NLP for Social Good: Research Paper Review


In this article I discuss a research paper by  Lin et al (2021) [1] named “How Good Is NLP? A Sober Look at NLP Tasks through the Lens of Social Impact”.   

This paper presents an in-depth overview of research areas for NLP in 2020,  effect of research in NLP on targets for a better world and improved society. The analysis for choice of research topic for a productive NLP based application and technology development. Existing NLP  techniques and missing points are analysed too. Focus on Social Good from NLP based technologies.

My Analysis About Article:

 The following are key points as per my readings:

  1. The authors introduces various positive applications of NLP in recent years especially the challenging pandemic times, and how NLP helped us in it.  
  2. The negative impacts of NLP has also being mentioned, which includes bot using anonymous language in comments, privacy issues in various software’s using NLP engines and deep learning based gender and other bias.
  3. AI for social good and ethics in general in AI had been explored. United Nations (UN) sustainable development goals (SDGs) have also been touched in the research.
  4. Mathematical expression of the following has been expressed: “Which best technology can improve Social Impact?”
  5. Connection to answering social good has been linked to theories of philosophy along with their acceptance to the people in this domain (which are philosophers)
  6. ACL 2020, research papers have been analysed for getting the key NLP areas in trend, in which country, what topics and other statistics about the same.
  7. NLP technology development is divided in 4 specific tasks including (1)inception (2) model development (3) application and finally (4) deployment.
  8. Rest of the paper computes impact of contribution in a technology by a researcher.
  9. Finally, discussion has been made for development of deciding research priority given the targets- social impact, SDGs to mention a few.

My Comments:

The paper presents an in-depth analysis of the authors for NLP for social good and global challenges. They have also provided mathematical representation of the computation of suitability of a research topic as a viable topic for an improved socially impactable and relevant NLP research.  But the experiments related to these models were missing. Though statistics were presented for ACL 2020 papers, in a well-documented tabular form. Conclusions were drawn of what is missing from NLP research and how to include it, for instance, the gender bias in NLP tasks. Further, case studies were provided of several key NLP areas including Green NLP (which saves major energy resources in model development) to NLP for social media. Overall a good read.


[1]. Jin, Z., Chauhan, G., Tse, B., Sachan, M., & Mihalcea, R. (2021). How Good Is NLP? A Sober Look at NLP Tasks through the Lens of Social Impact. arXiv preprint arXiv:2106.02359.

Published by Nidhika

Hi, I have an excellent problems solving skills in the domain of Engineering, Sciences, and Modelling Solutions for any topic. I gained my strong problem-solving skills with the problems I solved in elite schools with top grades, and hence can now think in direction of problem solving as a generic thing. Apart from profession, I have inherent interest in writing especially about Global Issues of Concern, poems, stories, doing painting, cooking, photography, music to mention a few! And most important on my website you can find my suggestions to latest problems, views and ideas, my poems, stories, novels, my comments, my proposals for people with funding's to implement, my blogs, my interests, my personal experiences and glimpses of my research work as well.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: