DeepNLP AI APP Store
DeepNLP AI Store (http://www.deepnlp.org/store) is a newly released website to let users write genuine reviews, ratings, human evaluation, prompts and share use cases about detailed aspects of AI services from users' perspective (different from researchers' perspective like the LLM/MultiModal benchmarks). DeepNLP wants to build the platform referred as the "Yelp" of AI services, which stems from its role in providing a platform where users can discover and connect with various AI services, AI Apps and Robotics community and ease the burden of customers choosing various AI services, similar to how Yelp helps users find local businesses and services.
Detailed AI Service Use Case
Users can write detailed reviews about some functions about an AI service (ChatGPT/Gemini/Perplexity/Midjourney/) like how AI is doing on "Correct Grammarly Mistakes in Essays", "Acting like doctors about Illness", or text-to-image ability like "Generate Cartoon Characters", "Draw Picture of Fantasy Humanoid", by uploading the screenshots of a conversation or the generated images of AI image generator.
Multi-Aspect Rating
Users can write review and give overall rating from 1 to 5 to each function of AI service, as well as give detailed rating of different aspects of each function, including "Correctness", "Helpfulness" and "Interesting", and customized aspects of each function, such as "Clarity of image", "Image Resolution", "Artistic", "Grammar", "Succintness" and more.
All Categories and People Groups
We cover 30+ different categories of use cases, such as AI Image Generators, AI Assistant & Chatbot, AI Translator, AI Search Engine, AI for kids, AI for adults, AI for elderly, AI in TRAVEL, AI in TRANSPORTATION AI in HEALTHCARE, AI in BUSINESS, AI in FINANCE, AI in EDUCATION, AI in PRODUCTIVITY TOOL, AI in POLITICS, AI in ENTERTAINMENT, AI in NEWS, AI in ART AND SPORTS, AI in LIFESTYLE AI in PAYMENT, AI in SOCIAL, AI in AGRICULTURE AI in SCIENCE, AI in TECHNOLOGY, AI in TRAVEL, AI in TRANSPORTATION, AI in CAR, AI in CHARITY AI in PUBLIC SERVICE, AI in HOUSING, AI in LAW, AI in COMMUNICATION, AI in FOOD
Reviews and Rating of AI App Store By Categories
AI App Store Major AI Service Publisher
Chatbot Assistant
Social AI
Image Generator
AI Writer
Robots Humanoid Quadruped
Tesla Optimus|pub-tesla-optimus
Electric Vehicle
Video Generator
VR Headset/AI Glasses
LLM Reasoning
AI Education
Comparison of AI Services Reviews and Ratings in Multiple Aspects
AI Chatbot Assistant
AI Image Generator Comparison
midjourney vs stable diffusion
stable diffusion vs midjourney
AI Video Generator Comparison
AI Chatbot
AI Writing
Robotaxi
Baidu VS Waymo Robotaxi Reviews
Tesla Cybercab vs Baidu Apollo
Robots
AI Store of Image Generator
Best AI Tools in Image Generator User Ratings Reviews and Showcase
DeepNLP AI store is a platform and community for user to write genuine user reviews and ratings of AI apps and services. User can upload pictures as showcase of AI Image Generator, such as Midjourney, Canva, Stable Diffusion, etc. People can write reviews about some common user prompts (questions or intents), such as "Generate Cartoon Characters", "Draw Fantacy and Humanoids", "Plot Architecture" and more.
