qa system.
- Paper: Look before you Hop: Conversational Question Answering over Knowledge Graphs Using Judicious Context Expansion
- Authors: Philipp Christmann, Rishiraj Saha Roy, Abdalghani Abujabal, Jyotsna Singh, and Gerhard Weikum, CIKM 2019.
- Original github repo: https://github.com/PhilippChr/CONVEX
- License: MIT license by Philipp Christmann
- Website: http://qa.mpi-inf.mpg.de/convex/
- PDF preprint of the CIKM'19 paper: https://arxiv.org/abs/1910.03262
This project by Romain Claret is licensed under the MIT license. Note that used modules (see above) may be licensed differently; however, the license capabilities are maintained.