New Zealand School of Education
Wednesday 3 October 2018
Concurrent Session 1A
Thursday 4 October 2018
Concurrent Session 2B
An AI-Based Conversation Bot For Facilitating STEM Education
Education system around the world currently places more emphasis on Science, Technology, Engineering, and Mathematics (STEM) education for a viable workforce in the high-growth technology sector. In the last decade, government has been encouraging schools in New Zealand to promote STEM programs to prepare students to find their interest for tertiary education. Teaching STEM is important to spark the interest of young minds for future success at later stages of education. Using innovative means to teach STEM makes it more attractive for the students, and easier to prepare and deliver for the teachers who normally need to manage a heavy workload. Using conversation bot or chatbot in the learning environment have been spotted as a tool that can improve learner interest, memory retention and knowledge transference. It also can have a great impact on student’s interaction with computer which have significant effect in students learning. Students also can clear their doubt and find the answer without embarrassment of asking question.
We introduce an AI-based conversation bot for facilitating STEM education. Although there have been many tools and online courses, keeping students engaged has always been a major problem. Our conversation bot is able to conduct a two-way communication with students and answer their questions on different subjects. It can be customised for different age groups and for the Teriary education particularly early levels. It also evolves and learns from ongoing conversations and the feedback provided by users. Preliminary feedback from teachers and students show that there is a deep interest in our solution.
A Boosted Student Management System
New Zealand School of Education is a well-established and category 1 PTE in Auckland which uses aPlus+ a course and assessment management system and PEPi for the student records management. aPlus+ offers various features for managing classes and students information such as student profile, attendance data, assessment information, grades and etc. Also, it visualises students’ attendance data and attendance rate for a course as well as a whole programme. It is also used by the lecturers to verify and approve the students grades and transfer the information to the student management system PEPi.
We have developed a complementary and a customised tool on top of the existing tools which pulls out data from them and perform various analysis, and provides information and visulaisations to the user via a web-based dashboard. The tools integrates the data extracted from aPlus+ and PEPi and provides required information for EER via a dashboard. It also sends alerts with related information to the course facilitators when any risk, unusual activity, and or incomplete task are identified. For example, the tool traces the attendance data for a given period and identifies the dates and classes that a lecturer has missed to complete the attendance for students. It also performs prediction on the students attendance and performance which helps students support team to identify students susceptible to high absence rate in advance. It can provide forecast and insights to enrollment, and marketing team as well on the prospective enrollments and business risks in the future. This system automates some of repetitive and tedious tasks and facilitates various improvements in the organisations.
The proposed system enables customised improvements in an educational institution with some of disruptive
technologies. Our solution addresses real-life issues which are experienced by the employees and it embraces
technology to provide a solution. As vocational institutions, polytechnics and private tertiary educations (PTEs) usually go through different NZQA and self-assessment processes, they need to have a constant access for the assessment of their data. Our solution facilitates the tedious task of extracting data from databases and processing them for certain purposes. This reduces the required time, effort and human resources for some of the routine and repetitive tasks which are mostly performed manually such as the data extraction required for EER (External Evaluation and Review) process. It helps different users across organisation to use this system for various purposes. Enrollment, student support, and marketing teams are able to obtain insights about the
future prospects of the enrollment, success and retention rates etc. Therefore, it brings value to the organisation and improves the productivity as well. The system is non-invasively integrated and deployed to an existing management system without interrupting business and employees.
Although existing course and student management systems provide variety of features, they are not customised for some of common processes of an educational organisation. The academic and non-academic staff are required to extract data from the system and they also sometimes need to perform data cleaning, transformation, processing and analytics on the data. Due to experiencing these challenges we decided to create an automated system on top of the existing one to extract information from the databases. One of the
main challenges is to access the back end and pull out data from databases without interrupting business and staff and with the minimum risk to the system security. Currently, our developed system access the data through a shadow copy of the databases however, there is a possibility to directly link this with the main system. The access to databases is one of the challenging parts, though this has been facilitated via an API provided by the back end system. Also, deploying our system on the cloud (Azure) as a web-based dashboard enables users to easily access the service, though this increases the cost to the organisation. Thus an operational challenge is to convince the finance department to pay for the cloud service cost. The system should be expanded for the larger number of users in terms of the Azure virtual machines which could be a potential challenge in terms of scalability in the future.
We have implemented an analytic engine with R and integrated it into power BI. The analytic part is responsible for extracting data from the back end, pre-processing and analysing data. The Power BI is used for visualising the insights and providing forecast information. Also, the R engine sends alerts to the users for certain observations (e.g. alert to lectures if attendance data is missing in the records for a particular period such as a week). This is deployed on the Azure platform running a virtual machine. Using R in Power BI enables us to expand the system with machine learning tools to provide required insights to the organisation.
Farhad Mehdipour received his PhD in computer science and completed his post-doctoral research on high performance and low power computing systems in Kyushu University, Japan between 2007 and 2010.
From 2006 to 2016, he was a senior R&D engineer and an academic in Japan, where he participated in several joint projects with the industry in Japan.He is currently a research leader, and senior lecturer at Techschool at NZSE (New Zealand School of Education), a part of New Zealand Skills & Education Group. Farhad has a broad experience both in industry and academia and has initiated and led several interdisciplinary R&D projects particularly on AI, data analytics, and IoT-enabled smart systems.