Menu

Education Today Logo

An AI assistant reins in dropout risk

News Image

UNSW Sydney has around 17,000 students enrolled in the engineering faculty and there can be 500 students enrolled in a class at any one time. To help their learning and reduce the risk of drop out, Dr David Kellermann, a senior lecturer in the school of mechanical and manufacturing engineering has employed artificial intelligence and rich data analytics together with Microsoft Teams to promote collaboration and communication

Kellermann made Question bot – an AI infused chatbot that can to answer student queries. It uses machine learning to build a body of knowledge that students can ask for support 24×7.

Because all of Kellermann’s lectures are captured in Stream, Question Bot provides an answer to their question and can direct students to the exact point in the lecture video where the relevant issue is being addressed.

Kellermann, Microsoft partner Cloud Collective, and just one developer, used Graph API and Bot Framework, to develop Question bot in just eight weeks. Kellermann uses the QnA Maker Azure machine learning cognitive service to train the bot.

“I use QnA Maker, a Cognitive Service on Azure, in order to train the AI of the bot. And within a couple of weeks, it started answering questions on its own. But not just that, Question Bot was also able to direct the students back to the conversations where their peers had been talking about similar problems. That’s reconnecting people and building learning communities,” said Kellermann.

He uses the data collected across the platform to help identify at-risk students, and then personally engages with them to see if they need extra support to help them through rocky periods and ultimately to complete their studies.

“Question bot is actually creating a study resource for the students filtered by topic. It’s not a textbook, it’s made out of their own collaboration, their own discussion automatically. In fact, in the first two weeks alone, Question bot created 200 high-quality topic filtered question and answer pairs.

“I use QnA Maker, a Cognitive Service on Azure, in order to train the AI of the bot. And within a couple of weeks, it started answering questions on its own. But not just that, Question Bot was also able to direct the students back to the conversations where their peers had been talking about similar problems. That’s reconnecting people and building learning communities.”

Kellermann has been able to boost impact by using QR codes in student workbooks that instantly alert both lecturers and Question bot to the topic being tackled when a student poses a question.

The data collection that’s amassed from across the platform establishes a digital feedback loop for UNSW Sydney that is available to students and lecturers. Both can see how they are travelling with their studies, and also glean context about their results.

That’s important because a student might be worried about what seems like a low score on one test, for example – but the context from the data might show it was a really tough question and they actually did quite well. The data also provides Kellermann with early warning of students at risk of dropping out – if their marks suddenly slide, or engagement falls away – allowing him to personally reach out to offer support and help them through what might otherwise be a rocky period.

“We’re actually able to identify at risk students by week four of the course. I reach out personally to every one of those at-risk students to say, ‘Hey, how can I help?’,” he said.

Kellermann explained that by integrating systems, he now has access to very highly structured data on SQL. This allows one-click Power BI dashboards to be created – which can be accessed from smartphones – for students to get their marks and see them in context.

He has also used artificial intelligence to create personalised study plans for each student to improve their learning outcomes.

“I used 2017 data to train an Azure Machine Learning model to correlate all of the information against student performance. And using a database of competency-ranked resources – every single topic against every difficulty level, and an algorithm on .NET, we automatically assembled 500 individual personalised optimised study packs for every student based on the prediction of not only their exam results, but their exam results for every individual question two weeks before. And uploaded to SharePoint with personal access, one click in Teams.”

According to Steven Miller, Education lead of Microsoft Australia; “Solutions like AI Builder in the PowerApp family make it easier for customers to create their own AI-infused applications, while the work we have done to percolate intelligence into business applications like Dynamics 365 can be a game changer for business of all sizes.

“The AI-led transformation led by Dr Kellermann is attracting attention from learning institutions all over the world.”


6 Aug 2019 | National
Awesome Destination Moon Resource Book for National Science Week News Image

With more than 2000 planned events around Australia, National Science Week, 10–18 August, is going to be huge. One great free accompanying resource is the free Destination Moon Resource Book. Read More

6 Aug 2019 | Melbourne
A STEAMy degree, finally News Image

La Trobe University has launched the brand-new Bachelor of Humanities, Innovation and Technology – a fusion of humanities with business and emerging technologies. Read More

6 Aug 2019 | Melbourne
The STEAM Futures Conference at Viewbank College News Image

Want to be a part of a community developing exciting initiatives in STEAM? The STEAM Futures Conference by Viewbank College on August 23rd 2019 is your chance to do it. Read More

6 Aug 2019 | National
Digging in to save amazing dinosaur gem News Image

A collection of ancient rock with opalised bones from Lightning Ridge in NSW has been brought to life through a novel project involving scientists from around Australia. Read More

6 Aug 2019 | National
Devika’s Virtual worlds in classrooms News Image

Start-up Devika has been focusing on introducing Virtual Reality (VR) to Australian classrooms to make experiential learning software a feature in learning.
Read More