# Scientometrics helps learning?

Jinshan Wu

The Institute of Educational Systems Science

School of Systems Science

Beijign Normal University

The 3rd International Conference on Data-driven Knowledge Discovery (CDKD2021)

# Presentation goals

• Share some progresses on merging scientometrics and teaching/learning
• concept networks from textbooks and papers
• usage frequency from papers
• algorithms for learning orders and tests
• Inspire your critical input

# Outlines

• Motivation
• Network of Chinese characters and algorithms for teaching/learning
• Concepts network of physics and some prelimenary results
• Concepts Network of scientometrics
• The author-paper-concept network of scientometrics
• Call for collaborations

# Motivation

• Research can be seen as evolution of concept network and we scientometricians, in principle, know a lot about it
• How this network and its evolution can help the development of science
• as a researcher
• as a research administrator/policy makers
• as a student/teacher
• To make scientometrics relevant to researchers, research administrators, students and teachers

# Network of Chinese characters

## An algorithm for learning order

• Matrix $a^{i}_{j}=1$, $i$ is used in $j$, from the network (knowledge base)
• Vector $W$ represents usage frequency
• Vector $\tilde{W}$ the learning order $$\tilde{W}= \left(1-A\right)^{-1}W= W + AW+A^{2}W+A^{3}W + \cdots$$
• integrates usage frequency and structural role of each concept in the network

## An algorithm for tests

• probability graph model, (not) knowing ($i$)$j$ means very likely also (not) knowing ($j$)$i$ when $a^{i}_{j}=1$
• combined with the above to form personalized learning orders
• more efficient than random sampling

## Summary on network of Chinese characters

• A concept network
• A vector of usage frequency
• Algorithms for learning orders and tests
• Currently working on experimental tests
• Can this be extended to other fields and how?

## Other fields

• English words
• Mathematics
• Physics
• Scientometrics

# The 3-layer network for scientometrics

## Nodes and connections

• The concept layer: concepts, logically connected
• The paper layer: papers, connected by citations
• The author layer: Authos, connected as mentor-mentee
• Intralayers: author-writes-paper, paper-works on (makes use of)-concept

## What can this 3-layer network framework offer?

• As a unified scientometrical data descriptor
• As a platform for asking and answering scientometrical questions
• science map
• research activities
• evaluations
• teaching and learning

# Call for collaborations

• The concept network of scientometrics by and for the research community
• Even the concept network of each discipline by scientometricians and for researchers in the discipline
• Making scientometrics to be a fundamental tool for researchers and students/teachers of each discipline

# Question time

• Thank you for your attention
• Take-home: via the author-paper-concept network for scientometrics and the algrithms over the network, scientometrics can help researcher, research adminstrator, students and teachers
• Let’s make it happen all together