AI Image Generator ShowCases
AI Image Generator of FOOD
Ask Midjourney to generate FOOD related images
Ask Stable Diffusion to generate FOOD related images
Ask Canva to generate FOOD related images
Ask Doubao to generate FOOD related images
Ask Dreamina to generate FOOD related images
Ask Zhipu AI to generate FOOD related images
Ask Qwen to generate FOOD related images
AI Image Generator of ANIMAL
Ask Midjourney to generate ANIMAL related images
Ask Stable Diffusion to generate ANIMAL related images
Ask Canva to generate ANIMAL related images
Ask Doubao to generate ANIMAL related images
Ask Dreamina to generate ANIMAL related images
Ask Zhipu AI to generate ANIMAL related images
Ask Qwen to generate ANIMAL related images
AI Image Generator of CELEBRITY
AI Image Generator of CELEBRITY
Ask Midjourney to generate CELEBRITY related images
Ask Stable Diffusion to generate CELEBRITY related images
Ask Canva to generate CELEBRITY related images
Ask Doubao to generate CELEBRITY related images
Ask Dreamina to generate CELEBRITY related images
Ask Zhipu AI to generate CELEBRITY related images
Ask Qwen to generate CELEBRITY related images
AI Image Generator of SELFIES
Ask Midjourney to generate SELFIES related images
Ask Stable Diffusion to generate SELFIES related images
Ask Canva to generate SELFIES related images
Ask Doubao to generate SELFIES related images
Ask Dreamina to generate SELFIES related images
Ask Zhipu AI to generate SELFIES related images
Ask Qwen to generate SELFIES related images
AI Image Generator of SPORTS
Ask Midjourney to generate SPORTS related images
Ask Stable Diffusion to generate SPORTS related images
Ask Canva to generate SPORTS related images
Ask Doubao to generate SPORTS related images
Ask Dreamina to generate SPORTS related images
Ask Zhipu AI to generate SPORTS related images
Ask Qwen to generate SPORTS related images
AI Image Generator of TECHNOLOGY
AI Image Generator of TECHNOLOGY
Ask Midjourney to generate TECHNOLOGY related images
Ask Stable Diffusion to generate TECHNOLOGY related images
Ask Canva to generate TECHNOLOGY related images
Ask Doubao to generate TECHNOLOGY related images
Ask Dreamina to generate TECHNOLOGY related images
Ask Zhipu AI to generate TECHNOLOGY related images
Ask Qwen to generate TECHNOLOGY related images
AI Image Generator of TRAVEL
Ask Midjourney to generate TRAVEL related images
Ask Stable Diffusion to generate TRAVEL related images
Ask Canva to generate TRAVEL related images
Ask Doubao to generate TRAVEL related images
Ask Dreamina to generate TRAVEL related images
Ask Zhipu AI to generate TRAVEL related images
Ask Qwen to generate TRAVEL related images
AI Image Generator of NATURE
Ask Midjourney to generate NATURE related images
Ask Stable Diffusion to generate NATURE related images
Ask Canva to generate NATURE related images
Ask Doubao to generate NATURE related images
Ask Dreamina to generate NATURE related images
Ask Zhipu AI to generate NATURE related images
Ask Qwen to generate NATURE related images
AI Image Generator of DAILY LIFE
AI Image Generator of DAILY LIFE
Ask Midjourney to generate DAILY LIFE related images
Ask Stable Diffusion to generate DAILY LIFE related images
Ask Canva to generate DAILY LIFE related images
Ask Doubao to generate DAILY LIFE related images
Ask Dreamina to generate DAILY LIFE related images
Ask Zhipu AI to generate DAILY LIFE related images
Ask Qwen to generate DAILY LIFE related images
ShowCases
Use Midjourney to Generate Cartoon Characters
Use Midjourney to Draw Fantacy and Humanoids
Use Midjourney to Plot Architecture Use Stable Diffusion to Generate Cartoon Characters Use Stable Diffusion to Generate Fantacy and Humanoids Use Canva to Draw Fantacy and Humanoids
AI Store of AI Video Generator
Use Sora to generate FANTASY related Videos
Use Pika to generate ALL related Videos
Use Kling to generate FANTASY related Videos
Use Runway to generate FANTASY related Videos
Use Sora to generate FANTASY related Videos
AI Store in Law
Best AI in Law and Legal User Ratings Reviews and Showcase
Unlike most Large Language Models(LLM) benchmark and arena's win-rate metric, which is quite difficult for customers to understand (1 vs 1 metric), users' reviews on DeepNLP AI store is about AI tool's performance in a detailed subfield of the industry, such as "Ask ChatGPT Employment Law Related Questions" and "Ask Gemini Contracts and Agreements Related Questions". Users can give rating from 1 to 5 stars on different aspects of the answeres generated by AI systems, including "Overall", "Correctness", "Helpfulness", "Interesting". They can also give rating to any customized aspect of the service, such as "whether the information is realtime", "generation speed", and many others. Here, we are going to cover different sub areas of AI in Law, includine Employment Law, Contracts and Agreements, Business and Corporate Law, Real Estate, Family Law, Personal Injury, Criminal Law, Immigration Law, and Civil Rights Law, etc.
Prompts (Questions) exmaples
- You are a lawyer specialized in employment Law. I will consult you a few questions. And my questions include "Can my employer fire me without cause and how much compensation can I get if I live in California?".
