Persona-based dialogue generation is an important milestone towards building conversational artificial intelligence. Despite the ever-improving capabilities of large language models (LLMs), effectively integrating persona fidelity in conversations remains challenging due to the limited diversity in existing dialogue data. We propose a novel framework SBS (Score-Before-Speaking), which outperforms previous methods and yields improvements for both million and billion-parameter models. Unlike previous methods, SBS unifies the learning of responses and their relative quality into a single step. The key innovation is to train a dialogue model to correlate augmented responses with a quality score during training and then leverage this knowledge at inference. We use noun-based substitution for augmentation and semantic similarity-based scores as a proxy for response quality. Through extensive experiments with benchmark datasets (PERSONA-CHAT and ConvAI2), we show that score-conditioned training allows existing models to better capture a spectrum of persona-consistent dialogues. Our ablation studies also demonstrate that including scores in the input prompt during training is superior to conventional training setups. Code and further details are available at https://arpita2512.github.io/score_before_you_speak.

In this age of life-wide learning, video-based learning has increasingly become a crucial method of education. However, the challenge lies in watching numerous videos and connecting key points from these videos with relevant study domains. This requires video characterization. Existing research on video characterization focuses on manual or automatic methods. These methods either require substantial human resources (experts to identify domain related videos and domain related areas in the videos) or rely on learner input (by relating video parts to their learning), often overlooking the assessment of their effectiveness in aiding learning. Manual methods are subjective, prone to errors and time consuming. Automatic supervised methods require training data which in many cases is unavailable. In this paper we propose a weakly supervised method that utilizes concepts from an ontology to guide models in thematically classifying and characterising video segments. Our research is concentra ted in the health domain, conducting experiments with several models, including the large language model GPT-4. The results indicate that CorEx significantly outperforms other models, while GLDA and Guided BERTopic show limitations in this task. Although GPT-4 demonstrates consistent performance, it still falls behind CorEx. This study offers an innovative perspective in video-based learning, especially in automating the detection of learning themes in video content.

Using either a search engine or website navigation can pose difficulties for web users in certain situations. Therefore, creating a useful visualization of the website structure can aid web users in navigating through a website. By employing visualization techniques, the website structure can be presented as regions that provide clarity on all pages, links, and labels. To address these challenges, our goal is to identify the top three existing visualization layouts that improve website navigation. To achieve this, we aim to evaluate different types of existing visualization layouts by applying the region mechanism to visualize website structures. Initially, we designed a region mechanism that ensures scalability and ease of navigation and understanding for website structures. Subsequently, we conducted a technical evaluation to assess the performance of 12 existing visualization layouts on a website structure. During the evaluation, we generated 12 layouts for each webpage in the University of London (UoL) website. We then calculated three metrics for each layout: average aspect ratio, visualization area, and label overlapping. Finally, based on the results obtained, we identified the three most suitable visualization layouts for visualizing a website structure. These selections were made based on low aspect ratio and stability, effective control of the visualization area, and support for readability. Overall, this research focuses on improving website navigation by evaluating existing visualization layouts and selecting the most suitable ones. The chosen layouts are determined based on their ability to maintain low aspect ratio, ensure stability, control visualization area, and support readability.

Abstract Whitby is the server-side of an Intelligent Tutoring System application for learning System-Theoretic Process Analysis (STPA), a methodology used to ensure the safety of anything that can be represented with a systems model. The underlying logic driving the reasoning behind Whitby is Situation Calculus, which is a many-sorted logic with situation, action, and object sorts. The Situation Calculus is applied to Ontology Authoring and Contingent Scaffolding: the primary activities within Whitby. Thus many fluents and actions are aggregated in Whitby from these two sub-applications and from Whitby itself, but all are available through a common situation query interface that does not depend upon any of the fluents or actions. Each STPA project in Whitby is a single situation term, which is queried for fluents that include the ontology, and to determine what pedagogical interventions to offer. Initially Whitby was written in Prolog using a module system. In the interest of a cleaner architecture and implementation with improved code reuse and extensibility, the initial application was refactored into Logtalk. This refactoring includes decoupling the Situation Calculus reasoner, Ontology Authoring framework, and Contingent Scaffolding framework into third-party libraries that can be reused in other applications. This extraction was achieved by inverting dependencies via Logtalk protocols and categories, which are reusable interfaces and components that provide functionally cohesive sets of predicate declarations and predicate definitions. In this paper the architectures of two iterations of Whitby are evaluated with respect to the motivations behind the refactor: clean architecture enabling code reuse and extensibility.

