All Projects → yongzhuo → Nlp_xiaojiang

yongzhuo / Nlp_xiaojiang

Licence: mit
自然语言处理(nlp),小姜机器人(闲聊检索式chatbot),BERT句向量-相似度(Sentence Similarity),XLNET句向量-相似度(text xlnet embedding),文本分类(Text classification), 实体提取(ner,bert+bilstm+crf),数据增强(text augment, data enhance),同义句同义词生成,句子主干提取(mainpart),中文汉语短文本相似度,文本特征工程,keras-http-service调用

Programming Languages

python
139335 projects - #7 most used programming language

Projects that are alternatives of or similar to Nlp xiaojiang

Eda nlp for chinese
An implement of the paper of EDA for Chinese corpus.中文语料的EDA数据增强工具。NLP数据增强。论文阅读笔记。
Stars: ✭ 660 (-30.82%)
Mutual labels:  chinese, data-augmentation, text-classification
BERT-chinese-text-classification-pytorch
This repo contains a PyTorch implementation of a pretrained BERT model for text classification.
Stars: ✭ 92 (-90.36%)
Mutual labels:  text-classification, chinese
Weibo terminater
Final Weibo Crawler Scrap Anything From Weibo, comments, weibo contents, followers, anything. The Terminator
Stars: ✭ 2,295 (+140.57%)
Mutual labels:  chatbot, chinese
Snips Nlu
Snips Python library to extract meaning from text
Stars: ✭ 3,583 (+275.58%)
Mutual labels:  chatbot, text-classification
Cnn Question Classification Keras
Chinese Question Classifier (Keras Implementation) on BQuLD
Stars: ✭ 28 (-97.06%)
Mutual labels:  chinese, text-classification
Cnn intent classification
CNN for intent classification task in a Chatbot
Stars: ✭ 90 (-90.57%)
Mutual labels:  chatbot, text-classification
Text Classification Cnn Rnn
CNN-RNN中文文本分类,基于TensorFlow
Stars: ✭ 3,613 (+278.72%)
Mutual labels:  chinese, text-classification
Cluedatasetsearch
搜索所有中文NLP数据集,附常用英文NLP数据集
Stars: ✭ 2,112 (+121.38%)
Mutual labels:  chinese, text-classification
Chatito
🎯🗯 Generate datasets for AI chatbots, NLP tasks, named entity recognition or text classification models using a simple DSL!
Stars: ✭ 678 (-28.93%)
Mutual labels:  chatbot, text-classification
Nlp chinese corpus
大规模中文自然语言处理语料 Large Scale Chinese Corpus for NLP
Stars: ✭ 6,656 (+597.69%)
Mutual labels:  chinese, text-classification
Rasa nlu chi
Turn Chinese natural language into structured data 中文自然语言理解
Stars: ✭ 1,166 (+22.22%)
Mutual labels:  chatbot, chinese
Eda nlp
Data augmentation for NLP, presented at EMNLP 2019
Stars: ✭ 902 (-5.45%)
Mutual labels:  data-augmentation, text-classification
Unilmchatchitrobot
Unilm for Chinese Chitchat Robot.基于Unilm模型的夸夸式闲聊机器人项目。
Stars: ✭ 67 (-92.98%)
Mutual labels:  chatbot, chinese
Nlp pytorch project
Embedding, NMT, Text_Classification, Text_Generation, NER etc.
Stars: ✭ 153 (-83.96%)
Mutual labels:  chatbot, text-classification
Cnn Text Classification Tf Chinese
CNN for Chinese Text Classification in Tensorflow
Stars: ✭ 237 (-75.16%)
Mutual labels:  chinese, text-classification
Chinese Text Classification
Chinese-Text-Classification,Tensorflow CNN(卷积神经网络)实现的中文文本分类。QQ群:522785813,微信群二维码:http://www.tensorflownews.com/
Stars: ✭ 284 (-70.23%)
Mutual labels:  chinese, text-classification
Cluepretrainedmodels
高质量中文预训练模型集合:最先进大模型、最快小模型、相似度专门模型
Stars: ✭ 493 (-48.32%)
Mutual labels:  chinese, text-classification
Lightnlp
基于Pytorch和torchtext的自然语言处理深度学习框架。
Stars: ✭ 739 (-22.54%)
Mutual labels:  chinese, text-classification
Nlp tensorflow project
Use tensorflow to achieve some NLP project, eg: classification chatbot ner attention QAetc.
Stars: ✭ 27 (-97.17%)
Mutual labels:  chatbot, text-classification
Text2gender
Predict the author's gender from their text.
Stars: ✭ 14 (-98.53%)
Mutual labels:  text-classification

nlp_xiaojiang

AugmentText

- 回译(效果比较好)
- EDA(同义词替换、插入、交换和删除)(效果还行)
- HMM-marko(质量较差)
- syntax(依存句法、句法、语法书)(简单句还可)
- seq2seq(深度学习同义句生成,效果不理想,seq2seq代码大都是 [https://github.com/qhduan/just_another_seq2seq] 的,效果不理想)

ChatBot

- 检索式ChatBot
    - 像ES那样直接检索(如使用fuzzywuzzy),只能字面匹配
    - 构造句向量,检索问答库,能够检索有同义词的句子
- 生成式ChatBot(todo)
    - seq2seq
    - GAN

