Tensorflow Implementation of "Attention-Based Bidirectional Long Short-Term Memory Networks for Relation Classification" (ACL 2016)

Home Page: http://www.aclweb.org/anthology/P16-2034

License: Apache License 2.0

Python 56.96% Perl 43.04%

attention-based-bilstm-relation-extraction's Introduction

Attention-Based Bidirectional Long Short-Term Memory Networks for Relation Classification

Tensorflow Implementation of Deep Learning Approach for Relation Extraction Challenge(SemEval-2010 Task #8: Multi-Way Classification of Semantic Relations Between Pairs of Nominals) via Attention-based BiLSTM.

Usage

Train

Train data is located in "SemEval2010_task8_all_data/SemEval2010_task8_training/TRAIN_FILE.TXT".
"glove.6B.100d" is used as pre-trained glove model.
Performance (accuracy and f1-socre) outputs during training are NOT OFFICIAL SCORE of SemEval 2010 Task 8. To compute the official performance, you should proceed the follow Evaluation step with checkpoints obtained by training.

Display help message:

$ python train.py --help

Train Example:

$ python train.py --embedding_path "glove.6B.100d.txt"

Evaluation

You can get an OFFICIAL SCORE of SemEval 2010 Task 8 for test data by following this step. README describes how to evaluate the official score.
Test data is located in "SemEval2010_task8_all_data/SemEval2010_task8_testing_keys/TEST_FILE_FULL.TXT".
MUST GIVE --checkpoint_dir ARGUMENT, checkpoint directory from training run, like below example.

Evaluation Example:

$ python eval.py --checkpoint_dir "runs/1523902663/checkpoints/"

SemEval-2010 Task #8

Given: a pair of nominals
Goal: recognize the semantic relation between these nominals.
Example:
- "There were apples, pears and oranges in the bowl."
  → CONTENT-CONTAINER(pears, bowl)
- “The cup contained tea from dried ginseng.”
  → ENTITY-ORIGIN(tea, ginseng)

The Inventory of Semantic Relations

Cause-Effect(CE): An event or object leads to an effect(those cancers were caused by radiation exposures)
Instrument-Agency(IA): An agent uses an instrument(phone operator)
Product-Producer(PP): A producer causes a product to exist (a factory manufactures suits)
Content-Container(CC): An object is physically stored in a delineated area of space (a bottle full of honey was weighed) Hendrickx, Kim, Kozareva, Nakov, O S´ eaghdha, Pad ´ o,´ Pennacchiotti, Romano, Szpakowicz Task Overview Data Creation Competition Results and Discussion The Inventory of Semantic Relations (III)
Entity-Origin(EO): An entity is coming or is derived from an origin, e.g., position or material (letters from foreign countries)
Entity-Destination(ED): An entity is moving towards a destination (the boy went to bed)
Component-Whole(CW): An object is a component of a larger whole (my apartment has a large kitchen)
Member-Collection(MC): A member forms a nonfunctional part of a collection (there are many trees in the forest)
Message-Topic(CT): An act of communication, written or spoken, is about a topic (the lecture was about semantics)
OTHER: If none of the above nine relations appears to be suitable.

Distribution for Dataset

SemEval-2010 Task #8 Dataset [Download]

Relation	Train Data	Test Data	Total Data
Cause-Effect	1,003 (12.54%)	328 (12.07%)	1331 (12.42%)
Instrument-Agency	504 (6.30%)	156 (5.74%)	660 (6.16%)
Product-Producer	717 (8.96%)	231 (8.50%)	948 (8.85%)
Content-Container	540 (6.75%)	192 (7.07%)	732 (6.83%)
Entity-Origin	716 (8.95%)	258 (9.50%)	974 (9.09%)
Entity-Destination	845 (10.56%)	292 (10.75%)	1137 (10.61%)
Component-Whole	941 (11.76%)	312 (11.48%)	1253 (11.69%)
Member-Collection	690 (8.63%)	233 (8.58%)	923 (8.61%)
Message-Topic	634 (7.92%)	261 (9.61%)	895 (8.35%)
Other	1,410 (17.63%)	454 (16.71%)	1864 (17.39%)
Total	8,000 (100.00%)	2,717 (100.00%)	10,717 (100.00%)

Reference

Attention-Based Bidirectional Long Short-Term Memory Networks for Relation Classification (ACL 2016), P Zhou et al. [paper]
roomylee's cnn-relation-extraction repository [github]

attention-based-bilstm-relation-extraction's People

Contributors

Stargazers

Watchers

Forkers

13lingzi itsmengzaime wellbetter monk1337 fendaq lcy081099 leowood topgunforone huaxinyuan charlesfufu vhientran wangmingxjtu syunzhou joanzhou zhoudayang shenliyan01 10183308 samithaj allensmile shubhampachori12110095 alchemist1024 machine4life yangchangli lyjsz johmy592 moolighty alucardmini circirmaa ghadaalfattni yaozhian ojasvin wtfenfenmiao littleflow3r davidliremini xuhaiming1996 lichao88 liuriver123 qiyyyue carolingao fantasydreams sonali856 qniguoym xuehui0725 xiaojie2018 cjm1044642385 chenny0808 bigboyooo wilsonsky18 xuanhanyu coffeebeanustb xiaoyu5301 yyy0921 plr123 jerryten jingkl pnorest dorothychai crystal22 ssossp guoyin90 chongtwo xingboliu hxlszxy weibobo2015 sxrczh tjuzcguo dongpoli youngflyasd taufique74 panna19951227 donjon86 zihaoyue pythonjwm adixxov hiyky yaolezju chen1310054465 anki54 baozi-lala ling718 zbn123 90217 huliflove newtusst wuzaijun914 husin123 hsimwong sparrowljq zhaoxuangithub weidqi wzpy 1798064760 lizhaofu kkklia pkucp jaykimbravekjh pachongchong hackbuteer001 jiangxubin jgr98

attention-based-bilstm-relation-extraction's Issues

contact information

I am currently studying relationship extraction. Can you leave a contact information ,so i can discuss it with you?