Nowadays, the profound digital transformation has upgraded the role of the computational system into an intelligent multidimensional communication medium that creates new opportunities, competencies, models and processes. The need for human-centered adaptation and personalization is even more recognizable since it can offer hybrid solutions that could adequately support the rising multi-purpose goals, needs, requirements, activities and interactions of users. HAAPIE workshop embraces the essence of the "human-machine co-existence" and brings together researchers and practitioners from different disciplines to present and discuss a wide spectrum of related challenges, approaches and solutions. In this respect, the fifth edition of HAAPIE includes 5 long papers.

Urban infrastructure assets (e.g. roads, water pipes) perform critical functions to the health and well-being of society. Although it has been widely recognised that different infrastructure assets are highly interconnected, infrastructure management in practice such as planning, installation and maintenance are often undertaken by different stakeholders without considering these dependencies due to the lack of relevant data and cross-domain knowledge, which may cause unexpected cascading social, economic and environmental effects. In this paper, we present a knowledge based decision support system for urban infrastructure inter-asset management. By considering various infrastructure assets (e.g. road, ground, cable), triggers (e.g. pipe leaking) and potential consequences (e.g. traffic disruption) as a holistic system, we model each sub-domain using a modular ontology and encapsulate the interdependence between them using a set of rules. Moreover, qualitative likelihood is assigned to each rule by domain experts (e.g. civil engineers) to encode the uncertainty of knowledge, and an inference engine is applied to predict the potential consequences of a given trigger with location specific data and the encoded rules. A web-based prototype system has been developed based on the above concept and demonstrated to a wide range of stakeholders. The system can assist in the process of decision making by aiding data collation and integration, as well as presenting potential consequences of possible triggers, advising on whether additional information is needed or suggesting ways of obtaining such information. The work shows an intelligent approach to integrate and process multi-source data to pioneer a novel way to aid a complex decision process with a high social impact.

We present an interactive decision support system for assisting city infrastructure inter-asset management. It combines real-time site specific data retrieval, a knowledge base co-created with domain experts and an inference engine capable of predicting potential consequences and risks resulting from the available data and knowledge. The system can give explanations of each consequence, cope with incomplete and uncertain data by making assumptions about what might be the worst case scenario, and making suggestions for further investigation. This demo presents multiple real-world scenarios, and demonstrates how modifying assumptions (parameter values) can lead to different consequences.

In this paper we leverage the explicit user profile (relating to experience, knowledge, and self-regulation) to predict user engagement in active video watching. Data from two user studies for informal learning of presentation skills in a Higher Education context is used to develop and validate the prediction models. Our results show that these user characteristics can reasonably predict the overall engagement (inactive, passive and constructive learners). Our approach can be used to inform adaptive interventions that prevent disengagement and enhance the learning experience.

We present an uncertainty management scheme in rule-based systems for decision making in the domain of urban infrastructure. Our aim is to help end users make informed decisions. Human reasoning is prone to a certain degree of uncertainty but domain experts frequently find it difficult to quantify this precisely, and thus prefer to use qualitative (rather than quantitative) confidence levels to support their reasoning. Secondly, there is uncertainty in data when it is not currently available (missing). In order to incorporate human-like reasoning within rule-based systems we use qualitative confidence levels chosen by domain experts in urban infrastructure. We introduce a mechanism for the representation of confidence of input facts and inference rules, and for the computation of confidence in the inferred facts. We also present a mechanism for computing inferences in the presence of missing facts, and their effect on the confidence of inferred facts.

