OpenKE (sub-project of OpenSKL)

OpenKE is a sub-project of OpenSKL, providing an Open-source Knowledge Embedding toolkit for knowledge representation learning (KRL), with TransR and PTransE as key features to handle complex relations and relational paths in large-scale knowledge graphs.

Overview

OpenKE is an efficient implementation based on PyTorch for knowledge embedding. We use C++ to implement some underlying operations such as data preprocessing and negative sampling. For each specific model, it is implemented by PyTorch with Python interfaces so that there is a convenient platform to run models on GPUs. OpenKE contains 4 repositories:

OpenKE-PyTorch: the repository based on PyTorch, which provides the optimized and stable framework for knowledge graph embedding models.

OpenKE-Tensorflow1.0: OpenKE implemented with TensorFlow, also providing the optimized and stable framework for knowledge graph embedding models.

TensorFlow-TransX: light and simple version of OpenKE based on TensorFlow, including TransE, TransH, TransR and TransD.

Fast-TransX: efficient lightweight C++ inferences for TransE and its extended models utilizing the framework of OpenKE, including TransH, TransR, TransD, TranSparse and PTransE.

More information is available on our website http://openke.thunlp.org/

*** UPDATE ***

We are now developing a new version of OpenKE-PyTorch. The project has been completely reconstructed and is faster, more extendable and the codes are easier to read and use now. If you need get to the old version, please refer to branch OpenKE-PyTorch(old).

*** New Features ***

• RotatE
• More enhancing strategies (e.g., adversarial training)
• More scripts of the typical models for the benchmark datasets.
• More extendable interfaces

Models

OpenKE (Tensorflow):

• RESCAL, HolE
• DistMult, ComplEx, Analogy
• TransE, TransH, TransR, TransD

OpenKE (PyTorch):

• RESCAL
• DistMult, ComplEx, Analogy
• TransE, TransH, TransR, TransD
• SimplE
• RotatE

We welcome any issues and requests for model implementation and bug fix.

Experimental Settings

For each test triplet, the head is removed and replaced by each of the entities from the entity set in turn. The scores of those corrupted triplets are first computed by the models and then sorted by the order. Then, we get the rank of the correct entity. This whole procedure is also repeated by removing those tail entities. We report the proportion of those correct entities ranked in the top 10/3/1 (Hits@10, Hits@3, Hits@1). The mean rank (MRR) and mean reciprocal rank (MRR) of the test triplets under this setting are also reported.

Because some corrupted triplets may be in the training set and validation set. In this case, those corrupted triplets may be ranked above the test triplet, but this should not be counted as an error because both triplets are true. Hence, we remove those corrupted triplets appearing in the training, validation or test set, which ensures the corrupted triplets are not in the dataset. We report the proportion of those correct entities ranked in the top 10/3/1 (Hits@10 (filter), Hits@3(filter), Hits@1(filter)) under this setting. The mean rank (MRR (filter)) and mean reciprocal rank (MRR (filter)) of the test triplets under this setting are also reported.

More details of the above-mentioned settings can be found from the papers TransE, ComplEx.

For those large-scale entity sets, to corrupt all entities with the whole entity set is time-costing. Hence, we also provide the experimental setting named "type constraint" to corrupt entities with some limited entity sets determining by their relations.

Experiments

We have provided the hyper-parameters of some models to achieve the state-of-the-art performace (Hits@10 (filter)) on FB15K237 and WN18RR. These scripts can be founded in the folder "./examples/". Up to now, these models include TransE, TransH, TransR, TransD, DistMult, ComplEx. The results of these models are as follows,

Model	WN18RR	FB15K237	WN18RR (Paper*)	FB15K237 (Paper*)
TransE	0.512	0.476	0.501	0.486
TransH	0.507	0.490	-	-
TransR	0.519	0.511	-	-
TransD	0.508	0.487	-	-
DistMult	0.479	0.419	0.49	0.419
ComplEx	0.485	0.426	0.51	0.428
ConvE	0.506	0.485	0.52	0.501
RotatE	0.549	0.479	-	0.480
RotatE (+adv)	0.565	0.522	0.571	0.533

We are still trying more hyper-parameters and more training strategies (e.g., adversarial training and label smoothing regularization) for these models. Hence, this table is still in change. We welcome everyone to help us update this table and hyper-parameters.

