Implemented shortest-circuit disambiguation, maximum probability disambiguation, HMM-based lexical annotation and BiLSTM+CRF-based named entity recognition

Overview

To Startup

进入根目录(ner文件夹 或 seg_tag文件夹),执行:

pip install -r requirements.txt

等待环境配置完成

程序入口为main.py文件,执行:

python main.py

seg_tag文件夹中将会一次性输出:

  1. 最大概率分词结果与P、R、F
  2. 最大概率分词(加法平滑)结果与P、R、F
  3. 最大概率分词(Jelinek-Mercer插值法平滑)结果与P、R、F
  4. 最短路分词结果与P、R、F
  5. 词性标注结果与两种评分的P、R、F
  6. 各算法耗时

ner文件夹中将会输出:

  1. 各标签的数量和各自的P、R、F
  2. 测试集上的P、R、F
  3. 混淆矩阵
  4. 算法耗时

自动分词与词性标注部分

文件结构

D:.
│  clean.ipynb # 处理数据集dag.py # 建图dictionary.py # 建立词典main.py # 程序入口mpseg.py # 最大概率分词模块pos.py # 词性标注模块spseg.py # 最短路分词模块requirements.txttrie.py # trie树score.py # 函数
│
├─data # 数据集sequences.txtwordpieces.txt
│          
└─__pycache__

每个模块均经过单元测试和集成测试

代码注释采用Google风格

建立词典

定义class Trie作为词典数据结构,在Trie的尾节点保存该词出现的次数与词性。

使用Trie可以最大化节约空间开销。

定义class Dictionary作为词典,并统计词频、词性、转移矩阵、发射矩阵等。

基于词典的最短路分词

给定句子sentence[N],调用类SPseg中的spcut方法,代码依次执行:

  1. 依据词典建立有向无环图(调用类DAG
  2. 最短路dp (调用dp函数)
  3. 回溯得到最短路径
  4. 返回分词结果

最短路分词获得的是尽可能小的分词集合。

基于统计的最大概率分词

给定句子sentence[N],调用类MPseg中的mpcut方法,代码依次执行:

  1. 依据词典建立有向无环图(调用类DAG
  2. 根据类Dictionary中统计的词频计算边权(边权为该词出现的概率)
  3. 最短路dp (调用dp函数)
  4. 回溯得到最短路径
  5. 返回分词结果

最大概率分词得到的分词结果y满足 $$ y = argmax{P(y|x)} = argmax \frac{P(x|y)P(y)}{P(x)} $$ 其中$P(x), P(x|y)$是常数,即: $$ y & = argmax P(y|x)\ & = argmax P(y) \ & = argmax \prod_1^n P(w_i) \ & = argmax log(\prod_1^n P(w_i))\ & = argmin (- \sum_i^m log(P(w_i)) )\ $$ 最大概率即可等价于在DAG上求边权为$-log(P)$的最短路径

数据平滑

考虑到unseen event,对于频率为0的事件,我们也应分配一定的概率。

代码给出了两种数据平滑方式:

  1. Adding smoothing (加法平滑方法)
  2. Jelinek-Mercer interpolation (JM插值法)

Adding smoothing: $$ P(w_i) = \frac{\delta + c(w_i)}{\delta|V| + \sum_j c(w_j)} $$ 代码中取$\delta = 1$

Jelinek-Mercer interpolation $$ P(w_i) = \lambda P_{ML}(w_i) + (1-\lambda)P_{unif} $$ 思想为n元模型的概率由n元模型和n-1元模型插值而成

代码中取0元模型为均匀分布:$P_{unif} = \frac{1}{|V|}$,并给出$\lambda$的默认值为0.9

基于HMM的词性标注

HMM是一种概率图模型,基于统计学习得到emission matrix和transition matrix,推断给定观测序列(分词结果)的隐状态(词性序列)。

给出分词结果,调用类WordTagging中的tagging方法,代码依次执行:

