Analytics are no longer just a matter of “crunching the numbers” in a database or spreadsheet. In fact, by some estimates nearly 80% of all enterprise information is unstructured data, such as documents, web content, or social media. But ordinary analytics techniques have trouble extracting meaning from the complexity and nuances of human language. That’s the purview of specialized AI algorithms that can mimic human insight—but generally require specialized data science expertise.
That’s why OpenText Release 16 Enhancement Pack 5 (EP5) spotlights text mining and other language-related enhancements in the latest upgrade to OpenText Magellan, our AI-enhanced analytics platform.
As part of EP5, OpenText has given Magellan a series of improvements to help it process the fine nuances of human language more adeptly, support the processing demands of Big Data, and make AI easier for non-experts to use. This helps enterprises leverage and analyze all their terabytes of data, both structured and unstructured, deriving more value from Big Data and Big Content.
We’ve enhanced Magellan in four major areas:
- An improved user interface for Magellan Text Mining content crawlers, along with an enhanced dashboard and smart UI.
- Leveraging the proven distributed data storage and processing technologies of Apache Spark, Hadoop, and other open source Big Data tools
- Wider array of languages and text processing features
- Deeper integration into other OpenText solutions – including the Capture and Workforce Engagement Management portfolios.
“User experience is really key,” said Martin Brousseau, Senior Technical Product Manager in charge of the Magellan Text Mining component. “Our overall objective, from Release 16.4 (EP4) of Magellan through EP5 and future releases, is to demystify text mining and analytics. We’re continually making these capacities more accessible to non-technical, business users.”
A better text mining user experience
The first change customers will notice in Magellan is the handsomely streamlined Smart UI user interface, which has been revised throughout with a tile-based dashboard design for a more immediate, intuitive experience.
The colorful Launchpad improves the content mining experience with enhanced views, such as an aggregation overview of all current and recent crawling projects. We’ve unified the interface for Magellan Text Mining web and social media crawlers with this new dashboard for greater convenience.
Under the hood, the updated Magellan Launchpad improves integration of its various components by adding a consistent SSO (Single Sign-On), while increasing security with a centralized user repository and the option of limiting user access to the different component UIs.
More flexible open source deployment options
The ease with which Magellan meets today’s computation challenges of processing very large amounts of data is due in part to the open-source Hadoop Big Data distributed data infrastructure; an instance of the Hortonworks Hadoop distribution is built in. However, we realize many customers have a storage cluster from other vendors already in place. That’s why in Release 16 EP5 of Magellan, we’ve added certification for the MapR platform (with more platforms contemplated for future releases).
Meanwhile, we’ve upgraded our Apache Spark data processing repository to its latest version. This improves performance and provides new transformers, as well as machine learning algorithms such as robust linear regression with Huber loss. The Spark upgrade also let us add new visual prescriptive analytics capabilities, including Pareto analysis.
Come si dice “dati non strutturati”?
Our natural language processing (NLP) researchers have been busy adding in-depth support in Magellan Text Mining for two new languages: Dutch and Italian. You can now extract complex concepts (noun phrases) from unstructured content in Dutch and Italian, joining English, Spanish, French, German and Portuguese. (Magellan already supports basic content analytics in more than 30 other global languages.)
We’ve also “taught” Magellan Text Mining how to recognize when digits or words describe various forms of ID, including driver’s licenses, US Social Security Numbers, passport, credit card, and bank account numbers, addresses, and phone numbers.
Expanding Magellan across the range of OpenText solutions
Magellan Text Mining is now integrated into the OpenText Workforce Engagement Management portfolio solutions OpenText™ Qfiniti and OpenText™ Explore. These solutions can more accurately provide inline summarization when transcribing speech (such as recordings of customer service calls) into text, as well as analyze callers’ sentiment. Magellan has also been integrated into the new reporting and analytics engine for OpenText™ Captiva Data Capture.
Reporting capabilities of Magellan have now been made available in the cloud as a service, so they are even easier to use and benefit from across organizations and product lines.
“OpenText is committed to leveraging the ubiquity of text mining throughout modern EIM solutions, helping customers unlock all the value within their stores of unstructured data,” Brousseau said.
OpenText Magellan uses industry-leading AI, data analytics, and machine learning to speed up, enhance, and automate business processes. You can learn more about Magellan by visiting our website, and you can learn more about the EP5 releases throughout the OpenText family of solutions here.