Data and Code for paper Outlining and Filling: Hierarchical Query Graph Generation for Answering Complex Questions over Knowledge Graph is available for research purposes.

Last update: Nov 10, 2022

Related tags

Deep Learning HGNet

Overview

Data and Code for paper Outlining and Filling: Hierarchical Query Graph Generation for Answering Complex Questions over Knowledge Graph is available for research purposes.

Results

We apply three KGQA benchmarks to evaluate our approach, ComplexWebQuestions (Talmor and Berant, 2018), LC-QuAD (Trivedi et al., 2017), and WebQSP (Yih et al., 2016).

Dataset	Structure Acc.	Query Graph Acc.	Precision	Recall	F1-score	[email protected]
ComplexWebQuestions	66.96	51.68	65.27	68.44	64.95	65.25
LC-QuAD	78.00	60.90	75.82	75.22	75.10	76.00
WebQSP	79.91	62.63	70.22	74.38	70.61	70.37

Requirements

Python == 3.7.0
cudatoolkit == 10.1.243
cudnn == 7.6.5
six == 1.15.0
torch == 1.4.0
transformers == 4.9.2
numpy == 1.19.2
SPARQLWrapper == 1.8.5
rouge_score == 0.0.4
filelock == 3.0.12
nltk == 3.6.2
absl == 0.0
dataclasses == 0.6
datasets == 1.9.0
jsonlines == 2.0.0
python_Levenshtein == 0.12.2
Virtuoso SPARQL query service

Data

Download and unzip our preprocessed data to ./, you can also running our scripts under ./preprocess to obtain them again.
Download our used Freebase and DBpedia. Both of them only contain English triples by removing other languages. Download and install Virtuoso to conduct the SPARQL query service for the downloaded Freebase and DBpedia. Here is a tutorial on how to install Virtuoso and import the knowledge graph into it.
Download GloVe Embedding glove.42B.300d.txt and put it to your_glove_path.
Download our vocabulary from here. Unzip and put it under ./. It contains our used SPARQL cache for Execution-Guided strategy.

Running Code

1. Training for HGNet

Before training, first set the following hyperparameter in train_cwq.sh, train_lcq.sh, and train_wsp.sh.

--glove_path your_glove_path

Execute the following command for training model on ComplexWebQuestions.

sh train_cwq.sh

Execute the following command for training model on LC-QuAD.

sh train_lcq.sh

Execute the following command for training model on WebQSP.

sh train_wsp.sh

The trained model file is saved under ./runs directory.
The path format of the trained model is ./runs/RUN_ID/checkpoints/best_snapshot_epoch_xx_best_val_acc_xx_model.pt.

2. Testing for HGNet

Before testing, need to train a model first and set the following hyperparameters in eval_cwq.sh, eval_lcq.sh, and eval_wsp.sh.

--cpt your_trained_model_path
--kb_endpoint your_sparql_service_ip

You can also directly download our trained models from here. Unzip and put it under ./.

Execute the following command for testing the model on ComplexWebQuestions.

sh eval_cwq.sh

Execute the following command for testing the model on LC-QuAD.

sh eval_lcq.sh

Execute the following command for testing the model on WebQSP.

sh eval_wsp.sh

Data and Code for paper Outlining and Filling: Hierarchical Query Graph Generation for Answering Complex Questions over Knowledge Graph is available for research purposes.

Related tags

Overview

Results

Requirements

Data

Running Code

1. Training for HGNet

2. Testing for HGNet

Owner

Yongrui Chen

Pre-trained NFNets with 99% of the accuracy of the official paper

Official PyTorch implementation of "Improving Face Recognition with Large AgeGaps by Learning to Distinguish Children" (BMVC 2021)

Wandb-predictions - WANDB Predictions With Python

CIFAR-10 Photo Classification

This is the pytorch implementation for the paper: Learning Accurate Performance Predictors for Ultrafast Automated Model Compression, which is in submission to TPAMI

YOLOv5 detection interface - PyQt5 implementation

Codes for NAACL 2021 Paper "Unsupervised Multi-hop Question Answering by Question Generation"

A Flow-based Generative Network for Speech Synthesis

Code for Parameter Prediction for Unseen Deep Architectures (NeurIPS 2021)

Time-series-deep-learning - Developing Deep learning LSTM, BiLSTM models, and NeuralProphet for multi-step time-series forecasting of stock price.

Text Extraction Formulation + Feedback Loop for state-of-the-art WSD (EMNLP 2021)

Code repository for Self-supervised Structure-sensitive Learning, CVPR'17

[CVPR 2021] Unsupervised Degradation Representation Learning for Blind Super-Resolution

😇A pyTorch implementation of the DeepMoji model: state-of-the-art deep learning model for analyzing sentiment, emotion, sarcasm etc

OpenLT: An open-source project for long-tail classification

Implementation of the state of the art beat-detection, downbeat-detection and tempo-estimation model

Contains code for the paper "Vision Transformers are Robust Learners".

Symbolic Parallel Adaptive Importance Sampling for Probabilistic Program Analysis in JAX

The code release of paper Low-Light Image Enhancement with Normalizing Flow

DUE: End-to-End Document Understanding Benchmark

Data and Code for paper Outlining and Filling: Hierarchical Query Graph Generation for Answering Complex Questions over Knowledge Graph is available for research purposes.

Related tags

Overview

Results

Requirements

Data

Running Code

1. Training for HGNet

2. Testing for HGNet

Owner

Yongrui Chen

Pre-trained NFNets with 99% of the accuracy of the official paper

Official PyTorch implementation of "Improving Face Recognition with Large AgeGaps by Learning to Distinguish Children" (BMVC 2021)

Wandb-predictions - WANDB Predictions With Python

CIFAR-10 Photo Classification

This is the pytorch implementation for the paper: *Learning Accurate Performance Predictors for Ultrafast Automated Model Compression*, which is in submission to TPAMI

YOLOv5 detection interface - PyQt5 implementation

Codes for NAACL 2021 Paper "Unsupervised Multi-hop Question Answering by Question Generation"

A Flow-based Generative Network for Speech Synthesis

Code for Parameter Prediction for Unseen Deep Architectures (NeurIPS 2021)

Time-series-deep-learning - Developing Deep learning LSTM, BiLSTM models, and NeuralProphet for multi-step time-series forecasting of stock price.

Text Extraction Formulation + Feedback Loop for state-of-the-art WSD (EMNLP 2021)

Code repository for Self-supervised Structure-sensitive Learning, CVPR'17

[CVPR 2021] Unsupervised Degradation Representation Learning for Blind Super-Resolution

😇A pyTorch implementation of the DeepMoji model: state-of-the-art deep learning model for analyzing sentiment, emotion, sarcasm etc

OpenLT: An open-source project for long-tail classification

Implementation of the state of the art beat-detection, downbeat-detection and tempo-estimation model

Contains code for the paper "Vision Transformers are Robust Learners".

Symbolic Parallel Adaptive Importance Sampling for Probabilistic Program Analysis in JAX

The code release of paper Low-Light Image Enhancement with Normalizing Flow

DUE: End-to-End Document Understanding Benchmark

This is the pytorch implementation for the paper: Learning Accurate Performance Predictors for Ultrafast Automated Model Compression, which is in submission to TPAMI