Big Data Bootcamp: What Managers Need to Know to Profit from the Big Data Revolution (2014)
Chapter 11. Big Data Opportunities in Education
The Rise of Big Data Learning Analytics
Netflix can predict what movie you should watch next and Amazon can tell what book you’ll want to buy. With Big Data learning analytics, new online education platforms can predict which learning modules students will respond better to and help get students back on track before they drop out.1 That’s important given that the United States has the highest college dropout rate of any OECD (Organisation for Economic Co-operation and Development) country, with just 46% of college entrants completing their degree programs.2,3 In 2012, the United States ranked 17th in reading, 20th in science, and 27th in math in a study of 34 OECD countries.4 The country’s rankings have declined relative to previous years.
Many students cite the high cost of education as the reason they drop out. At private for-profit schools, 78% of attendees fail to graduate after six years compared with a dropout rate of 45% for students in public colleges, according to a study by the Pew Research Center.5
Among 18 to 34 year olds without a college degree, 48% of those surveyed said they simply couldn’t afford to go to college. Yet 86% of college graduates say that college was a good investment for them personally.
The data tells us that staying in school matters. But it also tells us that finishing school is hard. Paul Bambrick-Santoyo, Managing Director of Uncommon Schools, Newark and author of Driven By Data: A Practical Guide to Improve Instruction, has shown that taking a data-driven approachdoes make a difference.
During the eight years in which Bambrick-Santoyo has been involved with the Uncommon Schools, which consist of seven charter schools focused on helping students prepare for and graduate from college, the schools have seen significant gains in student achievement, reaching 90% proficiency levels on state assessments in many categories and grade levels.6
Using a data-driven approach can help us teach more effectively. At the same time, technology that leverages data can help students with day-to-day learning and staying in school. Netflix and Amazon present us with offerings we’re more likely to buy, delivering a more personalized and targeted experience. Pandora figures out our music tastes and recommends new music to listen to. In the future, this kind of personalized experience won’t just be used just for entertainment and shopping, but for education as well.
Note Just as Pandora understands our musical tastes, tomorrow’s education companies—built on Big Data analytics—will tailor custom educational experiences to specific students and their needs.
Adaptive Learning Systems
How can computers help students learn more effectively? Online learning systems can evaluate past student behavior, both for individuals and in aggregate, to predict future student behavior. Within a given course or courseware framework, an adaptive learning system can decide what material to show a student next or determine which areas a student might not yet fully understand. It can also show students, visually, how they are progressing with certain material, and how much material they’ve absorbed.
One of the strengths of an adaptive learning system is the built-in feedback loop it contains. Based on student interactions and performance, an adaptive learning system provides feedback to students, feedback to teachers, and feedback to the system itself, which humans or the system can then use to optimize the prediction algorithms used to help the students in the future. As a result, students, teachers, and educational software systems have a lot more visibility into what’s going on.
Software can also predict which students are likely to need help in a given course.7 Online courseware can evaluate factors such as login frequency and timeliness of turning in homework to predict whether students will pass or fail. Such software can then alert course instructors, who can reach out to students in danger of failing and offer them extra help or encouragement.
Knewton is one of the most well-known adaptive learning systems. Founded by a former executive of test prep company Kaplan, Knewton’s system identifies strengths and weaknesses of individual students. The company started out by offering GMAT test prep, but now universities are using it to improve education.
Arizona State University (ASU), the country’s largest public university with some 72,000 students, uses the Knewton system to improve students’ proficiency with math. After using the system for two semesters with 2,000 students, ASU saw withdrawal rates drop by 56% and pass rates improve from 64% to 75%. The company has raised $105 million in venture capital and the World Economic Forum named Knewton a Technology Pioneer.