- What are my rights regarding workplace discrimination?
- Act as an employment law expert and please answer this question "Can I fight for unjustified evaluation if I am placed on performance improvement plan?".
Employment Law
ask-chatgpt-employment-law-related-questions
ask perplexity ai employment law related questions
ask gemini employment law related questions
ask claude employment law related questions
Contracts and Agreements
Best AI in Contracts and Agreements
Ask perplexity ai contracts and agreements related questions
Ask chatgpt contracts and agreements related questions
Ask gemini contracts and agreements related questions
Ask claude contracts and agreements related questions
Business and Corporate Law
Best AI in Business and Corporate Law ask gemini business and corporate law related questions
ask chatgpt business and corporate law related questions
ask perplexity ai business and corporate law related questions
ask claude business and corporate law related questions
Real Estate
Best AI in Real Estate Law ask-chatgpt-real-estate-related-questions
ask-perplexity-ai-real-estate-related-questions
ask-gemini-real-estate-related-questions
ask claude real estate related questions
Civil Rights Law
ask perplexity ai civil rights law related questions
ask gemini civil rights law related questions
ask chatgpt civil rights law related questions
ask claude civil rights law related questions
Family Law
Best AI in Family Law Ask chatgpt family law and related question
Ask claude family law related questions
Ask perplexity ai family law related questions
Ask gemini family law related questions
Personal Injury
Ask perplexity ai personal injury law and accidents related question
Ask chatgpt personal injury law and accidents related question
Ask claude personal injury law and accidents related question
Ask gemini personal injury law and accidents related question
Criminal Law
Best AI in Criminal Law Ask gemini criminal law related questions
Ask perplexity ai criminal law related questions
Ask chatgpt criminal law related questions
Ask claude criminal law related questions
Immigration Law
Best AI in immigration law Ask perplexity ai immigration law related questions
Ask gemini immigration law related questions
Ask chatgpt immigration law related questions
Ask gemini criminal law related questions
AI in Finance: User Ratings, Reviews and Showcase
Best AI in Finance User Ratings Reviews and Showcase
Prompts (Questions)
- You are an expert in stock investment, and I will consult you a few questions. The questions include "Is Tesla a buy stock? Should I sell Nvidia stock after financial report season?"
- Please act as a trader. Please answer this question "Summarize Nvidia's GAAP Financial Report Q4 FY24 statistics and compare P/E ratio with other tech company such as google and apple"
- Act as a financial analyst and "Please draw a bar chart comparing the market capital of Google, Apple, Tesla, Nvidia and Microsoft."
AI in Investment Finance
Ask chatgpt facts about investment and seek advice
Ask Gemini Facts about Investment and Seek Advice
Ask Claude INVESTMENT Related Questions
Ask Perplexity AI INVESTMENT Related Questions
AI in INSURANCE Finance
Best AI in Finance INSURANCE User Ratings Reviews and Showcase
Ask Gemini INSURANCE Related Questions
Ask Perplexity AI INSURANCE Related Questions
Ask ChatGPT INSURANCE Related Questions
Ask Claude INSURANCE Related Questions
AI in Mortgage and Loan Finance
Best AI in Mortgage Loan Finance
Ask ChatGPT MORTGAGE & LOAN Related Questions
Ask Gemini MORTGAGE & LOAN Related Questions
Ask Claude MORTGAGE & LOAN Related Questions
Ask Perplexity AI MORTGAGE & LOAN Related Questions
AI in BANKING Finance
Ask Perplexity AI BANKING Related Questions
Ask Gemini BANKING Related Questions
Ask ChatGPT BANKING Related Questions
Ask Claude BANKING Related Questions
AI in DEBT Finance
Ask Claude DEBT Related Questions
Ask ChatGPT DEBT Related Questions
Ask Gemini DEBT Related Questions
Ask Perplexity AI DEBT Related Questions
AI in Healthcare
