Our customer, IBM received an order from one of the private US research universities for the development of an interactive informational panel, which tracks and analyzes topics discussed by students on social media.
The university administration requested that data from social media be collected and analyzed, with a view to identifying the common meeting places of university societies on campus, as well as preventing ahead of time the occurrence of unauthorized mass meetings. The developed web app analyzes posts and chats from open student Facebook, Instagram and Twitter groups, as well as anonymous mobile social media like Yik Yak.
The main objective of the app is to register posts that directly or indirectly mention alcohol, drugs, violence and depression, etc. This is collected when key words and the main topic of the message are analyzed; this analysis relies on categories that these identifiers fall under, such as mental health, alcohol, unacceptable sexual behavior, or a neutral group. As soon as the mentioning of certain key words reaches a frequent and critical level, the host is notified.
Data from social media is collected from open sources such as forums used by student groups and communities, as well as from anonymous yet publicly visible messages like Yik Yak, which are aggregated by connecting to the social media networking API’s. This data is then uploaded to one of the IBM Watson services for further analysis of text and graphic content. Not only does the developed web application reflect the main concept of the message, but it also displays the individual’s intrinsic personality characteristics, depending on the tonality of the messages. Graphs are constructed based on the obtained and analyzed data, , allowing information to be clearly visualized and enabling the scheduler to change dynamically.
Integration with social networking APIs
Social media focused data aggregation is performed by integrating with the social media networking API’s
Subject analysis of the aggregated content
Depending on the type of content, the collected data is uploaded to one of the IBM Watson services
Social media word cloud analysis
Real-time social media text clouds analysis and tag dynamics change scheduler
Email alerts and notifications
Sending keyword peak email notifications and/or alerts to admins and moderators
Messages from senders are registered only if they are physically on campus