DreamBox, another provider of adaptive learning systems, is trying to improve math performance at the elementary school level. The company has raised $35.6 million in funding from well-known investors, including John Doerr of venture capital firm Kleiner Perkins, and Reed Hastings, CEO of Netflix.8 DreamBox offers more than 1,300 lessons that help boost math proficiency. The company’s applications are available for both desktop computers and Apple iPads.
At a broader level, data mining can also recommend courses to students and determine whether college students are off track in their selected major. ASU uses an eAdvisor system to coach students through college. The university’s retention rate has risen from 77 to 84%, a change that the provost, Elizabeth Capaldi, attributes to eAdvisor.9
The eAdvisor system tracks whether students fail key courses or don’t sign up for them to begin with. It then flags such students so that academic advisors know to talk with them about their progress and recommend new majors, if necessary.
Such systems have vast amounts of data available to them, from individual course performance to standardized test scores to high school grades. They can compare data about any one student to data gathered about thousands of other students to make course suggestions.
This increased level of transparency extends all the way from students to teachers and administrators. Students get more information about their own progress. Teachers get more visibility into individual student progress as well as overall class progress, and administrators can look across all classes at a school to see what’s working and what isn’t. District administrators can then draw conclusions about what kinds of educational programs, software, and approaches are most useful and adjust curriculums accordingly.
Putting Education Online
One of the most interesting uses of Big Data as applied to education is the ability of adaptive learning systems to test many different educational approaches across a large number of students. Web sites use A/B testing to show one version of a web page to one visitor and another version to another visitor. Learning systems can do the same thing.
A learning system can evaluate whether students learn faster when they receive a lot of practice on a given type of problem all at once or when that practice is spread out over time. Learning systems can also determine how much material students retain after a given period of time and tie that back to the learning approaches used.
The intersection of Big Data and education doesn’t stop at understanding how students learn. New startups are bringing educational materials online and opening them up to a much larger audience in the process.
Note While Big Data educational startup opportunities rest on a foundation of analytics and insight into learning models, the most effective succeed through smart marketing that brings in masses of students and sustains interest.
Khan Academy is an online destination containing thousands of educational videos. Salman Khan, the founder of the site, originally started recording the videos himself and the site now has more than 6,000 lectures available on a variety of topics, including history, medicine, finance, biology, computer science, and more.10 The site’s videos are stored on YouTube and in aggregate have received more than 440 million views.
The site’s approach is simple yet effective. In addition to thousands of short videos, which highlight the material being taught rather than the person teaching it, the site uses hundreds of exercises to help teach concepts and evaluate the level of each student’s comprehension.
It’s a long way from there to being able to build and make money from your own iPhone app, but the site provides a great way to get started. Imagine using courses from either site to train data scientists or educate business users on how to use analytics software.
Major universities are also putting their courses online. Harvard University and MIT teamed up to form edX, a not-for-profit enterprise that features learning designed specifically for study over the web. The site’s motto is “The Future of Online Education: for anyone, anywhere, anytime.” A number of major universities now participate in the program. Along with MIT and Harvard, Columbia University, Cornell University, the University of California at Berkeley, The University of Texas, Wellesley, Georgetown University, and many others, are also participants.
University faculty members teach the classes, which typically consist of short lecture videos accompanies by assignments, quizzes, and exams. In addition to enabling these universities to deliver course material electronically, edX provides a platform for learning about how students learn. EdX can analyze student behavior to determine which courses are most popular and which result in the greatest learning. EdX has said it wants to teach a billion students, and the MIT Technology Review called offerings like edX the most important educational technology in the last 200 years.11
How Big Data Changes the Economics of Education
As the Technology Review points out, online learning isn’t new. Some 700,000 students in the United States already use distance learning programs. What’s different is the scale at which new offerings operate, the technology used to deliver those offerings, and the low-cost or free delivery models.
As in other areas of Big Data, what’s changed is not that Big Data never existed before, but the scale and cost at which it can be accessed. The power of Big Data is its ability not just to gather and analyze more data, but to open access to that data to a much larger number of people and at a much lower cost. Free and low-cost education offerings such as those from edX are called massive open online courses, or MOOCs for short.