  1. 根据词频计算发射概率和转移概率
  2. Viterbi decoding,找到具有最大概率的隐状态序列
  3. 回溯,得到隐状态序列

HMM经Viterbi解码得到的词性序列满足: $$ y & = argmax P(y|x)\ & = argmax \frac{P(y)P(x|y)}{P(x)} \ & = argmax P(y)\ & = argmax {\pi[t_i]b_1[w_1] \prod_1^{n-1} a[t_i][t_{i+1}]b_{i+1}[w_{i+1}]} \ & = argmax {log(\pi[t_i]b_1[w_1] \prod_1^{n-1} a[t_i][t_{i+1}]b_{i+1}[w_{i+1}])}\ & = argmin {-( log(\pi[t_i]) + log(b_1[w_1]) + \sum_i^m {log(a[t_i][t_{i+1}])+log(b_{i+1}[w_{i+1}])} )}\ $$

准确率、召回率、F1 score与性能

由公式: $$ P = \frac{系统输出的正确结果}{系统输出的全部结果个数} \ R = \frac{系统输出的正确结果}{测试集中的结果个数} \ F = \frac{2\times P \times R}{P+R} $$ 执行python main.py命令,在测试数据上推断,可得到上述全部分词、词性标注结果,并得到准确率、召回率、F1 score和性能指标

分词准确率:MP(with JM smoothing) = MP(with Add1 smoothing) > MP(no smoothing) = SP

使用平滑技术能得到更好的分词效果,统计方法(MP)比词典法能得到更好的分词效果。

HMM词性标注中,先利用MP(with JM smoothing) 法分词,再对分词结果进行词性标注。同时采用了粗略的评价指标(不考虑顺序)和严格的评价指标(考虑顺序)。

对于给定的长为N的序列:

Methods Inference Time Complexity
MP分词 $O(N+M)$
SP分词 $O(N+M)$
HMM词性标注 $O(T^2N)$

其中,$M$为DAG中的边数,$T$词性总数。因此三个算法的推断复杂度都是线性的

命名实体识别部分

采用BiLSTM+CRF模型

img

其中,BiLSTM输入是给定的sentence(embedding sequence),输出为该词对应的命名实体标签。它通过双向的设置学习到观测序列(输入的字)之间的依赖,在训练过程中,LSTM能够根据目标(比如识别实体)自动提取观测序列的特征。但是,BiLSTM无法学习到输出序列之间的依赖与约束关系。

CRF等同于在BiLSTM的输出上添加了一层约束,使得模型也能学习到输出序列内部之间的的依赖。传统的CRF需要人为给出特征模板,但在该模型中,特征函数将由模型自行学习得到。

文件结构

D:.
│  dataloader.py # 载入数据集evaluation.py # 评估模型main.py # 程序入口model.py # BiLSTM、BiLSTM+CRF模型utils.py # 函数requirements.txt
│
├─data_ner # 数据集dev.char.bmestest.char.bmestrain.char.bmes
│
├─results # 训练好的模型BiLSTM+CRF.pkl
│
└─__pycache__

参数设置

Total epoches Batch size learning rate hidden size embedding size
30 64 0.001 128 128

每结束一个epoch,模型在验证集上评估,选取在验证集上效果最好的模型作为最终模型(optimal model)。

模型在测试集上能达到95%以上的准确率。

Reference

[1] 宗成庆 《统计自然语言处理》

[2] Lample G, Ballesteros M, Subramanian S, et al. Neural architectures for named entity recognition[J]. arXiv preprint arXiv:1603.01360, 2016.

[3] blog: 1. Understanding LSTM Networks -- colah's blog, 2. CRF Layer on the Top of BiLSTM - 1 | CreateMoMo

[4] code: 1. hiyoung123/ChineseSegmentation: 中文分词 (github.com) ,2. luopeixiang/named_entity_recognition: 中文命名实体识别(github.com), 3. Advanced: Making Dynamic Decisions and the Bi-LSTM CRF — PyTorch Tutorials 1.9.1+cu102 documentation

[5] dataset: 1. jiesutd/LatticeLSTM: Chinese NER using Lattice LSTM. Code for ACL 2018 paper. (github.com), 2. 人民日报1998

Code to reprudece NeurIPS paper: Accelerated Sparse Neural Training: A Provable and Efficient Method to Find N:M Transposable Masks

Accelerated Sparse Neural Training: A Provable and Efficient Method to FindN:M Transposable Masks Recently, researchers proposed pruning deep neural n

itay hubara 4 Feb 23, 2022
Topic Modelling for Humans

gensim – Topic Modelling in Python Gensim is a Python library for topic modelling, document indexing and similarity retrieval with large corpora. Targ

RARE Technologies 13.8k Jan 02, 2023
Yet Another Neural Machine Translation Toolkit

YANMTT YANMTT is short for Yet Another Neural Machine Translation Toolkit. For a backstory how I ended up creating this toolkit scroll to the bottom o

Raj Dabre 121 Jan 05, 2023
Bot to connect a real Telegram user, simulating responses with OpenAI's davinci GPT-3 model.