Best AI in Healthcare User Ratings Reviews and Showcase
HOSPITAL APPOINTMENT
Best AI in HOSPITAL APPOINTMENT
Ask Gemini HOSPITAL APPOINTMENT Related Questions
Ask ChatGPT HOSPITAL APPOINTMENT Related Questions
Ask Perplexity AI HOSPITAL APPOINTMENT Related Questions
Ask Claude HOSPITAL APPOINTMENT Related Questions
HOSPITAL
Best AI in Hospital Healthcare
AI in ILLNESS
Ask Gemini Food to Avoid About Illness
Ask Perplexity AI ILLNESS Related Questions
Ask ChatGPT ILLNESS Related Questions
Ask Claude ILLNESS Related Questions
MEDICINE
Best AI in Medicine Healthcare
Ask ChatGPT MEDICINE Related Questions
Ask Perplexity AI MEDICINE Related Questions
Ask Gemini MEDICINE Related Questions
Ask Claude MEDICINE Related Questions
NURSING
Ask Gemini NURSING Related Questions
Ask Claude NURSING Related Questions
Ask Perplexity AI NURSING Related Questions
Ask ChatGPT NURSING Related Questions
BEAUTY
Ask Gemini NURSING Related Questions
Ask Claude NURSING Related Questions
Ask Perplexity AI NURSING Related Questions
Ask ChatGPT NURSING Related Questions
ELDERLY CARE
Ask Perplexity AI ELDERLY CARE Related Questions
Ask Gemini ELDERLY CARE Related Questions
Ask ChatGPT ELDERLY CARE Related Questions
Ask Claude ELDERLY CARE Related Questions
MEDICAL INSTRUMENT
Ask ChatGPT MEDICAL INSTRUMENT Related Questions
Ask Claude MEDICAL INSTRUMENT Related Questions
Ask Perplexity AI MEDICAL INSTRUMENT Related Questions
Ask Gemini MEDICAL INSTRUMENT Related Questions
AI for Elderly: User Ratings, Reviews and Showcase
Best AI for Elderly User Ratings, Reviews and Showcase
Prompts (Questions) exmaples What are the side effects of [medication]? How can I manage [illness]?
e.g. What are the side effects of Tylenol? How can I manage arthritis pain?
ShowCases
Elderly People Ask Gemini Personal Hobbies Related Questions
Elderly People Ask Gemini Legal Questions
Elderly People Ask Gemini Health and Medicine Related Questions
Elderly People Ask ChatGPT Health and Medicine Related Questions
Elderly People Ask Perplexity AI Health and Medicine Related Questions
Elderly People Ask Claude Health and Medicine Related Questions
AI for Kids: User Ratings, Reviews and Showcase
Best AI for Kids User Ratings, Reviews and Showcase
Prompts (Questions) exmaples
STORY TELLING for KIDS
- Please tell me a bedtime story about unicorns for my 4 years old daughter.
- Help me write a story about Paw Patrol Dogs Fighting Bad Guys
DRAWING|AIGC
- Help me Draw a picture about Paw Patrol Dogs Chase Driving Police Car
- Help me Draw a picture Peppa Pig playing in the swimming pool
ShowCases Ask Gemini to Tell a Bedtime Story Ask Gemini to Draw Cartoon Characters Ask Gemini to Tell a Bedtime Story Ask ChatGPT to Draw Cartoon Character
AI in LifeStyle
Best AI in LifeStyle User Ratings, Reviews and Showcase
Prompts (Questions) exmaples
Act as ${role}
I want you to respond and answer like ${role} using the tone, manner that ${role} would use. Do not write any explanations. My first sentence is ${your_question}.
ShowCases
Ask ChatGPT to Act As Lovers in a Relationship
Ask Gemini to Act As Lovers in a Relationship
Ask Character AI to Act As Lovers in a Relationship
Ask Doubao to Act As Lovers in a Relationship
Ask Qwen AI from Alibaba to Act As Lovers in a Relationship
Ask Zhipu AI to Act As Lovers in a Relationship
AI in Productivity Tool
Best AI in Productivity Tool User Ratings, Reviews and Showcase
Writing Tool
Research Analysis
Best AI in Research & Analysis
Programming
Prompts: Show me the ${programming language} code of ${algorithm}, no explanation. Write the ${programming language} to implement ${function}, no explanation. Generate the ${programming language} code of a ${description} website, no explanation. Find the ${statistic} from ${data_source} and display in ${format}.
e.g.
- Show me the latex code of KL Divergence
- Write the python code for QuickSort Implementation
- Generate the html code of a login page of a community website, no explanation.
- Find the 2023 divorce rate of all states in US and display in the table and line chart.