Note While MOOCs have received a lot of press coverage, the completion rates for courses are very low. As a result, the design of these courses needs to be revisited and I predict that a next generation of MOOCs—data-aware MOOCs—will emerge in the years ahead.
In 2002, only about 9.6% of college students were enrolled in at least one online course. By 2013, 33% of students were, according to studies by Babson Survey Research Group. That means some 7.1 million college students are taking at least one course online every year.12
Another offering, Coursera, was started by computer science professors at Stanford University. The company originally launched with Stanford, Princeton, the University of Michigan, and the University of Pennsylvania. Now it offers courses from more than 80 universities and organizations. It has received some 22.2 million enrollments for its 571 courses, with students from 190 different countries.13 Notably, Data Analysis is one of the top courses on the site, highlighting just how much interest there is in Data analytics.
Meanwhile, Udemy, which has the tagline “the academy of you,” brings together online instruction from a range of CEOs, best-selling authors, and Ivy League professors. The site takes a somewhat less academic approach to its offering and many of its courses are about practical business issues, such as raising venture capital. Unlike some of the other sites, Udemy allows course creators to provide their courses for free or charge for them.
Udacity, founded by Google vice president and part-time Stanford University professor Sebastian Thrun, has the goal of democratizing education. The company’s initial courses have focused primarily on computer science related topics, but it is continuing to expand its offerings. To date, the company has raised $48 million in funding with the goal of expanding its course marketplace into lots of markets outside the United States.
As existing academic institutions search for ways to remain relevant in an online world, it is clear that the proliferation of such digital offerings will offer insight into the most effective ways to deliver educational content. While you may not yet be able to get a degree from Harvard, MIT, or Stanford over the web, getting access to their materials as well as to materials from anyone with something to teach is becoming a lot easier. At the same time, courses from leading academics on Big Data-related topics are now available to everyone—not just students at the universities at which they teach.
Of course, online courses can’t provide the same kind of social or physical experience that classrooms or laboratories can provide. Courses in biology, chemistry, and medicine require hands-on environments. And just as social encouragement and validation is important when it comes to exercising and dieting, it may also be important when it comes to learning. The most promising educational systems of the future may be those that combine the best of the online and offline worlds.
Virtual environments may also provide a way to bring offline experiences online. The Virtual Medical Environments Laboratory of the Uniformed Services University adapts leading edge technology to provide medical training through simulation. Such simulation environments take advantage not only of software but also of hardware that simulates actual medical procedures. These environments can also simulate the noise or distractions that medical personnel may experience in the real world.
Using Big Data to Track Performance
The U.S. government, across federal, state, and local governments spends about $820 billion per year on education. That doesn’t count all of the investment made at private institutions. But it does mean that administrators want visibility into how school systems are performing, and new systems are providing that visibility, according to Darrel M. West of the Brookings Institution.14
DreamBox, the adaptive learning systems provider, also provides visibility for administrators. In addition to delivering adaptive learning tools, it has a dashboard capability that aggregates data for administrators to view. Administrators can track student progress and see the percentage of students who have achieved proficiency.
At a government level, the U.S. Department of Education has created a dashboard that summarizes public school performance for the entire country. The interactive dashboard is available on the web at dashboard.ed.gov/dashboard.aspx.
States use a variety of systems to report on educational progress. The state of Michigan provides a dashboard at www.michigan.gov/midashboard that indicates whether performance is improving, declining, or staying the same in areas such as third grade reading proficiency, college readiness, and academic proficiency between grades three and eight.
The state’s third grade reading proficiency, for example, improved from 63.1% of students during the 2007–2008 school year to 70.0% of students during the 2013–2014 school year. According to the site, this measure is a strong indicator of future academic success.
