School Days May See Semantic Tech Help With Online Learning Assessments

By on

It’s getting to be that time again – yup, school days are getting into full swing. Education, of course, is going through a lot of change these days. The Common Core State Standards initiative is changing what students must learn and what teachers must teach in the early grades, while school-specific online courses now are joined by massive online learning courses (MOOCs) that are bringing new learning experiences on a large-scale to everyone from high school and college students to adults who haven’t taken courses inside a live classroom for decades.

It’s under these circumstances that startup Cognii is hoping to make its mark by applying natural language processing and semantic technology to automate assessments for online learning for students and to grade essays for educators. Its initial focus is on the online education sector, though founder and CTO Dharmendra Kanejiya – whose background involves developing algorithms to improve speech recognition at Vlingo, which were applied to Nuance Communications’ solutions when it acquired the company – says it also can have applicability in the real-world classroom.

In online scenarios, Cognii’s several linguistic features, including syntactic and deeper semantic analysis that extends to enable a hierarchical conceptual representation of students’ and teachers’ information, comes into play. It is to be used in content area subjects where a clear conceptual understanding is required for essay answers, so that when a student types a response, “we compare that to what is expected and measure it more accurately,” says Kanejiya. It isn’t, for example, appropriate for essay questions that could have very varied manifestations, like those asking students to describe their favorite summer experience, for instance.

As an example, if a student is asked to explain the five causes of a certain war, Cognii would expect the teacher to provide the ideal answer listing the reasons, which would be put into its database and where hierarchical analysis would be performed on it. A student’s answer is matched against that. The technology can be set up such that the student writing an online essay can receive instant feedback from the system should they miss components of the answer, asking them about further elaborating on the topic. That likely would be more appropriate for formative rather than summative assessments, he notes.

For both formative and summative assessments, the technology kicks in to automatically mark the responses for the teacher, based on a rubric the teacher provides that includes the maximum score to give a student per answer. Should technology be grading students? As Kanejiya notes, in online learning, particularly in MOOCs, the ratio can be 1,000-students to one-teacher, so the help comes in handy.

The ratio in live classroom environments at lower grades isn’t as dramatic as it can be with a MOOC, of course, but demands on teachers’ time are as great. “Teachers spend about 30 percent of their time grading work. They are very much burdened in how much they can grade vs. instruct or counsel students,” he says. Automating grading on formative assessments means that teachers save time that they then can apply back to instruction and student counseling, to better prepare them for summative testing.

“The goal is to improve the productivity of teachers so they can give more coaching. That is where the technology can definitely enhance,” he says. “Then the high-stakes tests would be graded by the teacher.”

Its NLP capabilities require customization for every scenario; its assessment engine for one educational course isn’t going to be directly applicable to another. But, Kanejiya says, the approach still has advantages over other solutions that leverage machine learning. “They need more than100 answers pre-graded by a teacher for the technology to predict scores for new answers,” he says. “We propose our technology can work with just one correct answer for each question, and we can provide the best quality assessment for the answers.”

While the company is targeting school/online learning environments now for interactive learning, assessment and grading, the opportunities to apply its solution in other respects are there, he says. In student scenarios, for instance, it could be used not just for testing but as an intelligent tutoring system for a student to learn at his own pace, or to measure the ability of students to understand any concept being taught at the time. But corporations’ employe training programs also could benefit, he says.

Currently,  Cognii is developing custom demos for a few possible customers, including large publishing companies with educational content that is static, and which they want to make interactive and accessible. It’s also working with an online distance education company that provides credits to students while they are working, so they can enter college fulltime in their appropriate year if they pass their online assessments, he says.


Leave a Reply