Big Data as an Enabler of Primary Education

This is an excerpt of an article that was published on Mary Ann Liebert,.inc publishers online publication on 1 September, 2016

The journey of a thousand miles begins with one step.

-Lao Tzu

Tens of millions of children across the globe lack a basic primary education and are unable to exercise their full potential as citizens and participate as productive members of society. This is arguably one of the worst shortcomings of the human condition. We are addressing this problem of early childhood learning through the EkStep platform,* which is currently under development in India. “EkStep” means “one step” in Hindi.

We realise that there are no easy solutions, but we might have answers if we think about education in new ways. We have learned a lot through the successes and failures of various government initiatives in many countries and “Edtech” solutions that attempt to improve some aspect of education. A common lesson is that we must simultaneously consider the pedagogical, technological, and cultural issues to address the critical constraints that hinder basic learning. In this article, we identify these constraints and explain how they may be addressed, specifically by considering the role of machines and “big data” to augment traditional processes of education.

The EkStep platform leverages the accessibility of smartphones and mobile devices that are becoming increasingly common in the world. The platform provides the pedagogic digital infrastructure for learning that includes fundamental concepts in numeracy and literacy. These concepts are the foundation to further learning. The platform also anchors analytics for content creation, tutoring, assessment, and reporting. All these functions are available programmatically through Application Programming Interfaces. The platform is “open” and designed to allow large-scale crowd-sourced collaboration, curation, and assembly of relevant content in multiple languages. The latter capability is especially important in India considering its linguistic diversity.

The conventional solution to the problem would be to train millions of gifted teachers over the next decade and place them in schools across the country that are equipped with state-of-the-art facilities. This is wishful thinking. We cannot wait for 10 years considering the tens of millions of children who desperately need help today, nor do we have the resources to create physical facilities and train large numbers of well-paid and qualified teachers.

Our solution is to integrate technology creatively into the learning process by addressing the scarce resources involved, namely expertise, time/attention, and varied content. This is not about delivering paper-based learning through a new medium, but about using the new medium’s unique strengths to create new types of engagement and more learning opportunities by addressing the three constraints explicitly. For example, touch-based mobile devices, unlike personal computers, do not require a literate person to interact with them. Unlike a personal computer, the experience in a touch-based mobile device is naturally intuitive and multimedia, so that a child does not need to know how to read, write, or have technology awareness to interact with the device—playing with it using simple touch-based gestures makes the device come alive. Ironically, a machine also has the sociological advantage of being nonjudgmental, basing its advice and actions on data it ingests through its listening, speaking, and responding functions. Indeed, expert human teachers do much the same, but such teachers are in very short supply.

Expertise is clearly a scarce resource, as is the time or attention that can be allocated to each child when it is needed the most. Teachers have expertise but they have limited time and attention to devote to individual students or to prepare tailored content. In contrast, parents, especially mothers who typically spend the most time with their children after school and can supervise their homework, usually lack the expertise and cannot provide critical inputs when needed. Private tutors are often used after school, but they are not cheap and have limited time to customise content to each pupil.

Expertise and attention play a critical role in the various phases of learning, which consists of a series of “Instruction, Practice, and Assessment” cycles. Learning depends on the number of learning opportunities, in effect, the number of such cycles and the quality of assistance available in each cycle. Children vary considerably in terms of their listening, speaking, reading, and skills, which a skilled teacher recognises to tailor assistance accordingly. Could a machine learn to do this effectively?

The short answer is yes in theory, but it will take some effort to design such machines. How might we do so? Fortunately, many of the basic concepts covered in primary education have been well researched and defined, which provides a good basis for evaluation. Numeracy, for example, involves a few dozen concepts that cover basic arithmetic operations, relationships, and shapes.

Literacy is somewhat more complex, involving phonology and visual representation of combinations of symbols known as orthography. Every language has a set of symbols, sounds, and a “threshold vocabulary” that serve as a basis for expression. Indian languages use compound conjugations and modifications of consonants and vowels compared with the linear left to right structure of English. This additional compositional complexity provides an unambiguous correspondence between the visual and the phonetic. Figure 1 shows two similar sounding words in Telugu—“sthree” (woman) and “shree” (auspicious)—where the base consonant “s” is modified twice in the first case (with a soft “t” and an “r”) and the base consonant “sh” is modified once in the second case (with the “r”), and the vowel “e” is stretched as in “eel” in both cases. In addition to the compounding of symbols, Indian languages also allow relatively free word order at the sentence level, meaning there is no “one right” order for a sentence…..for more, click here.