Such systems improve accountability and provide more visibility into educational performance, according to West. Much of the information that goes into dashboards like these already exists, but web-based systems that have simple user interfaces and easy-to-view graphics are a big step forward in making such data accessible and actionable.
Note A big part of the opportunity in education for entrepreneurs lies not just in accessing data that was difficult to access in the past, but also in the product developer’s ability to interpret the data for end-users and to show it in visually powerful ways.
Data mining, data analytics, adaptive learning solutions, and web dashboards all present opportunities to improve education and increase access to it. But one of the biggest challenges, states West, is the focus on “education inputs, not outputs.” Quite frequently, schools are measured on seat-time, faculty-student ratios, library size, and dollars spent rather than on results. “Educational institutions should be judged not just on what resources are available, but whether they do a good job delivering an effective education,” says West.
With that in mind, it’s clear that the approach taken by today’s MOOCs needs to be revisited. MOOC completion rates are abysmally low, with just 5% of edX registrants earning a certificate of completion.15 The typical long-form course content is similar to traditional university lectures, but without the social nature of a local community of fellow students to help everyone stay engaged. But what looks like failure in one context is often opportunity in another. In this case, the failure of the first generation of MOOCs presents an opportunity for entrepreneurs to create a next generation of MOOCs that are more engaging and result in higher completion rates.
In particular, data-aware MOOCs could automatically detect if students are becoming disengaged and present different content or interactive quiz modules to keep students engaged. Data-aware MOOCs could predict, based on the behavior of past students, which students are likely to drop out and alert advisors, who can then communicate with the students.
Data-aware MOOCs could even use data to determine which kinds and pieces of content, such as videos and interactive modules, are most effective. Based on this data, educational content creators could improve their content, while savvy application developers would create interactive MOOCs that use data to create learning interfaces that are as engaging, exciting, and social as today’s most popular video games.
Education faces many of the same challenges when it comes to Big Data as other areas. Incompatible technology systems make it hard for schools to aggregate data within schools let alone compare data across different academic institutions. For example, some schools use separate systems for tracking academic performance and attendance. Transforming complex data sets about educational performance into key metrics is critical to making such data actionable.
How Big Data Deciphers What We Learn
As discussed in the Introduction, data not only makes computers smarter, it also makes human beings smarter. But the biggest question of all when it comes to education and Big Data may be the fundamental question about education itself: how do we learn?
Different people learn in different ways. Some students do better with visual learning while others do better with hands-on studies or when they write things down. Psychologists spent much of the last century constructing theories about how we learn, but they made little actual progress.
About ten years ago, scientists started taking a different approach. They used neuroscience and cognitive psychology to study how the brain learns.16 What they discovered was that our ability to learn is shaped in large part not by what is taught but by the effectiveness of the learning process. A more efficient learning process can result in more effective learning.
Learning consultant Clive Shepherd captured some of the key insights from a talk by Dr. Itiel Dror of Southampton University on the science of learning.17 While pop psychology has it that we only use five to ten percent of our brains, in reality, we use the entire capacity.
One of the keys to understanding how we learn is to recognize that our brains have limited resources for processing the huge amount of data we receive through our different senses. Our brains rely on all kinds of shortcuts to avoid getting swamped—what is known as cognitive overload.
As a result, to make learning more efficient, teachers can provide less information or take a very careful approach to how they communicate information. People have an easier time absorbing information if there’s less noise that goes with it. But less noise also means less context.
One of the shortcuts the brain uses is to group things together. Teachers can make learning more efficient by grouping material so that students don’t have to. Another approach to reducing cognitive overload is to remove every word or picture that isn’t necessary to a particular learning goal. Challenging the brain helps with learning; researchers found that students learn more when they try to read a book for the first time than when they try to read the same book again.
Of course, all that still doesn’t answer exactly how we learn. To cope with the vast amount of information it receives, the brain does a lot of filtering. The brain has evolved over many years, and one of the first things it needed to do was deal with basic survival.
