deeplaying/Text-Pairs-Relation-Classification

About Text Pairs (Sentence Level) Classification (Similarity Modeling) Based on Neural Network.

Deep Learning for Text Pairs Relation Classification

This project is used by my bachelor graduation project, and it is also a study of TensorFlow, Deep Learning(CNN, RNN, LSTM, etc.).

The main objective of the project is to determine whether the two sentences are similar in sentence meaning (binary classification problems) by the two given sentences based on Neural Networks (Fasttext, CNN, LSTM, etc.).

Requirements

• Python 3.6
• Tensorflow 1.8 +
• Numpy
• Gensim

Data

Research data may attract copyright protection under China law. Thus, there is only code.

实验数据属于实验室与某公司的合作项目，涉及商业机密，在此不予提供，还望谅解。

Innovation

Data part

1. Make the data support Chinese and English.(Which use jieba seems easy)
2. Can use your own pre-trained word vectors.(Which use gensim seems easy)
3. Add embedding visualization based on the tensorboard.

Model part

1. Deign two subnetworks to solve the task --- Text Pairs Similarity Classification.
2. Add the correct L2 loss calculation operation.
3. Add gradients clip operation to prevent gradient explosion.
4. Add learning rate decay with exponential decay.
5. Add a new Highway Layer.(Which is useful based on the performance)
6. Add Batch Normalization Layer.
7. Add several performance measures(especially the AUC) since the data is imbalanced.

Code part

1. Can choose to train the model directly or restore the model from checkpoint in train.py.
2. Add test.py, the model test code, it can show the predict value of label of the data in Testset when create the final prediction file.
3. Add other useful data preprocess functions in data_helpers.py.
4. Use logging for helping recording the whole info(including parameters display, model training info, etc.).

Data Preprocessing

Depends on what your data and task are.

Text Segment

You can use jieba package if you are going to deal with the chinese text data.

Pre-trained Word Vectors

• Use gensim package to pre-train data.
• Use glove tools to pre-train data.
• Even can use a fasttext network to pre-train data.

Network Structure

FastText

References:

• Bag of Tricks for Efficient Text Classification

---

TextANN

---

TextCNN

References:

• Convolutional Neural Networks for Sentence Classification
• A Sensitivity Analysis of (and Practitioners' Guide to) Convolutional Neural Networks for Sentence Classification

---

TextRNN

References:

• Recurrent Neural Network for Text Classification with Multi-Task Learning

---

About Me

黄威，Randolph

SCU SE Bachelor; USTC CS Master

Email: chinawolfman@hotmail.com

My Blog: randolph.pro

LinkedIn: randolph's linkedin