All Projects → Agwave → PDF-Resume-Information-Extraction

Agwave / PDF-Resume-Information-Extraction

Licence: other
天池比赛作品整理。实现从pdf中提取出姓名、出生年月、性别、电话、最高学历、籍贯、落户市县、政治面貌、毕业院校、工作单位、工作内容、职务、项目名称、项目责任、学位、毕业时间、工作时间、项目时间共18个字段。

Programming Languages

python
139335 projects - #7 most used programming language
Jupyter Notebook
11667 projects

Projects that are alternatives of or similar to PDF-Resume-Information-Extraction

NER-in-Chinese-Text
NLP Keras BiLSTM+CRF
Stars: ✭ 53 (-17.19%)
Mutual labels:  ner, bilstm-crf
ChineseNER
中文NER的那些事儿
Stars: ✭ 241 (+276.56%)
Mutual labels:  ner, bilstm-crf
naacl2019-select-pretraining-data-for-ner
BiLSTM-CRF model for NER
Stars: ✭ 15 (-76.56%)
Mutual labels:  ner, bilstm-crf
sequence tagging
Named Entity Recognition (LSTM + CRF + FastText) with models for [historic] German
Stars: ✭ 25 (-60.94%)
Mutual labels:  ner, bilstm-crf
NER BiLSTM CRF Chinese
BiLSTM_CRF中文实体命名识别
Stars: ✭ 46 (-28.12%)
Mutual labels:  ner, bilstm-crf
BiLSTM-CRF-NER-PyTorch
This repo contains a PyTorch implementation of a BiLSTM-CRF model for named entity recognition task.
Stars: ✭ 109 (+70.31%)
Mutual labels:  ner, bilstm-crf
resume
Simple and clean resume in Jekyll
Stars: ✭ 40 (-37.5%)
Mutual labels:  resume
nisrulz.github.io
Everything about Nishant Srivastava/nisrulz (online alias)
Stars: ✭ 41 (-35.94%)
Mutual labels:  resume
resumos-aulas-cubos-academy
Exercícios, testes, e experimentos das aulas do curso.
Stars: ✭ 12 (-81.25%)
Mutual labels:  resume
xinlp
把李航老师《统计学习方法》的后几章的算法都用java实现了一遍,实现盒子与球的EM算法,扩展到去GMM训练,后来实现了HMM分词(实现了HMM分词的参数训练)和CRF分词(借用CRF++训练的参数模型),最后利用tensorFlow把BiLSTM+CRF实现了,然后为lucene包装了一个XinAnalyzer
Stars: ✭ 21 (-67.19%)
Mutual labels:  bilstm-crf
mohusman360.github.io
Simple Resume Template with Tailwind CSS
Stars: ✭ 38 (-40.62%)
Mutual labels:  resume
tensorflow-ml-nlp-tf2
텐서플로2와 머신러닝으로 시작하는 자연어처리 (로지스틱회귀부터 BERT와 GPT3까지) 실습자료
Stars: ✭ 245 (+282.81%)
Mutual labels:  ner
easy-resume
🎉 A less is more online resume editor!
Stars: ✭ 116 (+81.25%)
Mutual labels:  resume
jsonresume-theme-caffeine
Caffeine theme for the JSON Resume project
Stars: ✭ 78 (+21.88%)
Mutual labels:  resume
molminer
Python library and command-line tool for extracting compounds from scientific literature. Written in Python.
Stars: ✭ 38 (-40.62%)
Mutual labels:  ner
BlackIQ
Who I am !?
Stars: ✭ 19 (-70.31%)
Mutual labels:  resume
nerus
Large silver standart Russian corpus with NER, morphology and syntax markup
Stars: ✭ 47 (-26.56%)
Mutual labels:  ner
genalog
Genalog is an open source, cross-platform python package allowing generation of synthetic document images with custom degradations and text alignment capabilities.
Stars: ✭ 234 (+265.63%)
Mutual labels:  ner
cv
📄 My CV
Stars: ✭ 31 (-51.56%)
Mutual labels:  resume
SkillNER
A (smart) rule based NLP module to extract job skills from text
Stars: ✭ 69 (+7.81%)
Mutual labels:  ner

0. 摘要

参加了天池的一个pdf简历信息提取的比赛,这里进行回顾、整理和分享

赛题要求从pdf简历中提取出信息,比如说名字,籍贯等。这里搭建了一个BiLSTM-CRF模型,能够从PDF简历中提取出所需的信息。

模型的线上得分是0.727,排名 21/1200+

1. 赛题相关

模型目标:pdf简历 --> 类别信息

2. 思路

在这里插入图片描述 使用python库pdfminer,将pdf简历中的文本提取出来。利用json标注文件,对提取出来的文本进行匹配和BIO标注,每一个字对应一个标注。最后,将标注后的文本送到BiLSM-CRF模型中进行训练。

3. BiLSTM-CRF 模型

在这里插入图片描述 将文本中的每个字进行one-hot编码,经过Embedding层后,每一个字对应一个字向量,所以文本可以用一个矩阵表示。将文本矩阵输入BiLSTM层,输出中每一个字会对应一个类别概率向量,此类别概率向量表示了该字属于各个类别的概率。所以所有字属于各个类别的概率可以用一个类别概率矩阵表示。将此类别概率矩阵输入CRF层,即可得到得分最高的文本标注序列

此处留一个pytorch官方的BiLSTM-CRF教程链接: https://pytorch.org/tutorials/beginner/nlp/advanced_tutorial.html#

4. 代码地址

https://github.com/Agwave/PDF-Resume-Information-Extraction

5. 不足

  1. 没有利用外部文本来训练语言模型。语言模型的文本只利用了训练集的pdf中的文本。
  2. 只使用了字嵌入。中文文本的话还可以结合词嵌入。
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].