Revisiting five decades of educational technology research： a content analysis of the British Journal of Educational Technology

Melissa Bond， Olaf Zawacki-Richter and Mark Nichols

Reflecting on 50 years of educational technology research， a content analysis was conducted of 1，777 research article titles and abstracts， published in the British Journal of Educational Technology （BJET） from 1970 to Issue 3， 2018. A text-mining tool （Leximancer） was used to identify key concepts and themes emerging throughout each of the five decades， which were then compared to those found in a previous analysis of Computers & Education， as well as the most cited BJET publications in each decade. Common themes in BJET throughout the past 50 years have included the evolution of teaching and learning in distance education， the emergence of instructional design， misunderstanding between practitioners and learning designers， issues of pre- and in-service teacher education and technology uptake by educators and students， including the confidence to do so， the technology skills of educators and students， as well as a lack of institutional support to provide space and time for training and integration to occur. Suggested future research areas include finding alternative models of educator professional development， and further exploration of the role of theory and policy.

Keywords： educational technology; research; content analysis; online learning; distance education; open learning; text-mining; research trends

An empirical study of knowledge production from the perspective of connectivism： comparative analysis based on machine learning

Xiaoshan Li， Li Chen， Wenjing Wang and Yanyan Li

Unlike behaviorism， cognitivism and constructivism， connectivism focuses on the role of learner community in knowledge production. Nevertheless， how knowledge is produced by the learner community is an under-researched issue. Using the tool of Word2vec semantic analysis， this study attempts to explore this issue by comparing how knowledge of the same theme is produced in the connectivist model， using a cMOOC as an example， and in the traditional model， which refers to findings from relevant Chinese core journal publications. Findings show that knowledge production in cMOOC is characterized by theme focus， theoretical freshness， and broad vision. In contrast， the traditional model of knowledge production has such features as hierarchical structure， relevance to specific groups， and supply-side orientation. Implications of these findings are then discussed in terms of knowledge producer， knowledge ideology and knowledge application.

eywords： Internet +; online learning; learner community; knowledge production; collective intelligence and collaboration; machine learning; semantic analysis; Word2vec; knowledge production feature

Professional talent cultivation in the context of AI： an empirical study of the profession of accounting

Yijun Wang and Youran Yang

Artificial intelligence （AI） is commonly recognized as a General Purpose Technology （GPT） that will have a profound influence on labor skills and labor markets， hence leading to concerns and controversies over technological unemployment. This study sets out to investigate how to cultivate qualified professionals in the age of AI. It starts by distinguishing Artificial Narrow Intelligence and Artificial General Intelligence. Informed by classical theories of labor economics， it then adopts the ALM model created by Autor， Levy and Murnane as the research framework. Based on the O net online data and China-related data， it analyzes the changes in the overall skill demands from the accounting profession. Findings show that instead of routine cognitive skills traditionally required by the profession， it is non-routine analytic skills and non-routine interactive skills that are in increasing demand. Further， Python technology is used to identify word frequency of the accounting job recruitment texts from major recruitment websites in China. The results verify the findings from the O net online data and China-related data and refine the specific skill elements. Implications and suggestions are also discussed.

Keywords： Artificial Intelligence; professional talent cultivation; accounting; routine skills; non-routine skills

Integration of learning spaces in the intelligent age： pattern and path

Xianmin Yang， Yifei Li， Dongli Wang and Beibei Xing

In the intelligent age， the application of emerging technologies such as big data， cloud computing， and smart technology in education has led to the emergence of new learning methods such as ubiquitous learning， seamless learning， and smart learning. As a major place for learning， the learning space has been significantly transformed. One of the important trends in change is the convergence of learning spaces， which provides a seamless environment for learners to integrate virtual reality with reality， enabling them to engage in formal and informal learning easily， effectively， and enthusiastically. This article expounds on the main modality of the learning space and the essential connotation of spatial integration in the contemporary context. It is argued that spatial integration has two core characteristics -"consistency in teaching design" and "continuity of learning chain" and that there are three typical forms of integration， namely integration of physical spaces， integration of information spaces， and integration of physical and information spaces. Paths of integration are also discussed in terms of learning and teaching elements.

Keywords： intelligent age; learning space; physical space; information space; spatial integration; instructional design; path of integration; mode of integration

Deep learning： occurrence， design model and mechanism interpretation

Hang Hu， Yaxin Li， Qie Lang， Hairu Yang， Qiuhua Zhao and Yifan Cao

Drawing upon cognitive and learning psychology theories such as schema， gestalt， cognitive load， ACT-R and APOS， this evidence-based study reviews the research processes， conclusions and cases of study in relation to empirical research into deep learning in the past three years. It analyzes the way learning takes place in the S-AICG model of deep learning and develops a design model of deep learning classroom. Using mathematic learning as an example， it then constructs a model of the mechanism of deep learning， explaining the model in terms of representation， interaction， adaptation， construction and generation.

Keywords： deep learning; empirical induction; comparison of taxonomy of educational objectives; cognitive psychology; occurrence; design model; occurrence mechanism; CTCL paradigm

（英文目次、摘要譯者：肖俊洪）