ClassificationText

- bert+bi-lstm(keras) approach 0.78~0.79% acc of weBank Intelligent Customer Service Question Matching Competition
- bert + text-cnn(keras) approach 0.78~0.79% acc of weBank Intelligent Customer Service Question Matching Competition
- bert + r-cnn(keras) approach 0.78~0.79% acc of weBank Intelligent Customer Service Question Matching Competition
- bert + avt-cnn(keras) approach 0.78~0.79% acc of weBank Intelligent Customer Service Question Matching Competition

Ner

- bert命名实体提取(bert12层embedding + bilstm + crf)
    - args.py(配置一些参数)
    - keras_bert_embedding.py(bert embedding)
    - keras_bert_layer.py(layer层, 主要有CRF和NonMaskingLayer)
    - keras_bert_ner_bi_lstm.py(主函数, 定义模型、数据预处理和训练预测等)
    - layer_crf_bojone.py(CRF层, 未使用)

FeatureProject

- bert句向量、文本相似度
    - bert/extract_keras_bert_feature.py:提取bert句向量特征
    - bert/tet_bert_keras_sim.py:测试xlnet句向量cosin相似度
- xlnet句向量、文本相似度
    - xlnet/extract_keras_xlnet_feature.py:提取bert句向量特征
    - xlnet/tet_xlnet_keras_sim.py:测试bert句向量cosin相似度
- normalization_util指的是数据归一化
    - 0-1归一化处理
    - 均值归一化
    - sig归一化处理
- sim feature(ML)
    - distance_text_or_vec:各种计算文本、向量距离等
    - distance_vec_TS_SS:TS_SS计算词向量距离
    - cut_td_idf:将小黄鸡语料和gossip结合
    - sentence_sim_feature:计算两个文本的相似度或者距离,例如qq(问题和问题),或者qa(问题和答案)

run(可以在win10下,pycharm下运行)

  • 1.创建tf-idf文件等(运行2需要先跑1):
    python cut_td_idf.py
  • 2.计算两个句子间的各种相似度,先计算一个预定义的,然后可输入自定义的(先跑1):
    python sentence_sim_feature.py
  • 3.chatbot_1跑起来(fuzzy检索-没)(独立):
    python chatbot_fuzzy.py
  • 4.chatbot_2跑起来(句向量检索-词)(独立):
    python chatbot_sentence_vec_by_word.py
  • 5.chatbot_3跑起来(句向量检索-字)(独立):
    python chatbot_sentence_vec_by_char.py
  • 6.数据增强(eda): python enhance_eda.py
  • 7.数据增强(marko): python enhance_marko.py
  • 8.数据增强(translate_account): python translate_tencent_secret.py
  • 9.数据增强(translate_tools): python translate_translate.py
  • 10.数据增强(translate_web): python translate_google.py
  • 11.数据增强(augment_seq2seq): 先跑 python extract_char_webank.py生成数据, 再跑 python train_char_anti.py 然后跑 python predict_char_anti.py
  • 12.特征计算(bert)(提取特征、计算相似度): run extract_keras_bert_feature.py run tet_bert_keras_sim.py

Data

- chinese_L-12_H-768_A-12(谷歌预训练好的模型)
   github项目中只是上传部分数据,需要的前往链接: https://pan.baidu.com/s/1I3vydhmFEQ9nuPG2fDou8Q 提取码: rket
   解压后就可以啦
- chinese_xlnet_mid_L-24_H-768_A-12(哈工大训练的中文xlnet, mid, 24层, wiki语料+通用语料)
    - 下载地址[https://github.com/ymcui/Chinese-PreTrained-XLNet](https://github.com/ymcui/Chinese-PreTrained-XLNet)
- chinese_vector
    github项目中只是上传部分数据,需要的前往链接: https://pan.baidu.com/s/1I3vydhmFEQ9nuPG2fDou8Q 提取码: rket
    - 截取的部分word2vec训练词向量(自己需要下载全效果才会好)
    - w2v_model_wiki_char.vec、w2v_model_wiki_word.vec都只有部分,词向量w2v_model_wiki_word.vec可以用这个下载地址的替换[https://pan.baidu.com/s/14JP1gD7hcmsWdSpTvA3vKA](https://pan.baidu.com/s/14JP1gD7hcmsWdSpTvA3vKA)

- corpus
    github项目中只是上传部分数据,需要的前往链接: https://pan.baidu.com/s/1I3vydhmFEQ9nuPG2fDou8Q 提取码: rket
    - ner(train、dev、test----人民日报语料)
    - webank(train、dev、test)
    - 小黄鸡和gossip问答预料(数据没清洗),chicken_and_gossip.txt
    - 微众银行和支付宝文本相似度竞赛数据, sim_webank.csv
- sentence_vec_encode_char
    - 1.txt(字向量生成的前100000句向量)
- sentence_vec_encode_word
    - 1.txt(词向量生成的前100000句向量)
- tf_idf(chicken_and_gossip.txt生成的tf-idf)

requestments.txt

- python_Levenshtei
    - 调用Levenshtein,我的python是3.6,
    - 打开其源文件: https://www.lfd.uci.edu/~gohlke/pythonlibs/
    - 查找python_Levenshtein-0.12.0-cp36-cp36m-win_amd64.whl下载即可
- pyemd
- pyhanlp
    - 下好依赖JPype1-0.6.3-cp36-cp36m-win_amd64.whl

参考/感谢

其他资料

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].