Vivekanandan Suresh Kumar, PhD

Researcher in Learning Analytics, Technology-Enhanced Learning, Artificial Intelligence in Education

Research

Focus and Vision

Big Data Learning Analytics and Causal Modelling

Imagine a software agent that calls your computer its home, helps you to study better, be your friend, and motivates you to act in your long-term best interest, consistent with your own deepest values, personal, social, and other standards.

My research aims to build, deploy, and evaluate a suite of such agents called anthropomorphic agents that mimic and do better on human-like traits to assist learners in their regulatory tasks. Further, these agents will self-regulate, as well as causally model their own anthropomorphic traits such as initiative and subject matter competence in order to inspire high level of trust, shared perspective, and entitativity, thus increasing the learners’ willingness to view them as competent partner for co-regulation and to inspire learners to make the optimal regulatory changes.

To sustain and to succeed in performing such tasks, the agents rely on big data learning analytics techniques where:

  1.  sensors continuously observe study episodes of learners,
  2.  distributed ontologies continuously instantiate study activities,
  3.  causal models continuously establish relationships among study experiences,
  4.  high performance platforms continuously offer real-time computing resources, and
  5.  pedagogical models continuously look for ways to motivate, engage, classify, learn, and promote subject knowledge and study skills.

These agents aim to attune themselves to the needs of an individual learner or a group of learners, over life, to optimise learning experiences.

Areas

Artificial Intelligence in Education

  • Big Data Learning Analytics
  • Intelligent Tutors and Smart Objects
  • User Modelling and Model Tracing

Self-Regulated and Co-Regulated Learning

  • Computational Models of self- and co-regulation
  • Computer-Supported Collaborative Learning
  • Computational Models of Grit

Human-Computer Interaction

  • Mixed-Initiative
  • Real-time feedback and regulation

Causal Modelling

  • Knowledge Engineering, Representation, and Reasoning
  • Game-based causality
  • Longitudinal and Observational causal models
  • Discovery of causal models

Technology-Enhanced Teaching, Learning, and Research

  • Big Data Learning Analytics
  • Intelligent Tutors and Smart Objects
  • User Modelling and Model Tracing

Pedagogical Innovations

Learning Analytics
A suite of domain-specific instructional techniques to help students reach their full potential through evidence-based, machine-learned, anthropomorphic, immersive, competence-measured and data-rich environments
Open Research / Science
A socially networked open research environment that virtualises interactions, blends reality, leverages affordances of a global community of researchers/scientists
ICT4D policy choices to improve educational diversity
Computational mechanisms that empower common public to validate research outcomes and conduct own research, guided by anthropomorphic pedagogical agents. Outcome: A mixed initiative interface for the consumption of research publications by the Public
Immutable student interactions repository
Hosts interaction patterns of students, open exploratory models, and reusable content, shared by inter- and intra-disciplinary faculty members, using blockchain technology. Outcome: a distributed data repository system
Lean and Agile way of teaching Research Methods
Measures and improves researching skills of student researchers, effectiveness of research supervisors and research capacity of institutions by modularising learning activities. Outcome: a Research Methods MOOC
Self- and Co-regulation (SRL/CRL)
Engages learners in study-oriented self- and co-regulatory processes, to create, monitor, and adapt regulation-oriented initiatives and measure their impact. Outcome: a pluggable SRL/CRL component for Moodle, Eclipse and Word
Learning outcomes mapping
Meaningfully connect learning outcomes, study activities, course content, students’ study episodes, assessment rubrics, competences and meta-competences. Outcome: a pluggable mapper for Moodle
Learner-initiated Instructional design
Engages graduate students in the instructional design process of a course, empowering students to scope topics, choose assessment types, change submission deadlines, and commit to regulation initiatives.
Engaged lectures
Measures and publishes information on student-engagement, at real-time, as and when students are engaged in study activities in lecture halls and in virtual reality worlds. Outcome: a video analyzer of body language, a video analyzer of facial expressions and a sentiment analyzer of textual interactions
Peer help
Examines incentive-based peer-help model to measure preparedness of students for a targeted learning activity. Outcome: a repository and causal model of study activities of students

Copyright © Claudia Guillot 2018. All Rights Reserved.