AI-BOT Bot to connect a real Telegram user, simulating responses with OpenAI's davinci GPT-3 model.

Thempra 2 Dec 21, 2022
Study German declensions (dER nettE Mann, ein nettER Mann, mit dEM nettEN Mann, ohne dEN nettEN Mann ...) Generate as many exercises as you want using the incredible power of SPACY!

Study German declensions (dER nettE Mann, ein nettER Mann, mit dEM nettEN Mann, ohne dEN nettEN Mann ...) Generate as many exercises as you want using the incredible power of SPACY!

Hans Alemão 4 Jul 20, 2022
Sequence-to-Sequence learning using PyTorch

Seq2Seq in PyTorch This is a complete suite for training sequence-to-sequence models in PyTorch. It consists of several models and code to both train

Elad Hoffer 514 Nov 17, 2022
A Multi-modal Model Chinese Spell Checker Released on ACL2021.

ReaLiSe ReaLiSe is a multi-modal Chinese spell checking model. This the office code for the paper Read, Listen, and See: Leveraging Multimodal Informa

DaDa 106 Dec 29, 2022
A2T: Towards Improving Adversarial Training of NLP Models (EMNLP 2021 Findings)

A2T: Towards Improving Adversarial Training of NLP Models This is the source code for the EMNLP 2021 (Findings) paper "Towards Improving Adversarial T

QData 17 Oct 15, 2022
File-based TF-IDF: Calculates keywords in a document, using a word corpus.

File-based TF-IDF Calculates keywords in a document, using a word corpus. Why? Because I found myself with hundreds of plain text files, with no way t

Jakob Lindskog 1 Feb 11, 2022
A retro text-to-speech bot for Discord

hawking A retro text-to-speech bot for Discord, designed to work with all of the stuff you might've seen in Moonbase Alpha, using the existing command

Nick Schorr 23 Dec 25, 2022
Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition

SEW (Squeezed and Efficient Wav2vec) The repo contains the code of the paper "Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speec

ASAPP Research 67 Dec 01, 2022
An assignment from my grad-level data mining course demonstrating some experience with NLP/neural networks/Pytorch

NLP-Pytorch-Assignment An assignment from my grad-level data mining course (before I started personal projects) demonstrating some experience with NLP

David Thorne 0 Feb 06, 2022
Code from the paper "High-Performance Brain-to-Text Communication via Handwriting"

Code from the paper "High-Performance Brain-to-Text Communication via Handwriting"

Francis R. Willett 305 Dec 22, 2022
Paddlespeech Streaming ASR GUI

Paddlespeech-Streaming-ASR-GUI Introduction A paddlespeech Streaming ASR GUI. Us

Niek Zhen 3 Jan 05, 2022
Tevatron is a simple and efficient toolkit for training and running dense retrievers with deep language models.

Tevatron Tevatron is a simple and efficient toolkit for training and running dense retrievers with deep language models. The toolkit has a modularized

texttron 193 Jan 04, 2023
AudioCLIP Extending CLIP to Image, Text and Audio

AudioCLIP Extending CLIP to Image, Text and Audio This repository contains implementation of the models described in the paper arXiv:2106.13043. This

458 Jan 02, 2023
Share constant definitions between programming languages and make your constants constant again

Introduction Reconstant lets you share constant and enum definitions between programming languages. Constants are defined in a yaml file and converted

Natan Yellin 47 Sep 10, 2022
Code for EMNLP20 paper: "ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training"

ProphetNet-X This repo provides the code for reproducing the experiments in ProphetNet. In the paper, we propose a new pre-trained language model call

Microsoft 394 Dec 17, 2022
One Stop Anomaly Shop: Anomaly detection using two-phase approach: (a) pre-labeling using statistics, Natural Language Processing and static rules; (b) anomaly scoring using supervised and unsupervised machine learning.

One Stop Anomaly Shop (OSAS) Quick start guide Step 1: Get/build the docker image Option 1: Use precompiled image (might not reflect latest changes):

Adobe, Inc. 148 Dec 26, 2022