ShowCase
Programming
Gemini for Programming ChatGPT for Programming
Plot Chat
Use ChatGPT to Draw Plot of Math Function
Data Analysis
Use Gemini to Conduct Data Analysis
Quadruped Robot Reviews
Boston Dynamics Spot Robot Dog Customer Reviews
Unitree Go2 Robot Dog Customer Reviews
Anybotics Robot Dog Customer Reviews
Humanoid Robot Reviews
Figure 02 Humanoid Robots Reviews
Tesla Cybercab Robotaxi User Reviews from Earlybirds
Tesla Optimus Humanoid Robots Reviews
Robotaxi Reviews
Take a Ride in Waymo ROBOTAXI And Write Your Reviews
Take a Ride in Baidu Apollo ROBOTAXI And Write Your Reviews
Take a Ride using Full Self-Driving FSD ability of Tesla
Electric Vehicle Reviews
BMW i4 EV Reviews By Car Owners
Hyundai IONIQ 6 EV Reviews By Car Owners
BYD Seal EV Reviews By Car Owners
Tesla Model 3 EV Reviews By Car Owners
DeepNLP AI & Robots Community
DeepNLP AI & Robots Community for AI Practitioner
Would you share your experience using AI Productivity Tools such as AI Writing Coding CoPilot
What are the features you need for AI Video Generator apps and tools
Which one is the best AI Video Generator Runway Luma Pika Kling and Why
What are the typical use scenarios of Quadruped Robot Dogs
Humanoid Robot Husband Vote for the most popular appearance of Humanoid Robot Husband
Humanoid Robot Wife Who would you choose your humanoid robot wife to look like human females
What are the most important features Humanoid Robot should have in the future
What are the typical use cases of Humanoid Robots
DeepNLP Agent Tools
AI Agent Visualization Review Asynchronous Multi-Agent Simulation
Dialogue Visualization Agent Multimodal Visualization Tools for AI Systems A Review
Related Blog
http://www.deepnlp.org/equation/
http://www.deepnlp.org/search/
http://www.deepnlp.org/workspace/ai_courses/
http://www.deepnlp.org/workspace/aigc_chart/
http://www.deepnlp.org/workspace/ai_writer/
http://www.deepnlp.org/workspace/detail/
Machine Learning Equation Formula
Introduction to multimodal generative models
Generative AI Search Engine Optimization: How to Improve Your Content
AI in Fashion: Tell IWC Schaffhausen Watches Real or Fake
AI in Fashion: Tell Fendi bags real or fake
AI in Fashion: Tell Coach bags real or fake
AI in Fashion: Tell Prada bags real or fake
AI in Fashion: Tell Gucci bags real or fake
AI in Fashion: Tell Dior bags real or fake
AI in Fashion: Tell Hermes bags real or fake
AI in Fashion: Tell Chanel bags real or fake
AI in Fashion: Tell Louis Vuitton bags real or fake
AI in Fashion: Tell Omega Watches real or fake
AI in Fashion: Tell Rolex Watches real or fake
DeepNLP Ecommerce Review Panel
DeepNLP Ecommerce Bags Review Panel
DeepNLP Watch Bags Review Panel
DeepNLP Review Ecommerce Brand List
AI Agent Visualization Review Asynchronous Multi-Agent Simulation
Dialogue Visualization Agent Multimodal Visualization Tools for AI Systems A Review
======================================================================================================================================================================================================
By End of 2020, deepnlp lib is archived and only support TF to 1.13
Deep Learning NLP Pipeline implemented on Tensorflow. Following the 'simplicity' rule, this project aims to use the deep learning library of Tensorflow to implement new NLP pipeline. You can extend the project to train models with your own corpus/languages. Pretrained models of Chinese corpus are distributed. Free RESTful NLP API are also provided. Visit http://www.deepnlp.org/api/v1.0/pipeline for details.
Brief Introduction
Modules
-
NLP Pipeline Modules:
- Word Segmentation/Tokenization
- Part-of-speech (POS)
- Named-entity-recognition(NER)
- Dependency Parsing (Parse)
- textsum: automatic summarization Seq2Seq-Attention models
- textrank: extract the most important sentences
- textcnn: document classification
- Web API: Free Tensorflow empowered web API
- Planed: Automatic Summarization
-
Algorithm(Closely following the state-of-Art)
- Word Segmentation: Linear Chain CRF(conditional-random-field), based on python CRF++ module
- POS: LSTM/BI-LSTM/LSTM-CRF network, based on Tensorflow
- NER: LSTM/BI-LSTM/LSTM-CRF network, based on Tensorflow
- Parse: Arc-Standard System with Feed Forward Neural Network
- Textsum: Seq2Seq with attention mechanism
- Texncnn: CNN
-
Pre-trained Model
- Chinese: Segmentation, POS, NER, Parse (1998 china daily corpus)
- Domain Specific NER Models are also provided: general, entertainment, o2o, etc... Contribution are welcome
- English: POS (brown corpus)
- For your Specific Language, you can easily use the script to train model with the corpus of your language choice.