Installation

1. Install PyTorch

2. Clone the OpenKE-PyTorch branch:

git clone -b OpenKE-PyTorch https://github.com/thunlp/OpenKE --depth 1
cd OpenKE
cd openke

3. Compile C++ files

bash make.sh

4. Quick Start

cd ../
cp examples/train_transe_FB15K237.py ./
python train_transe_FB15K237.py

Data

• For training, datasets contain three files:

  train2id.txt: training file, the first line is the number of triples for training. Then the following lines are all in the format (e1, e2, rel) which indicates there is a relation rel between e1 and e2 . Note that train2id.txt contains ids from entitiy2id.txt and relation2id.txt instead of the names of the entities and relations. If you use your own datasets, please check the format of your training file. Files in the wrong format may cause segmentation fault.

  entity2id.txt: all entities and corresponding ids, one per line. The first line is the number of entities.

  relation2id.txt: all relations and corresponding ids, one per line. The first line is the number of relations.

• For testing, datasets contain additional two files (totally five files):

  test2id.txt: testing file, the first line is the number of triples for testing. Then the following lines are all in the format (e1, e2, rel) .

  valid2id.txt: validating file, the first line is the number of triples for validating. Then the following lines are all in the format (e1, e2, rel) .

  type_constrain.txt: type constraining file, the first line is the number of relations. Then the following lines are type constraints for each relation. For example, the relation with id 1200 has 4 types of head entities, which are 3123, 1034, 58 and 5733. The relation with id 1200 has 4 types of tail entities, which are 12123, 4388, 11087 and 11088. You can get this file through n-n.py in folder benchmarks/FB15K

To do

The document of the new version of OpenKE-PyTorch will come soon.

If you use the code, please cite the following paper:

 @inproceedings{han2018openke,
   title={OpenKE: An Open Toolkit for Knowledge Embedding},
   author={Han, Xu and Cao, Shulin and Lv Xin and Lin, Yankai and Liu, Zhiyuan and Sun, Maosong and Li, Juanzi},
   booktitle={Proceedings of EMNLP},
   year={2018}
 }

This package is mainly contributed (in chronological order) by Xu Han, Yankai Lin, Ruobing Xie, Zhiyuan Liu, Xin Lv, Shulin Cao, Weize Chen, Jingqin Yang.

---

About THU_OpenSKL

THU-OpenSKL project aims to harness the power of both structured knowledge and unstructured languages via representation learning. All sub-projects of THU-OpenSKL are as follows.

• Algorithm:
  • OpenNE
    • An effective and efficient toolkit for representing nodes in large-scale graphs as embeddings, with TADW as key features to incorporate text attributes of nodes.
  • OpenKE
    • An effective and efficient toolkit for representing structured knowledge in large-scale knowledge graphs as embeddings, with TransR and PTransE as key features to handle complex relations and relational paths.
    • This toolkit also includes three sub-toolkits:
      • KB2E
      • TensorFlow-Transx
      • Fast-TransX
  • OpenNRE
    • An effective and efficient toolkit for implementing neural networks for extracting structured knowledge from text, with ATT as key features to consider relation-associated text information.
    • This toolkit also includes two sub-toolkits:
      • JointNRE
      • NRE
  • ERNIE
    • An effective and efficient toolkit for augmenting pre-trained language models with knowledge graph representations.
• Resource:
  • The embeddings of large-scale knowledge graphs pre-trained by OpenKE, covering three typical large-scale knowledge graphs: Wikidata, Freebase, and XLORE. The embeddings are free to use under the MIT license, and please click the following link to submit download requests.
  • OpenKE-Wikidata
    • Wikidata is a free and collaborative database, collecting structured data to provide support for Wikipedia. Wikidata contains 20,982,733 entities, 594 relations and 68,904,773 triplets.
    • TransE version: Knowledge graph embeddings of Wikidata pre-trained by OpenKE.
    • Plugin version of TransR: Knowledge graph embeddings of Wikidata pre-trained by OpenKE for the project Knowledge-Plugin.
  • OpenKE-Freebase
    • Freebase was a large collaborative knowledge base consisting of data composed mainly by its community members. It was an online collection of structured data harvested from many sources. Freebase contains 86,054,151 entities, 14,824 relations and 338,586,276 triplets.
    • TransE version: Knowledge graph embeddings of Freebase pre-trained by OpenKE.
  • OpenKE-XLORE
    • XLORE is one of the most popular Chinese knowledge graphs developed by THUKEG. XLORE contains 10,572,209 entities, 138,581 relations and 35,954,249 triplets.
    • TransE version: Knowledge graph embeddings of XLORE pre-trained by OpenKE.
• Application:
  • Knowledge-Plugin
    • An effective and efficient toolkit of plug-and-play knowledge injection for pre-trained language models. Knowledge-Plugin is general for all kinds of knowledge graph embeddings mentioned above. In the toolkit repository, we provide the example of plugging OpenKE-Wikidata embeddings into BERT.