Unstructured data is notoriously tricky to analyze using traditional methods, so it’s tough to suss out trends from “messy” sources like news articles and social media. That’s quickly changing, however, and on Monday, OpenText launched a timely example that’s focused on the 2016 U.S. presidential elections.
Election Tracker ’16 is an interactive online tool that lets anyone visually monitor, compare and discover new insights about election coverage from 48 major online news sources.
The tool automatically scans hundreds of online media publications from around the globe for sentiment and trends, generating summaries of stories about the race. It organizes them by categories including news agency, candidate, topic and tone.
Election Tracker translates its findings into visual summaries and delivers them in an online app with interactive dashboards and reports. Users can compare coverage of candidates based on categories like topic, date, geography, and sentiment.
OpenText Release 16 is the software driving the process. More specifically, OpenText InfoFusion automatically crawls the Web for election-focused articles and retrieves the raw text for evaluation. OpenText Content Analytics processes that text to determine sentiment and extracts people, places and topics following standard or custom taxonomies, providing the metadata necessary for analysis.
Visual summaries are designed using OpenText Analytics Designer, then housed and deployed on OpenText iHub, where they’re embedded into the app’s user interface with the iHub JavaScript application programming interface.
Mostly, it’s a timely demo of OpenText’s technology. But it’s also a vivid illustration of the technological muscle that’s increasingly available for tackling unstructured data. YouEye, Tamr and Taste Analytics are among the numerous young companies that have emerged recently to meet that challenge.