Our ancestors were a lot more likely to survive if they could remember dangerous situations, such as stalking the wrong prey, and avoid them in the future. Such situations were often associated with moments of high emotion. As a result, it is easier for us to remember information that’s associated with high emotion, whether it is positive or negative.18
Past experiences also influence how we retain information. Scientists believe that our brains store information in a sort of filing cabinet-like approach. This is one of the reasons it’s easier to add more information to an existing area we know—an existing base of learning—than to learn something new from scratch.
Using Big Data to Speed Language Acquisition and Math Skill
According to educational consultant Dr. David Sousa, citing Dr. Keith Devlin at Stanford University, mathematics is the study of patterns.19 Sousa argues that all too often math is taught as simply a series of numbers and symbols, without any discussion of how it applies to daily life. Since meaning is one of the criteria the brain uses to identify whether information should be stored long term, math students may struggle to understand the subject without meaning to give it context.
Devlin highlights a number of cases where math applies to real life. Using probability to determine odds, calculating the amount of interest you pay when you buy a car, and applying exponential growth curves to understand population changes are three such examples.
But math may have as much to do with the language we use to represent our numbers as with how we learn it. As Malcolm Gladwell talks about in Outliers, referencing Stanislas Dehaene’s book, The Number Sense, the English numbering system is highly irregular. Unlike English numbers, which use words like eleven, twelve, and thirteen, Chinese, Korean, and Japanese numbers use a more logical and consistent approach: ten-one for eleven, ten-two for twelve, and so on.
As a result, Asian children learn to count a lot faster. By four, Chinese children can count up to forty while American children of the same age can only count up to fifteen. They only learn to count to forty when they’re a year older, putting them a year behind their Chinese counterparts. Gladwell cites another example: fractions. In Chinese three-fifths is literally, “out of five parts, take three,” which makes such quantities much easier to work with. The language matches the concept.
The implications don’t stop there. The brain has a working memory loop that can store about two seconds of information at a time. Chinese numbers can, in general, be pronounced in a shorter span of time than their English counterparts, which means that if you think about math in Chinese, you can remember more numbers at a time.20
More math education is highly correlated with higher earnings. In a study by the Public Policy Institute of California, authors Heather Rose and Julian R. Betts found that students who had completed calculus courses had higher earnings than those who had only completed advanced algebra.21 They in turn had higher earnings than people who had completed only basic algebra.
Higher-level math education is also associated with higher college graduation rates. As the authors point out, correlation is not the same as causation, but they conclude that math education is highly associated with both earnings and college graduation rates.
If there’s one person who knows more about learning math than just about anyone else, it’s Arthur Benjamin, Professor of Mathematics at Harvey Mudd College. Benjamin is best known for his ability to perform mathemagics, in which he multiplies large numbers together in his head and produces the correct result.
As Benjamin shows, math doesn’t need to be boring. It can be fun and entertaining. Proving the point, his TED talk on mathemagics has received more than four million views. Benjamin has also authored a book, Secrets of Mental Math: The Mathemagician’s Guide to Lighting Calculation and Amazing Math Tricks as well as a DVD entitled The Joy of Mathematics. In his book, Benjamin shares a number of shortcuts to doing complex math in your head. When it comes to numbers, Big Data can be fun if given the right context!
So what about language? According to research by Dr. Patricia K. Kuhl at the Center for Mind, Brain, and Learning at the University of Washington, as infants, we store a lot of information about speech and language before we begin speaking. Simply listening to sounds helps our brains understand one language better than another.
Earlier, we talked about how the brain filters the vast amounts of information to which it is exposed. The infant brain does much the same thing with language. As infants master the language spoken by their caretakers, they ignore sound differences that aren’t relevant.