Installation
-
Requirements
- CRF++ (>=0.54)
- Tensorflow(1.4)
- Python (python2.7 and python3.6 are tested) This project is up to date with the latest tensorflow release. For tensorflow (<=0.12.0), use deepnlp <=0.1.5 version. tensorflow (1.0-1.3), use deepnlp = 0.1.6 version tensorflow (1.4), use deepnlp = 0.1.7 version See RELEASE.md for more details
-
Pip
# linux, run the script:
pip install deepnlp
Due to pkg size restriction, english pos model, ner domain specific model files are not distributed on pypi You can download the pre-trained model files from github and put in your installation directory .../site-packages/.../deepnlp/... model files: ../pos/ckpt/en/pos.ckpt ; ../ner/ckpt/zh/ner.ckpt
- Source Distribution, e.g. deepnlp-0.1.7.tar.gz: https://pypi.python.org/pypi/deepnlp
# linux, run the script:
tar zxvf deepnlp-0.1.7.tar.gz
cd deepnlp-0.1.7
python setup.py install
- Initial setup
# install crf++0.58 package using the script
sh ./deepnlp/segment/install_crfpp.sh
# Download all the pre-trained models
python ./test/test_install.py
# Or Download pre-trained models from below command lines
import deepnlp
deepnlp.download('segment')
deepnlp.download('pos')
deepnlp.download('ner')
deepnlp.download('parse')
- Running Examples
# ./deepnlp/test folder
cd test
python test_segment.py # segmentation
python test_pos_en.py # POS tag
python test_ner_zh.py # NER Zh
python test_ner_domain.py # NER domain-specific models
python test_ner_dict_udf.py # NER load user dict and UDF for disambiguation
python test_nn_parser.py # dependency parsing
python test_api_v1_module.py
python test_api_v1_pipeline.py
Tutorial
Set Coding
设置编码 For python2, the default coding is ascii not unicode, use future module to make it compatible with python3
#coding=utf-8
from __future__ import unicode_literals # compatible with python3 unicode
Download pretrained models
下载预训练模型 If you install deepnlp via pip, the pre-trained models are not distributed due to size restriction. You can download full models for 'Segment', 'POS' en and zh, 'NER' zh, zh_entertainment, zh_o2o, 'Textsum' by calling the download function.
import deepnlp
# Download all the modules
deepnlp.download()
# Download specific module
deepnlp.download('segment')
deepnlp.download('pos')
deepnlp.download('ner')
deepnlp.download('parse')
# Download module and domain-specific model
deepnlp.download(module = 'pos', name = 'en')
deepnlp.download(module = 'ner', name = 'zh_entertainment')
Segmentation
分词模块
#coding=utf-8
from __future__ import unicode_literals
from deepnlp import segmenter
tokenizer = segmenter.load_model(name = 'zh_entertainment')
text = "我刚刚在浙江卫视看了电视剧老九门，觉得陈伟霆很帅"
segList = tokenizer.seg(text)
text_seg = " ".join(segList)
#Results
# 我 刚刚 在 浙江卫视 看 了 电视剧 老九门 ， 觉得 陈伟霆 很 帅
POS
词性标注
#coding:utf-8
from __future__ import unicode_literals
import deepnlp
deepnlp.download('pos')
## English Model
from deepnlp import pos_tagger
tagger = pos_tagger.load_model(name = 'en') # Loading English model, lang code 'en', English Model Brown Corpus
text = "I want to see a funny movie"
words = text.split(" ") # unicode
print (" ".join(words))
tagging = tagger.predict(words)
for (w,t) in tagging:
pair = w + "/" + t
print (pair)
#Results
#I/nn want/vb to/to see/vb a/at funny/jj movie/nn
## Chinese Model
from deepnlp import segmenter
from deepnlp import pos_tagger
tagger = pos_tagger.load_model(name = 'zh') # Loading Chinese model, lang code 'zh', China Daily Corpus
text = "我爱吃北京烤鸭"
words = segmenter.seg(text) # words in unicode coding
print (" ".join(words))
tagging = tagger.predict(words) # input: unicode coding
for (w,t) in tagging:
pair = w + "/" + t
print (pair)
#Results
#我/r 爱/v 吃/v 北京/ns 烤鸭/n
NER
命名实体识别
from __future__ import unicode_literals # compatible with python3 unicode
import deepnlp
deepnlp.download('ner') # download the NER pretrained models from github if installed from pip
from deepnlp import ner_tagger
# Example: Entertainment Model
tagger = ner_tagger.load_model(name = 'zh_entertainment') # Base LSTM Based Model
#Load Entertainment Dict
tagger.load_dict("zh_entertainment")
text = "你 最近 在 看 胡歌 演的 猎场 吗 ?"
