Innovation Blueprint Series:  Social Media Monitoring

With 9 patents in Quality Management system software, Sparta has led the industry with innovation in the quality management software space.  This year at Sparta Connection, I spoke about our continued commitment to the advancement of quality management software systems. 

In this first edition of a multi-part series, I’ll explain one of our latest prototypes.

Sparta’s innovation prototypes were built on TrackWise Digital, our validated managed cloud platform. ­

Social Media Monitoring

Many of our customers, especially outside of Pharma, are starting to leverage public data feeds to proactively identify potential product issues.  While there is still some work to be done to leverage this data under the umbrella of protecting your company from compliance risk, we know there is great power in using the data to head off potentially serious issues affecting not only health and safety but brand reputation.

This prototype involved collecting data feeds from Twitter based on certain keywords and sentiment. Specifically, we were looking for negative sentiment related to a particular product.    


The technology used for this prototype included a combination of Node.js, AWS Comprehend and a MongoDB data store.

A monitoring application built via Node.JS is used to intercept tweets based on certain keywords and perform actions based on configurations.  Upon receipt, the application sends the tweet to Amazon Comprend for key phrase and sentiment analysis. The tweet and overall anaylsis is stored in a MongoDB database where future analytics may be performed.

Based on our action configrations, when negative sentiment is detected, we create a simple email alert sent to specific members of the quality organization whose primary role is to investigate complaints and determine next steps and actions to be taken.  Typically, this person is monitoring social media and other forms of analytics to make a decision.


Here is an example of a tweet used in this prototype to drive a negative sentiment alert.  In our case, we created a fictitious company “SneezeAway” (leading manufacturer of allergy medication) with a product by the same name.

In the prototype a customer enters a tweet and the feed is picked up because it contains the keyword “SneezeAway”.  Our filter is focused on receiving tweets that contain the company and/or product name.

The content of the tweet is then sent to Amazon Comprend, a natural language processing service. Via text and phrase analysis, the service determines the emotional sentiment of the tweet; positive, neutral, negative or mixed.  In our case, the above tweet is classified as negative sentiment and therefore triggers the configuration action to send an alert to the interested parties.

In addition to the alert, we also demonstrated the configuration of a trigger that creates a complaint directly in TrackWise Digital by invoking the TrackWise Disgital APIs.  

Try it out…

This prototype was created by Matthew Bambach and Ben Lamm, senior members of the R&D organization at Sparta.  The code used in their prototype is located on Sparta’s Github organization under the Innovation repository.  To view the code and/or try it out for yourself, visit   

Next Steps

Look for the next edition in this multi-part series where I explain our Hands-free innovation that facilitates working closely with your qulatity management system in a cleanroom environment.