Creating Smarter Online Communities with NLP and Network Analytics

Aadit Barua
9 min readAug 16, 2020
Photo by Marvin Meyer on Unsplash

Aadit Barua, Westlake High School, Austin, TX.

A valuable resource, but too much data overload?

Online communities are platforms for in-depth discussions and interactions between people with shared interests. From car and hi-fi enthusiasts to patients suffering from a chronic disease, members in these online communities have a variety of objectives, including obtaining advice from experts, understanding the perception of the crowd regarding a product or service, and seeking support from others in dealing with a problem. Starting with electronic bulletin boards in the 1980s, online communities have been around for well over 35 years, though they include a wide variety of platforms today, including online forums, social networking sites, and chat rooms. While Web 2.0 transformed the real time nature of online interactions, very little has fundamentally changed in the features of these platforms over the decades. The successful ones have tens or hundreds of thousands of users and even millions of posts, with some providing basic keyword search, and a set of frequently asked questions. The challenge for a user is to find relevant information and insights without spending an excessive amount of time and effort, or to easily connect with users who share strong similarities instead of looking through thousands of user profiles.

In this article I show how online community platform operators can increase the value of time spent by users by going beyond displaying posts from others. They can provide deep insights from the wisdom of the crowd, detect sub-communities of interest, and identify community members who are similar in terms of the topics they post on. We can add these time saving and relevance enhancing features without asking members a single question, relying instead on NLP techniques like word associations, topic modeling and similarity analysis, and network analytics in the form of community detection algorithms. Figure 1 shows my framework for adding smart functionality to online communities.

Figure 1: A framework for creating smarter online communities
Aadit Barua

Freshman, UT Austin. Interested in ML, NLP & network analytics applications. Writer for CodeX, TheStartup, TowardsDataScience & TowardsAI.