words = text.split(" ")
tagset_entertainment = ['actor', 'role_name', 'teleplay', 'teleplay_tag']
tagging = tagger.predict(words, tagset = tagset_entertainment)
for (w,t) in tagging:
pair = w + "/" + t
print (pair)
#Result
#你/nt
#最近/nt
#在/nt
#看/nt
#胡歌/actor
#演的/nt
#猎场/teleplay
#吗/nt
#?/nt
Parsing
依存句法分析
from __future__ import unicode_literals # compatible with python3 unicode coding
from deepnlp import nn_parser
parser = nn_parser.load_model(name = 'zh')
#Example 1, Input Words and Tags Both
words = ['它', '熟悉', '一个', '民族', '的', '历史']
tags = ['r', 'v', 'm', 'n', 'u', 'n']
#Parsing
dep_tree = parser.predict(words, tags)
#Fetch result from Transition Namedtuple
num_token = dep_tree.count()
print ("id\tword\tpos\thead\tlabel")
for i in range(num_token):
cur_id = int(dep_tree.tree[i+1].id)
cur_form = str(dep_tree.tree[i+1].form)
cur_pos = str(dep_tree.tree[i+1].pos)
cur_head = str(dep_tree.tree[i+1].head)
cur_label = str(dep_tree.tree[i+1].deprel)
print ("%d\t%s\t%s\t%s\t%s" % (cur_id, cur_form, cur_pos, cur_head, cur_label))
# Result
id word pos head label
1 它 r 2 SBV
2 熟悉 v 0 HED
3 一个 m 4 QUN
4 民族 n 5 DE
5 的 u 6 ATT
6 历史 n 2 VOB
Pipeline
#coding:utf-8
from __future__ import unicode_literals
from deepnlp import pipeline
p = pipeline.load_model('zh')
#Segmentation
text = "我爱吃北京烤鸭"
res = p.analyze(text)
print (res[0].encode('utf-8'))
print (res[1].encode('utf-8'))
print (res[2].encode('utf-8'))
words = p.segment(text)
pos_tagging = p.tag_pos(words)
ner_tagging = p.tag_ner(words)
print (pos_tagging.encode('utf-8'))
print (ner_tagging.encode('utf-8'))
Textsum
自动文摘
See details: README
Textrank
重要句子抽取
See details: README
TextCNN (WIP)
文档分类
Train your model
自己训练模型 ###Segment model See instructions: README
###POS model See instructions: README
###NER model See instructions: README
###Parsing model See instructions: README
###Textsum model See instructions: README
Web API Service
www.deepnlp.org provides free web API service for common NLP modules of sentences and paragraphs. The APIs are RESTful and based on pre-trained tensorflow models. Chinese language is now supported.
- RESTful API
- Segmentation: http://www.deepnlp.org/api/v1.0/segment/?lang=zh&text=我爱吃北京烤鸭
- POS: http://www.deepnlp.org/api/v1.0/pos/?lang=zh&text=我爱吃北京烤鸭
- NER: http://www.deepnlp.org/api/v1.0/ner/?lang=zh&text=我爱吃北京烤鸭
- Pipeline: http://www.deepnlp.org/api/v1.0/pipeline/?lang=zh&annotators=segment,pos,ner&text=我爱吃北京烤鸭
Testing API from Browser, Need to log in first
Calling API from python
See ./deepnlp/test/test_api_v1_module.py for more details.