For example, the different sounds for “r” and “l” are important in English (for words like “rake” and “lake”), but they aren’t important in Japanese. Japanese babies tested at the age of six months could tell the difference between the two sounds equally as well as their American counterparts. By the age of 10 to 12 months, however, infants in the United States improved in their ability to tell the difference between the two sounds, while their Japanese counterparts got worse.
Kuhl attributes such changes to the infant brain focusing on the sounds it hears, the sounds of the infant’s native language. During this period of rapid learning, it is also possible to reverse such declines by exposing infants to multiple languages. In one study, Kuhl had Chinese graduates students talk in Chinese with American infants. After 12 laboratory sessions, the American infants were able to recognize Chinese sounds just about as well as their Taiwanese counterparts. Kuhl concluded that the brains of infants encode and remember the patterns they hear well before those infants speak or even understand complete words.
By the age of six months, our infant brains are able to map the patterns of language having to do with vowels and consonants and by nine months, the patterns of words.22 Kuhl describes the infant brain as analogous to a computer without a printer hooked up to it.
When it comes to reading, Kuhl’s studies show that our ability to distinguish speech sounds at the age of six months correlates highly with language abilities like reading later in life. In other words, the better we are at distinguishing the basic building blocks of speech early in life, the better we are at complex language skills later in life.
According to Kuhl, we have about a trillion neurons (nerve cells) in place in our brains when we’re born, but there are relatively few synaptic connections between. From the time we’re born until about three years old, our brains form connections at a furious rate.
By age three, the brain of the average child has nearly twice as many connections as that of an adult. Moreover, the connections create three times more brain activity than in adults. At this point, the brain begins to prune unnecessary connections. Kuhl describes this as “quite literally like a rose bush, pruning some connections helps strengthen others.” The pruning process continues until the end of puberty.
Are you out of luck if you don’t start learning multiple languages at a young age? Common wisdom has it that learning a new language is difficult, if not impossible, after childhood. But one adventurous individual spent more than nine years traveling the world to see if how many new languages he could learn.
Much as mathematician Arthur Benjamin developed a set of shortcuts for doing rapid math calculations, Benny Lewis, author of the blog Fluent in 3 Months: Unconventional Language Hacking Tips From Benny The Irish Polyglot, developed a set of shortcuts for rapidly learning to speak new languages. Lewis, a former electrical engineering student with a self-proclaimed dislike of learning new languages, has shown that learning a new language as an adult is possible, if you take the right approach.
According to Maneesh Sethi, author of Hack the System, most of the challenge in learning a language later in life is that we go about learning it the wrong way. Sethi realized after studying Spanish for four years in high school that according to standardized tests he was an expert in Spanish. But, as he puts it, when it came to actually speaking Spanish, “I couldn’t even order a burrito.”23
Sethi breaks the strategy of rapid language learning down into four steps: having the right resources, which include a grammar book, memorization software, and films/books; getting a private tutor; speaking and thinking only in the new language; and finding friends and language partners to converse with.
Sethi points out that by memorizing 30 words a day, you can learn 80% of the words necessary to communicate in a language in just 90 days. In Russian, for example, the 75 most common words make up 40% of occurrences. The 2,925 most common words make up 80% of occurrences, 75 less than the number of words you’ll know by learning 30 new words per day. Sethi also highlights the importance of having the right mentality. Instead of thinking of himself as a blogger who wanted to learn Italian, he started thinking of himself as an “Italian learner (who blogs in his extra time)”.
Fortunately, modern technology helps in many of the key areas, from memorization to tutoring to finding language partners. On the Mac, the Genius app uses a spaced repetition approach to flashcards that chooses questions intelligently based on your past performance. The more often you make a mistake, the more often the app will test you on a given word. Many online sites now provides live, interactive tutoring sessions via webcam.
Dealing with Data Overload: How Big Data Helps Digest and Filter Information
As children, we learn language by hearing it and speaking it, not by studying textbooks in a classroom. It should be no surprise therefore that the same approach that Kuhl highlights as being critical for children also works for adults: intensive, regular listening to and speaking the language we want to learn.
