NLP Architect – An Awesome Open Source NLP Python Library from Intel AI Lab (with GitHub link)

Pranav Dar 10 May, 2019 • 3 min read


  • Intel AI Lab has introduced an open source python library for NLP, called NLP Architect
  • The library comes with state-of-the-art NLP models on a variety of topics, including dependency parsing, reading comprehension, text chunking, among others
  • The library also includes a neat looking visualizer which shows your model’s annotations



Have you noticed how common chatbots have become lately? Almost all business websites seem to have a chatbot tucked away on their home page. This, along with other multiple and diverse examples are applications of Natural language processing. It’s potential is seemingly unlimited and the general perception is that we are only just scratching the surface!

So it comes as no surprise that the big tech giants like Google, Facebook and Intel (among others) have opened research divisions to explore this field. The latest NLP offering, called ‘NLP Architect’ comes from the Intel AI Lab.

NLP Architect is an open source Python library that enables data scientists and developers to explore state-of-the-art deep learning techniques in the field of natural language processing (NLP) and natural language understandings (NLU). According to Intel, this library includes their past and currently ongoing research and development efforts.

The existing version of NLP Architect includes features which aim to provide support for both research and practical applications. These features are:

  • NLP core models which enable extraction of linguistic features for NLP workflow: dependency parser (BIST) and NP chunker
  • NLU modules that deliver high-class performance: intent extraction (IE), name entity recognition (NER)
  • Modules that look into semantic understanding: colocations, most common word sense, NP embedding representation
  • Components that key for conversational AI: ChatBot applications, including dialog system, sequence chunking, IE
  • End-to-end deep learning applications leveraging new topologies: Q&A, machine reading comprehension

Below is the framework of the library:

The library comes with the NLP Architect Server, which has been designed with the aim of making predictions across different models in NLP Architect. This server includes a visualizer that shows you your model’s annotations in a pretty neat way. Check out an example below:

Various popular open source frameworks have been leveraged in the NLP Architect repository:

  • Intel® Nervana™ graph
  • Intel® neon
  • Tensorflow or Intel-Optimized TensorFlow
  • Dynet
  • Keras

You can check out their GitHub library here to get a taste of what you can do with Intel’s resources.


Our take on this

Last month we saw the folks at Intel AI Lab open sourcing nGraph, a framework neutral tool that allows data scientists to focus on their data science work, rather than worrying about hardware related limitations. This research arm of Intel is proving to be a truly incredible boon for the machine learning community as a whole.

This library will help both beginners and advanced machine learning developers. For beginners, it helps by providing ready-made examples you can replicate and for advanced users, you can incorporate it into your existing frameworks (if possible) or build your next project using it.

In future releases, the researchers are planning to implement and show the use cases of this library on real-life applications of their customers. I’m looking forward to data scientists from the Analytics Vidhya community utilizing this library and building applications of their own!


Subscribe to AVBytes here to get regular data science, machine learning and AI updates in your inbox!


Pranav Dar 10 May 2019

Senior Editor at Analytics Vidhya. Data visualization practitioner who loves reading and delving deeper into the data science and machine learning arts. Always looking for new ways to improve processes using ML and AI.

Frequently Asked Questions

Lorem ipsum dolor sit amet, consectetur adipiscing elit,

Responses From Readers