Skip to content

AI framework/starter kit for content publishing tools

License

Notifications You must be signed in to change notification settings

nycdidar/Content-AI

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Content AI

Content AI framework/starter kit.

Goal

Content editors/writers are the heart of any digital news space. This starter kit API can make content publishing tools better, faster and smarter. For ex: curation, cms, api, any application backend or even FE apps.

Core Benefits
  • Reduces UI components.
  • Saves time in writing stories.
  • Helps writing better stories.
  • Reduces 'time to 1st publish' duration.

Example UI Integration

Content AI

Features

Predict next word based on current word.
Classify breaking news.
Classify content category or type.
Predict probability of high traffic content.
Display wikipedia summery as you type.
Using facial recognition detect political figures to automatically select taxonomy or content category. Image-AI
Find image caption by image URL. Image-AI
Analyze image based on several pre trained image models. Image-AI
Subject priority image cropping. Image-AI
Context specific spell checker.
Extract keywords & weights.

Algorithm Used

Recurrent Neural Network

Recurrent Neural Network (RNN)

Naive Bayes classifier

Naive Bayes classifier

Natural language processing

Natural language processing (NLP)

Examples

Classify title

curl http://localhost:5004/classify_title?title=Video shows jaw-dropping scale of Dorian's ferocious hit on the Bahamas

Example Response: [{"category":"news","probability":"0.9941731"},{"category":"tech","probability":"0.0047676545"},{"category":"politics","probability":"0.0007970525"},{"category":"health","probability":"0.00026003388"},{"category":"business","probability":"2.2758081e-06"}]

Classify breaking news

curl http://localhost:5004/classify_breaking?title=Watch live: Latest on Hurricane Dorian as it move up U.S. coast

Example Response: [{"category":"breaking","probability":"0.7305469"},{"category":"normal","probability":"0.26945308"}]

Named Entity Recognition (NER)

curl curl http://localhost:5004/ner?content=<YOUR LARGE CONTENT BLOB>

Example Response: [["London","B-LOC"],["you","O"],["Trump","B-PER"].....

Predict Content Category

curl 'http://localhost:5004/predict_vertical' --data 'content=<YOUR LARGE CONTENT BLOB>'

Example Response: {"vertical_type":"Mach"}

{"keywords":{"politics":0.179,"obama":0.179,"election":0.179,"sanders":0.179......}}

Predict Next Word

curl 'http://localhost:5004/output?string=financial&work'

Example Response: { predictions:[ "planning" ] }

Context Based Spell Checker

curl 'http://localhost:5004/output?string=financd'

Example Response:

{ spell_suggestion:"finance" }

=============================================

Deploy (Using Docker Compose)

cd docker

docker-compose up -d

Deploy (Native)

cd docker

pip install -r requirements.txt

cd ..

python server.py

Visit http://localhost:5004/test to play around with example GUI.

THIS IS WORK IN PROGRESS.

TO DO

  • Organize files/folders.
  • Use more OOP structure.
  • Add more comments.
  • Reduce code footprints.
  • and so on .. huge list.

FULL CREDIT GOES TO EVERYONE INVOLVED IN ML/AI FIELD. THEY ARE TOTALLY RESPONSIBLE FOR MANKINDS EXTINCTION. :-)