ValueError: not enough values to unpack (expected 7, got 5)

x_dev: (800, 4, 98)!!
x_dev: (4, 800, 98)!!
Train/Dev split: 7200/800

2018-06-02 20:32:13.323599: I tensorflow/core/platform/cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
(<tf.Tensor 'bidirectional_rnn/fw/fw/transpose_1:0' shape=(?, 98, 800) dtype=float32>, <tf.Tensor 'ReverseV2:0' shape=(?, 98, 800) dtype=float32>)
pos: (?, 98)
WARNING:tensorflow:From /home/mldl/ub16_prj/Attention-Based-BiLSTM-relation-extraction/lstm_attention.py:69: softmax_cross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and will be removed in a future version.
Instructions for updating:

Future major versions of TensorFlow will allow gradients to flow
into the labels input on backprop by default.

See @{tf.nn.softmax_cross_entropy_with_logits_v2}.

Writing to /home/mldl/ub16_prj/Attention-Based-BiLSTM-relation-extraction/runs/1527942734

Traceback (most recent call last):
File "train.py", line 264, in
tf.app.run()
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/platform/app.py", line 126, in run
_sys.exit(main(argv))
File "train.py", line 261, in main
train(x_text, dist1, dist2, y, pos)
File "train.py", line 248, in train
train_step(x_batch, y_batch)
File "train.py", line 204, in train_step
feed_dict)
ValueError: not enough values to unpack (expected 7, got 5)
(venv) mldl@mldlUB1604:~/ub16_prj/Attention-Based-BiLSTM-relation-extraction$ /usr/bin/python3
Python 3.5.2 (default, Nov 17 2016, 17:05:23)
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.

import tensorflow as tf
tf.version
'1.8.0'
quit()
(venv) mldl@mldlUB1604:

when use python3 train.py I got a problem like this: " ValueError: Input 0 of layer dense_1 is incompatible with the layer: its rank is undefined, but the layer requires a defined rank."

Train/Dev split: 7200/800

Traceback (most recent call last):
File "/Users/wuxikun/Downloads/Attention-Based-BiLSTM-relation-extraction-master/train.py", line 162, in
tf.app.run()
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 124, in run
_sys.exit(main(argv))
File "/Users/wuxikun/Downloads/Attention-Based-BiLSTM-relation-extraction-master/train.py", line 158, in main
train()
File "/Users/wuxikun/Downloads/Attention-Based-BiLSTM-relation-extraction-master/train.py", line 61, in train
l2_reg_lambda=FLAGS.l2_reg_lambda)
File "/Users/wuxikun/Downloads/Attention-Based-BiLSTM-relation-extraction-master/att_lstm.py", line 49, in init
self.logits = tf.layers.dense(self.h_drop, num_classes, kernel_initializer=initializer())
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/layers/core.py", line 253, in dense
return layer.apply(inputs)
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/layers/base.py", line 762, in apply
return self.call(inputs, *args, **kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/layers/base.py", line 629, in call
self._assert_input_compatibility(inputs)
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/layers/base.py", line 1122, in _assert_input_compatibility
self.name + ' is incompatible with the layer: '
ValueError: Input 0 of layer dense_1 is incompatible with the layer: its rank is undefined, but the layer requires a defined rank.

Process finished with exit code 1

I guess it cause by the function attention(inputs) which return an output with unknown shape, how can I fix it ? or would you please tell me my mistaken ?

unofficial F1 only 0.694117

Hi SeoSangwoo.I run this code,but the Macro-Average F1 Score (excluding Other) is 0.694117.I noticed that official label range is（1，10），but here is（1，19），does it cause F1 lower than paper(0.84).

Attention Weight of padding tokens?

Please let me know if this model architecture calculates attention weights for padding tokens or not.

Where is the embedding_path "Glove.6B.100d.txt"

python train.py --embedding_path "glove.6B.100d.txt",can you tell me the position of "glove.6B.100d.txt ",Where should I put it?

Images for README

测试集的使用

which version of tensorflow it should be?

Writing to /Users//Attention-Based-BiLSTM-relation-extraction/runs/1526525946

Traceback (most recent call last):
File "train.py", line 262, in
tf.app.run()
File "/usr/local/Cellar/python3/3.6.0_1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "train.py", line 259, in main
train(x_text, dist1, dist2, y, pos)
File "train.py", line 246, in train
train_step(x_batch, y_batch)
File "train.py", line 202, in train_step
feed_dict)

Can I use the code on text data without annotation

I was wondering, can I use a pre-trained model to annotate text data (without annotations)?

How to run?

How would you go about running this so one could input a given text of choice to determine the semantic relations within?

I've looked through and can't seem to identify any way in which to do this.

Thanks in advance.

seosangwoo / attention-based-bilstm-relation-extraction Goto Github PK