This study explores the potential to enhance the performance of convolutional neural networks (CNNs) for automated scoring of kinematic graph answers through data augmentation using Deep Convolutional Generative Adversarial Networks (DCGANs). By developing and fine-tuning a DCGAN model to generate high-quality graph images, we explored its…

Automatic short answer grading (ASAG) is a challenging task that aims to predict a score for a given student response. Previous works on ASAG mainly use nonneural or neural methods. However, the former depends on handcrafted features and is limited by its inflexibility and high cost, and the latter ignores global word cooccurrence in a corpus and…

As a first step toward automatic feedback based on students' strategies for solving histogram tasks we investigated how strategy recognition can be automated based on students' gazes. A previous study showed how students' task-specific strategies can be inferred from their gazes. The research question addressed in the present article is how data…

Online collaborative learning is implemented extensively in higher education. Nevertheless, it remains challenging to help learners achieve high-level group performance, knowledge elaboration, and socially shared regulation in online collaborative learning. To cope with these challenges, this study proposes and evaluates a novel automated…

The pedagogical modelling of everyday classroom practice is an interesting kind of evidence, both for educational research and teachers' own professional development. This paper explores the usage of wearable sensors and machine learning techniques to automatically extract orchestration graphs (teaching activities and their social plane over time)…

The purpose of this study is to describe a methodology to recover the item model used to generate multiple-choice test items with a novel graph theory approach. Beginning with the generated test items and working backward to recover the original item model provides a model-based method for validating the content used to automatically generate test…

We present a new system for automated scoring of graph construction items that address complex science concepts, feature qualitative prompts, and support a range of possible solutions. This system utilizes analysis of spatial features (e.g., slope of a line) to evaluate potential student ideas represented within graphs. Student ideas are then…

This paper describes theory-led design as a way of developing novel tools for learning analytics (LA). It focuses upon the domain of automated discourse analysis (ADA) of group learning activities to help an instructor to orchestrate online groups in real-time. The paper outlines the literature on the development of LA tools within the domain of…

Our study integrates automated natural language-oriented assessment and analysis methodologies into feasible reading comprehension tasks. With the newly developed T-MITOCAR toolset, prose text can be automatically converted into an association net which has similarities to a concept map. The "text to graph" feature of the software is based on…

To date there has been very little work on the machine understanding of imprecise diagrams, such as diagrams drawn by students in response to assessment questions. Imprecise diagrams exhibit faults such as missing, extraneous and incorrectly formed elements. The semantics of imprecise diagrams are difficult to determine. While there have been…

Describes research that investigated a method of automatically generating concept association maps that could be useful to abstractors. The use of word stems as concept surrogates, cooccurrence to define concept links, and a general-purpose graph-drawing algorithm is described; and results of evaluation of two link selection methods are reported.…

Describes work performed under the Spelling Error Detection Correction Project (SPEEDCOP) at Chemical Abstracts Service to devise effective automatic methods of detecting and correcting misspellings in scholarly and scientific text. The trigram analysis technique developed determined sites but not types of errors. Thirteen references are listed.…