#coding:utf-8
from __future__ import unicode_literals
import json, requests, sys, os
if (sys.version_info>(3,0)): from urllib.parse import quote
else : from urllib import quote
from deepnlp import api_service
login = api_service.init() # registration, if failed, load default empty login {} with limited access
conn = api_service.connect(login) # save the connection with login cookies
# Sample URL
# http://www.deepnlp.org/api/v1.0/pipeline/?lang=zh&annotators=segment,pos,ner&text=我爱吃上海小笼包
# Define text and language
text = ("我爱吃上海小笼包").encode("utf-8") # convert text from unicode to utf-8 bytes
# Set up URL for POS tagging
url_pos = 'http://www.deepnlp.org/api/v1.0/pos/?"+ "lang=" + quote('zh') + "&text=" + quote(text)
web = requests.get(url_pos, cookies = conn)
tuples = json.loads(web.text)
print (tuples['pos_str'].encode('utf-8')) # POS json {'pos_str', 'w1/t1 w2/t2'} return string
中文简介
deepnlp项目是基于Tensorflow平台的一个python版本的NLP套装, 目的在于将Tensorflow深度学习平台上的模块，结合 最新的一些算法，提供NLP基础模块的支持，并支持其他更加复杂的任务的拓展，如生成式文摘等等。
-
NLP 套装模块
- 分词 Word Segmentation/Tokenization
- 词性标注 Part-of-speech (POS)
- 命名实体识别 Named-entity-recognition(NER)
- 依存句法分析 Dependency Parsing (Parse)
- 自动生成式文摘 Textsum (Seq2Seq-Attention)
- 关键句子抽取 Textrank
- 文本分类 Textcnn (WIP)
- 可调用 Web Restful API
- 计划中: 句法分析 Parsing
-
算法实现
- 分词: 线性链条件随机场 Linear Chain CRF, 基于CRF++包来实现
- 词性标注: 单向LSTM/ 双向BI-LSTM, 基于Tensorflow实现
- 命名实体识别: 单向LSTM/ 双向BI-LSTM/ LSTM-CRF 结合网络, 基于Tensorflow实现
- 依存句法分析: 基于arc-standard system的神经网络的parser
-
预训练模型
- 中文: 基于人民日报语料和微博混合语料: 分词, 词性标注, 实体识别
API 服务
http://www.deepnlp.org 出于技术交流的目的, 提供免费API接口供文本和篇章进行深度学习NLP的分析, 简单注册后就可以使用。 API符合RESTful风格, 内部是基于tensorflow预先训练好的深度学习模型。具体使用方法请参考博客: http://www.deepnlp.org/blog/tutorial-deepnlp-api/
API目前提供以下模块支持：
- 分词: http://www.deepnlp.org/api/v1.0/segment/?lang=zh&text=我爱吃北京烤鸭
- 词性标注: http://www.deepnlp.org/api/v1.0/pos/?lang=zh&text=我爱吃北京烤鸭
- 命名实体识别: http://www.deepnlp.org/api/v1.0/ner/?lang=zh&text=我爱吃北京烤鸭
- Pipeline: http://www.deepnlp.org/api/v1.0/pipeline/?lang=zh&annotators=segment,pos,ner&text=我爱吃北京烤鸭
安装说明
-
需要
- CRF++ (>=0.54) 可以从 https://taku910.github.io/crfpp/ 下载安装
- Tensorflow(1.0) 这个项目的Tensorflow函数会根据最新Release更新，目前支持Tensorflow 1.0版本，对于老版本的Tensorflow(<=0.12.0), 请使用 deepnlp <=0.1.5版本, 更多信息请查看 RELEASE.md
-
Pip 安装
pip install deepnlp
- 从源码安装, 下载deepnlp-0.1.7.tar.gz文件: https://pypi.python.org/pypi/deepnlp
# linux, run the script:
tar zxvf deepnlp-0.1.7.tar.gz
cd deepnlp-0.1.7
python setup.py install
- 初始设置
# 运行脚本安装 crf++0.58 包
sh ./deepnlp/segment/install_crfpp.sh
# 运行脚本下载预训练模型测试
python ./test/test_install.py
Reference
- CRF++ package: https://taku910.github.io/crfpp/#download
- Tensorflow: https://www.tensorflow.org/
- Word Segmentation Using CRF++ Blog: http://www.52nlp.cn/中文分词入门之字标注法4
- Blogs http://www.deepnlp.org/blog/