Formative assessment methods for intelligent learning systems

Abstract

The dissertation analyzes the development trends of intelligent multiagent systems for personalized learning, proposes a new system’s conceptual model, and explores the perspectives of individual intelligent agents, increasing the efficiency of individualized teaching. Based on the results, computer adaptive testing algorithms are proposed for assessing student achievement. The dissertation consists of an introduction, three chapters, general conclusions, references, and lists of the author’s publications on the dissertation’s topic. The introductory chapter discusses the research problem, the relevance of the thesis, describes the object of research, formulates the aim and objectives of the work, describes the research methodology, scientific novelty of the work, the practical significance of the results, and defended statements. The introduction closes by listing the author’s publications and conference papers on the dissertation’s topic and presenting the dissertation’s structure. The first chapter introduces the latest literature review. This chapter focuses on current trends in smart learning, including personalization and formative assessment methods. Several assessment models and algorithms are reviewed. This chapter concludes by clarifying the main objective and tasks of the thesis and a summary of the literature review findings. The second chapter describes some review results, which were executed to get a view of students’ knowledge testing media and learning style preferences. Additionally, it presents the smart multiagent learning system’s conceptual model. It is based on integrating existing learning style estimation methods, data mining and neural network solutions, and newly presented formative assessment agents. The third chapter presents the developed computer-adaptive methodologies and algorithms based on Upper-Confidence Bound and Elo rating for formative assessment capable of suggesting a further course of study. The experiments and modeling results are presented to illustrate the benefits of the proposed algorithms compared to existing formative assessment methods. Eleven articles were published on the dissertation’s topic: two in journals with an impact factor included in the Clarivate Analytics Web of Science database, three in other peer-reviewed journals, and six in conference proceedings, five of which are included in the Clarivate Analytics Web of Science database. Six presentations were made on the dissertation’s topic at national and international conferences.

Disertacijoje nagrinėjamos intelektualių daugiaagenčių sistemų, skirtų mokymui ir/ar mokymuisi personalizuoti, vystymosi tendencijos, siūlomas naujas sistemos konceptualus modelis, tiriamos atskirų intelektualiųjų agentų perspektyvos didinant individualizuoto mokymo efektyvumą. Atsižvelgiant į rezultatus, pasiūlomi nauji kompiuteriniai adaptyvaus testavimo algoritmai, skirti studentų pasiekimų formuojamajam vertinimui.