By MZD
January 28, 2021 | Blog

How Modern NLP is changing the EdTech Industry

Hello, my name is Adam Louly and I am one of the lead engineers at MZD working on our next generation essay platform (EMMA) and an AI/Machine scoring enthusiast.  

Essay Machine Marking Automation

 

Automated Essay scoring has been used across the assessment industry for some time now, however, the recent advancements in NLP models have significantly improved the feedback, speed, and reliability of automated scoring. The dramatically improved technology is allowing for more assessments to include authentic assessment items, resulting in better assessments that can be automatically scored, at scale. 

NLP is a multidisciplinary field involving linguistics, computer science, and artificial intelligence, which aims to create natural language processing tools for various applications. It should not be confused with computational linguistics, which aims to understand languages by means of computer tools. Natural language processing is used in almost every industry including Education Technology.

The applications of NLP in EdTech are many, some of them are :

  • Text Summarization
  • Test / Exams generation
  • Automated Assessment Scoring

Natural language processing witnessed a remarkable development in recent years, especially with the new Attention Models Mechanism, Thank you to the Google brain team for this great Paper Attention is all you need.

With this new Attention Mechanism, it’s much easier now to understand context-heavy texts, unlike the old way using recurrent neural networks (RNN)  that always had the problem processing very long sequences due to its activation function. 

Models like BERT, RoBERTa, and GPT are making a revolution in NLP and beating every other model on almost every NLP task. In EMMA, we utilize BERT as one of our tools in training our Models for scoring assessments, adding BERT to the pipeline adds a big value in understanding the words in context and dealing with the off-topic responses.

In a couple of years, we will no longer hear about using Classic NLP when it comes to Language Modeling because NLP 2.0 and Attention models are delivering State-Of-The-Art results in many Language tasks, also with the new OpenAI Language model GPT-3, which is the biggest and strongest Language model that was ever trained for (NLG) natural language generation, we will obviously witness a big paradigm shift in many NLP applications and especially in the EdTech Industry.

Adam Louly
Machine Learning Engineer
MZD

Archives