zychen423 / uhop

UHop: An Unrestricted-Hop Relation Extraction Framework for Knowledge-Based Question Answering

UHop: An Unrestricted-Hop Relation Extraction Framework for Knowledge-Based Question Answering

This is the code for our UHop paper.

Getting Started

To train and evaluate this code:

git clone https://github.com/zychen423/UHop.git
cd UHop
wget http://doraemon.iis.sinica.edu.tw/UHop/UHop_data.zip
unzip UHop_data.zip
mkdir saved_model
cd ./script
CUDA_VISIBLE_DEVICES=0 python3.6 train.py

Then you can use scripts in script/allocate to get best result from parameter search

Prerequisites

The environment we use is listed:

python >= 3.6 pytorch 1.3

License

This Work is licensed under the GNU General Public License v3.0 without any warranties. The license text in full can be getting access at the file named COPYING-GPL-3.0. Any person obtaining a copy of this Work and associated documentation files is granted the rights to use, copy, modify, merge, publish, and distribute the Work for any purpose. However if any work is based upon this Work and hence constitutes a Derivative Work, the GPL-3.0 license requires distributions of the Work and the Derivative Work to remain under the same license or a similar license with the Source Code provision obligation.

uhop's Issues

Can you explain your data in WQ-train_data.txt

[0, "what is the name of TOPIC_ENTITY brother", [[["music.artist.origin", [], 0], ["broadcast.artist.content", [], 0], ["celebrities.celebrity.net_worth", [], 0], ["people.person.ethnicity", [], 0], ["celebrities.celebrity.legal_entanglements", [], 0], ["music.artist.concert_tours", [], 0], ["base.popstra.celebrity.infidelity_victim", [], 0], ["music.group_member.membership", [], 0], ["people.person.education", [], 0], ["people.person.places_lived", [], 0], ["base.myspace.myspace_user.username", [], 0], ["influence.influence_node.influenced", [], 0], ["music.featured_artist.albums", [], 0], ["tv.tv_program_guest.appeared_on", [], 0], ["people.person.profession", [], 0], ["music.artist.contribution", [], 0], ["base.schemastaging.context_name.nickname", [], .........

Can you explain your data in WQ-train_data.txt? it's all relation? 0 refer what? Can you explain this data in detail?

Sorry, where is your main.py?

Sorry, where is your main.py? You just list your train.py

How to predict

Perhaps my carelessness, I only find the process of train and evaluate. So if I want to predict the KB path for a test QUESTION, How to use those code to predict?

Could you open source your data processing code?

Hello author, could you open source your data processing code? Or could you provide ideas for data processing? I want to apply it to other datasets, but for the data processing part, I don’t know how to implement it. Thank you very much.