Big Data in K-12 (Part III): Getting From Sunrise to Noon

Productizing Data Science and Voice Recognition

Previous columns—“Big Data in K-12: Attack of the Recommendation Engines” and “Big Data in K-12: Is Voice Technology Talking to You Yet?”—have explored the big ideas and market drivers behind the surge of recent entrepreneurial initiatives seeing big data technology and, to a lesser extent, voice recognition technology, as vehicles for markedly enhanced intelligent educational products. The articles triggered reader feedback with interesting examples of innovative products and services, several of which are discussed here in this concluding piece that also pulls together the hurdles and business implications for education resource providers. Language acquisition products are the low-hanging fruit, but a growing community of entrepreneurs sees a host of other K-12 applications on the horizon.

(These articles are based in part on my recent “View From the Catbird Seatpresentation at EdNET 2012.)

Coming Out of the Woodwork

Illustrative of the overlap of big data and voice recognition technology (VRT), Yosi Glick, Co-Founder & CEO, Jinni, developers of a “Pandora-like” taste-and-mood-based recommendation engine for movies and TV shows, wrote after reading the previous article, “Jinni has developed the Entertainment Genome™, over 2,000 descriptive semantic tags to power discovery for entertainment… [and] we are beginning to develop the world’s first ‘search as you speak’ natural language understanding (NLU) video discovery guide.”

Also in response to a previous article, Christen Graham, Giving Strong, Inc., wrote to me on behalf of Betsy Peters, President, PossibilityU, about the start-up’s Admission Toolkit, which uses a recommendation engine to “help students find colleges and universities where they will thrive by combining student data with data reported by 4,000+ schools in the U.S. to increase the likelihood of admission, aid, and completion with a degree in four years or less.” Explaining how it works, she added, “The student begins by entering schools of interest. The recommendation engine then suggests more colleges to investigate based on common academic and social characteristics culled from 78 different ‘fit and feel’ criteria, as gleaned from government reports and other proprietary sources. Next, after personalizing the search with the student’s grades and test scores, the program sorts and plots the chances of the student getting in and getting aid from this set of schools. Unlike more conventional college search engines and guidebooks, PossibilityU quickly, simply, and accurately personalizes the search—improving each student’s chances of finding the college that fits. The technology is akin to how Pandora ‘learns’ from one’s listening history to create a custom channel for an individual. From there, students get a ‘heat map’ visualization of their chances not only of getting in but also of getting aid—to help prioritize which schools they should apply to.”

Glen Fisher, Chief Operating Officer, RedRock Reports, has a more skeptical view of the innovation attributed to recommendation engines. Responding to the previous articles in this series, he wrote, “While Grockit is closer, a lot of stuff today isn’t really far advanced from Plato—‘atomistically’ walking students through an adaptively provided sequence of skill learning. This is the perfection of industrial education—if you will, the personalized auto production line. What’s missing is what we know about how people learn (which Grockit has one piece of, as does Kaplan—the group interaction). But the idea of project-based learning (what most adults and preschool children do) is totally out the window in this scheme. I’m really excited about having usable data available in school—when I taught, I got my students’ test results after they had already graduated, and my school didn't bother to share previous years’ testing data since it was printed and there was no easy way to rearrange it by classroom. I think that having useful information available to both student and teacher (learning guide?) is a huge step forward and really can change learning. But I’m concerned that we still seem to have the view of students sitting in rows—now at computers—all ‘doing’ the same curriculum, albeit in different, personalized, adaptive order.”

Hurdles and Implications

Some of the challenges that remain before big data and VRT can have a significant impact on K-12 include:

  1. Data accuracy, integration: “Garbage in, garbage out” still applies. Until educational data are well-defined and of high quality, analyses using the data will be seriously compromised. Further, data interoperability standards and usage need to advance to the point that K-12 data systems will be able to talk intelligently to one another.
  2. Analytic models: The power of big data predictive and adaptive learning systems depends on their underlying mathematical models, concept maps, and embedded application-specific logical structures, the so-called “ontology” mentioned earlier in this series. These tools are still in their infancy and, for example, for Knewton-like products, are very labor-intensive to create for any given course. Their validity and power have yet to be definitively proven. Even those thought of as “model free,” based on multivariate statistics intended to capitalize on the relative success of individual users with alternate instructional branches, require the track records of many thousands of users to begin to understand how best to match instruction to the individual.
  3. Voice recognition’s still immature: A recent eWeek article by Robert Mullens,Speech Recognition Finally Finding Its Voice in Mobile Technology,” reported, “If speech-recognition technology were a human, it would be like a five- or six-year-old child. At the age of one, you can speak to a child, but you have to speak slowly and simply using small words. By five or six, it starts to better understand your words and, more importantly, your meaning.” For the near term, VRT will be very application- and user demographic-specific but that hardly precludes robust K-12 niche products for many disciplines.
  4. Social, ethical, legal, privacy, and ownership: It goes without saying that the kinds of privacy, data residency, and legal and ethical concerns that are being debated in the consumer arena, such as for tracking web use, and that have been an issue for some school authorities considering SaaS resources instead of campus-based will need resolution for using big data tools in K-12.
  5. Changing power structures in schools and companies: Like other major technological changes, widespread use of big data resources in K-12 will have organizational impacts for schools and the companies that offer the services. For example, it will further embed curricular authorities in the IT decision-making structure, raise the need for professional development, and resonate with the virtualization of suppliers. Change management skills will be essential in both arenas.
  6. Classroom bandwidth, device penetration, etc.: The usual list of suspects for hindering technological innovation in schools will still apply.

