Yong-Bin Kang

Existing research suggests that machine translations of literary texts remain unsatisfactory. Such quality assessment often relies on automated metrics and subjective human ratings, with little attention to the stylistic features of machine translation. Current understanding is limited regarding the extent to which AI may transform the literary translation landscape, with implications for other critical domains for translation such as the creative industries more broadly. This pioneering study investigates the stylistic features of AI translations, specifically examining GPT-4’s performance against human translations of Chinese online literature. Our computational stylometry analysis reveals that GPT-4 translations closely mirror human translations in lexical, syntactic and content features. In addition to showing the relevance of stylometry for analysing the features of AI translation, the study provides critical insights into the implications of AI for literary translation in the posthuman tradition, where the line between machine and human translation becomes increasingly blurry.

The increasing adoption of video-based learning has presented a significant challenge in effectively assessing learner's comprehension and engagement. This paper addresses this gap by conducting a comparative analysis of ten state-of-the-art large language models (LLMs) for Automated Question Generation (AQG) from video transcripts. Our evaluation centres on assessing the quality of four question types: short-answer, multiplechoice, boolean, and gap-fill. Our major contributions are threefold. First, we propose an evaluation framework for comparing multiple LLMs in their ability to generate different question types, identifying the models best suited for each based on five key evaluation metrics-relevance, reading comprehension, question difficulty, clarity, and context utilisation-that holistically assess question quality. Second, we release the prompts used for both question generation and evaluation, providing a valuable resource for future AQG research. Third, we uncover the strengths and weaknesses of each LLM, identifying which question types require further improvement, and uncovering statistically significant differences in AQG performance across the LLMs. By providing an in-depth comparative analysis, this study provides important insight into the development of more effective AQG systems, contributing to the broader integration of AI in automated assessment in video-based learning environments.

This paper investigates open public data and data sharing reforms in Australia (2018–2022) and their potential role in deepening the ‘data divide’. In the contemporary datafied welfare state, open public data and data sharing are increasingly vexed issues in times of data-driven artificial intelligence (AI). We scrutinise public consultation surrounding the establishment of the Australian Data Availability and Transparency Act 2022 (DAT Act). Through topic modelling and critical discourse analysis, the study examines the representation and concerns of marginalised groups in the reform process. We highlight the overlooked role of non-profits and civil society in the public data ecosystem. The analysis emphasises the significant yet unacknowledged contributions of these organisations in advocating for data equity and justice. We argue that responsible and equitable public data practices do not just depend on administrative and technical procedures for data sharing but are fundamentally entwined with the social and institutional hierarchies in which public data is produced and used. The study calls for greater inclusion and support for civil society organisations to bridge the data divide, contributing to broader debates on the merits and challenges of open data and data sharing practices within a data justice framework.

The selection of appropriate cutting tool (CT) is a critical part of machining. Selecting the right tool for the right job will enable customers to achieve economic machining, saving time and cost while delivering high-quality products. Nowadays, the complexity of CT and workpiece is increasing; this changes the input and output requirements for cutting tool selection paving way to automation. There are two types of CT selection methods, manual and automated CT selection. This article focuses on automated CT selection, which has different inputs and outputs based on the algorithm/AI technique used. The potential and promising aspects of CT selection could enhance the machining in terms of productivity, time, cost, and quality aspects. A comprehensive review of automated CT selection methods has been presented in this paper. The review surveys different automated CT selection methods in terms of inputs, outputs, and artificial intelligence (AI) techniques/different algorithms since most of the researchers have not focused on this perspective. It outlines the current status of research and application, which has the potential to improve the automated CT selection methods for the benefit of the manufacturing industry.

Online peer support mental health forums provide an effective and accessible form of support, augmenting scarce clinical and face-to-face assistance. However, to enhance their effectiveness, it is essential to understand the unique characteristics of peer support user groups, and how they participate, contribute and communicate in these forums. This paper proposes and tests a novel approach that leverages stylometry analysis to uncover the unique characteristics of peer support user groups in such forums. Our approach identifies how each group empowers and supports other users, and what distinguishes them from others. The analysis shows that emotion-related words are crucial in identifying and distinguishing user groups based on their writing style. Comparative analysis of emotion expressions across user groups also uncovers the significance of emotional content in these forums in promoting mental well-being. Valued ‘senior contributors’ were more likely than all other groups including trained community guides to use a wide range of both positive and negative emotions in their posts. These findings have significant implications for improving the training of peer-mentors and moderators, scaling forum services, and improving guidelines for emotional expression among peer support users. Our approach presents an objective approach to differentiating the characteristics and communication patterns of valued senior contributors, mentors, and guides, enabling service providers to foster the kinds of communication that supports positive outcomes for distressed users.

The transformation of cities into smarter and more efficient environments relies on proactive and timely detection and maintenance of city-wide infrastructure, including roadside infrastructure such as road signs and the cleaning of illegally dumped rubbish. Currently, these maintenance tasks rely predominantly on citizen reports or on-site checks by council staff. However, this approach has been shown to be time-consuming and highly costly, resulting in significant delays that negatively impact communities. This paper presents AIoT-CitySense, an AI and IoT-driven city-scale sensing framework, developed and piloted in collaboration with a local government in Australia. AIoT-CitySense has been designed to address the unique requirements of roadside infrastructure maintenance within the local government municipality. A tailored solution of AIoT-CitySense has been deployed on existing waste service trucks that cover a road network of approximately 100 kms in the municipality. Our analysis shows that proactive detection for roadside infrastructure maintenance using our solution reached an impressive 85%, surpassing the timeframes associated with manual reporting processes. AIoT-CitySense can potentially transform various domains, such as efficient detection of potholes and precise line marking for pedestrians. This paper exemplifies the power of leveraging city-wide data using AI and IoT technologies to drive tangible changes and improve the quality of city life.

We present CrossSum, a large-scale cross-lingual summarization dataset comprising 1.68 million article-summary samples in 1,500+ language pairs. We create CrossSum by aligning parallel articles written in different languages via cross-lingual retrieval from a multilingual abstractive summarization dataset and perform a controlled human evaluation to validate its quality. We propose a multistage data sampling algorithm to effectively train a cross-lingual summarization model capable of summarizing an article in any target language. We also introduce LaSE, an embedding-based metric for automatically evaluating model-generated summaries. LaSE is strongly correlated with ROUGE and, unlike ROUGE, can be reliably measured even in the absence of references in the target language. Performance on ROUGE and LaSE indicate that our proposed model consistently outperforms baseline models. To the best of our knowledge, CrossSum is the largest cross-lingual summarization dataset and the first ever that is not centered around English. We are releasing the dataset, training and evaluation scripts, and models to spur future research on cross-lingual summarization. The resources can be found at https://github.com/csebuetnlp/CrossSum

Assessing learners’ understanding and competency in video-based digital learning is time-consuming and very difficult for educators, as it requires the generation of accurate and valid questions from pre-recorded learning videos. This paper demonstrates VideoDL, a video-based learning framework powered by Artificial Intelligence (AI) that supports automatic question generation and answer assessment from videos. VideoDL comprises of various AI algorithms, and an interactive web-based user interface (UI) developed using the principles of human-centred design. Our empirical evaluation using real-world videos from multiple domains demonstrates the effectiveness of VideoDL.