As Sethi points out, to learn a new language you must be an active learner. “Most people allow themselves to be taught to, but you have to take an active role in asking questions.”
To cope with the vast amount of information it receives, the brain uses pattern matching and other shortcuts to make decisions. In this context, the approaches taken by Lewis, Benjamin, and Sethi make perfect sense. Rather than waiting for the brain to develop new pattern matching approaches and turn those into shortcuts, which is hard work, the key is to teach the brain new shortcuts instead.
“The challenge isn’t in learning a new language, but rather learning how to learn a language,” says Sethi. The same may hold true for other areas of learning.
Tip Product developers can strike gold when they uncover new ways of learning, rather than simply clothing old teaching methods—like rote memorization—in new apps. Each subject has unique content and data aspects you can discover, pursue, and ultimately package into a product that radically improves the learning experience.
Ultimately, learning is about taking in information and storing it, then drawing connections between what is already known and new information.24 When it comes to Big Data, one takeaway from research on the brain and how we absorb information is that more data may give us more insight, but ultimately to be useful it needs to be digested and filtered down to a set of insights that are actionable; insights that can have a direct impact on our decisions.
As applied to education, Big Data is already helping to keep more students in school by figuring out when they’re going to drop out. Adaptive learning solutions, whether in the form of complete systems or simple digital flashcard apps like Genius, are helping us learn more efficiently.
But there are many more possibilities. Much as our brains filter large amounts of information in order to make sense of it, Big Data technologies are also being designed to do the same. As more information is generated—and more of that information is in the form of unstructured data that is hard for traditional approaches like Master Data Management (MDM) to make sense of—new approaches are required.
Semantic search technology attempts to match searcher intent with the contextual meaning of vast amounts of data. Semantic search goes beyond simple word matching and tries to understand both the searcher’s concepts and the concepts represented by documents, web pages, and even images and videos.
For example, if you type the word “flights” into the Google search box, Google uses semantic intent to predict what you’re actually searching for.25 In the case of the word “flights,” Google might combine that search term with knowledge of your current location to suggest flights originating in your city. Such semantic searching techniques can help us tackle the information overload problem that is growing bigger every day.
Related to semantic search, semantic filtering is the application of intent and conceptual understanding algorithms to the filtering of vast quantities of information, such as news sources. Rather than filtering and suggesting relevant content to read or view based on simple word match techniques, semantic filtering analyzes the concepts contained in news articles, combined with knowledge of a user’s preferences or interests, to recommend highly relevant content. The underlying algorithms can become smarter over time about which content to recommend based on feedback from users about which content they want to see more of—and which they don’t.
We are already integrating such techniques into our daily routines through the use of recommendation systems, spam filtering, news filters, and search boxes. There are opportunities to build many more applications that help us filter the data overload into actionable, relevant information.
At the same time, Big Data can also help us with the actual practice of education. To be truly successful, data-driven educational applications will need to take lessons from today’s video games and social networking applications. Facebook is able to drive some 40 minutes of voluntary engagement per day while games like World of Warcraft engage players for hours at a time.
Both of these applications have cracked the code on delivering content that engages—and optimizing the display of that content over time—as well as providing a social experience driven by status and recognition. By combining the right aspects of social, content, and Big Data, application developers can apply the same principles to produce the break-through educational appellations of tomorrow. From enabling more students to graduate to making it possible for us learn more efficiently, Big Data holds the promise of helping us become not only better teachers but better students as well.
2The Organisation for Economic Co-operation and Development is an organization whose mission is to promote policies that will improve economic and social well-being and that counts 34 countries among its members.
9https://asunews.asu.edu/20111012_eAdvisor_expansion and http://www.nytimes.com/2012/07/22/education/edlife/colleges-awakening-to-the-opportunities-of-data-mining.html