While the implications of big data and VRT science for K-12 resource providers are still far from clear, some likely ones include:

  1. Data science on your team: A common thread among K-12 entrepreneurial big data and VRT initiatives is the staff presence of veterans of commercial recommendation engine and/or VRT businesses (e.g., the data architect with five years at X and the engine scientist from Y who used to throw out data sets with less than 10 million user responses). Much like the advent of K-12 digital products and services brought IT and web skills into the personnel mix for most firms, you’ll see data scientists increasingly occupying a slot on your roster.
  2. Data interoperability and portability: As mentioned in an earlier column, to work in tomorrow’s data-rich digital environment, products will need to “play nicely” with others, putting a premium on data interchange protocols, such as IMS, SIFA, and SCORM, and content tagging schemes, such as the Learning Registry (LR), Shared Learning Infrastructure (SLI), and Learning Resource Metadata Initiative (LRMI). For a nice overview of what these mean to K-12 resource providers, see “How Will Student Data Be Used?—a recent Mind/Shift article by Frank Catalano, Principal, Intrinsic Strategy, and EdNET InsightIndustry Analyst for MDR. (Another of his overviews clarifying this alphabet soup of programs is here.) A related initiative, the MyData Button, a joint project of the U.S. DOE’s Office of Educational Technology (OET) and the White House Office of Science and Technology Policy (OSTP), aims to encourage schools and software vendors that hold student data to allow students to download their data to create personal learning profiles that they can keep with them throughout their learning career.
  3. Data mining for product development and marketing too: While we’ve concentrated here on technology supportive of adaptive real-time instruction, recommendation engines and VRT will be finding a wide range of other K-12 uses in product development and marketing. MDR’s new Targeted Educator Web Advertising product is an example of the latter.
  4. Platforms and licensed technology: As with other sophisticated digital and web tools, most K-12 resource providers will find it more cost-efficient to license their big data technology, for example, from Junyo (talk to Wayne Poncia, VP Business Development) and VRT, perhaps from Nuance Communications, than to develop and keep them current.
  5. Measured product and company efficacy: On the far distant horizon, when recommendation engines become much more mature and support real-time formative assessment, it’s conceivable that practical measures of product and even company efficacy may become a reality. Way before that time, you can count on marketers flaunting their products’ impact on student competencies.

The jury is still out on how far big data science and voice recognition technology can move the efficacy needle for online K-12 products. Given their growing impact in commercial, political, and military applications, it’s a safe bet that this isn’t an “if” question but a “when.” As 1:1 proliferates, and particularly with mobile technology delivering more sophisticated cloud-based services, it’s likely we’ll see use not only in adaptive instruction but also in assessment, product development, and marketing. Now’s not a bad time to start thinking about how they might be useful for you.

(Note: For more on big data’s implications for K-12, check out AEP’s recent “Big Data Leadership Day - How to Go Boldly into the New Frontier,” CEO Roundtable held in New York City, Wednesday, November 28.)


Dr. Nelson Heller is President of The HellerResults Group, a global strategic consultancy serving business and non-profits seeking growth opportunities in the education market. He is the founder of The Heller Reports newsletters and EdNET: The Educational Networking Conference, both started in 1989. The EdNET News Alert, successor to The Heller Reports publications and now published by MDR, reaches over 31,000 education executives worldwide every week and features a regular column from The HellerResults Group each month. You can learn more about Nelson and his industry leadership at The HellerResults Group. If you need strategic insight, business partners, international connections, stronger boards, keynoters, or entrepreneurial savvy and want the insight of 30 years at the business and technology crossroads of the education market, you can reach Nelson at 858-720-1914, by email at nelson@hellerresults.com, and on Twitter @NelsonHeller.