Videos can be a powerful medium for acquiring soft skills, where learning requires contextualisation in personal experience and ability to see different perspectives. However, to learn effectively while watching videos, students need to actively engage with video content. We implemented interactive notetaking during video watching in an active video watching system (AVW) as a means to encourage engagement. This paper proposes a systematic approach to utilise learning analytics for the introduction of adaptive intervention - a choice architecture for personalised nudges in the AVW to extend learning. A user study was conducted and used as an illustration. By characterising clusters derived from user profiles, we identify different styles of engagement, such as parochial learning, habitual video watching, and self-regulated learning (which is the target ideal behaviour). To find opportunities for interventions, interaction traces in the AVW were used to identify video intervals with high user interest and relevant behaviour patterns that indicate when nudges may be triggered. A prediction model was developed to identify comments that are likely to have high social value, and can be used as examples in nudges. A framework for interactive personalised nudges was then conceptualised for the case study.

The domain of cultural variations in interpersonal communication is becoming increasingly important in various areas, including human–human interaction (e.g., business settings) and human–computer interaction (e.g., during simulations, or with social robots). User‐generated content (UGC) in social media can provide an invaluable source of culturally diverse viewpoints for supporting the understanding of cultural variations. However, discovering and organizing UGC is notoriously challenging and laborious for humans, especially in ill‐defined domains such as culture. This calls for computational approaches to automate the UGC sensemaking process by using tagging, linking, and exploring. Semantic technologies allow automated structuring and qualitative analysis of UGC, but are dependent on the availability of an ontology representing the main concepts in a specific domain. For the domain of cultural variations in interpersonal communication, no ontological model exists. This paper presents the first such ontological model, called AMOn+, which defines cultural variations and enables tagging culture‐related mentions in textual content. AMOn+ is designed based on a novel interdisciplinary approach that combines theoretical models of culture with crowdsourced knowledge (DBpedia). An evaluation of AMOn+ demonstrated its fitness‐for‐purpose regarding domain coverage for annotating culture‐related concepts mentioned in text corpora. This ontology can underpin computational models for making sense of UGC.

The transport network in many countries relies on extended portions which run underground in tunnels. As tunnels age, repairs are required to prevent dangerous collapses. However repairs are expensive and will affect the operational efficiency of the tunnel. We present a decision support system (DSS) based on supervised machine learning methods that learns to predict the risk factor and the resulting repair urgency in the tunnel maintenance planning of a European national rail operator. The data on which the prototype has been built consists of 47 tunnels of varying lengths. For each tunnel, periodic survey inspection data is available for multiple years, as well as other data such as the method of construction of the tunnel. Expert annotations are also available for each 10m tunnel segment for each survey as to the degree of repair urgency which are used for both training and model evaluation. We show that good predictive power can be obtained and discuss the relative merits of a number of learning methods.

Learning by watching videos has become the dominant way of learning for millennials. However, watching videos is a passive form of learning which usually results in a low level of engagement. As the result, video-based learning often results in poor learning outcomes. One of the proven strategies to increase engagement is to integrate interactive activities such as quizzes and assessment problems into videos. Although this strategy increases engagement, it requires changing existing videos and therefore substantial effort from the teacher. We have developed the Active Video Watching (AVW) system that enables the teacher to use existing videos from YouTube without modifications. The teacher is required to define a set of aspects for videos, which serve as reflective scaffolds in order to increase engagement and focus learners’ thinking. AVW provides a Personal Space for individual learners to link their personal experiences while watching videos. The comments collected can be used by the individuals to reflect on their own thoughts or to be shared with other learners in the Social Space. We conducted a study with postgraduate students on presentation skills. The results show that the level of engagement with AVW was high, and that the aspects were effective as reflection prompts. We plan to conduct further studies related to other types of soft skills, and also to further extend AVW to provide individualized feedback to students.

Assessing the Underworld (ATU) is a large interdisciplinary UK research project, which addresses challenges in integrated inter-asset maintenance. As assets on the surface of the ground (e.g. roads or pavements) and those buried under it (e.g. pipes and cables) are supported by the ground, the properties and processes of soil affect the performance of these assets to a significant degree. In order to make integrated decisions, it is necessary to combine the knowledge and expertise in multiple areas, such as roads, soil, buried assets, sensing, etc. This requires an underpinning knowledge model, in the form of an ontology. Within this context, we present a new ontology for describing soil properties (e.g. soil strength) and processes (e.g. soil compaction), as well as how they affect each other. This ontology can be used to express how the ground affects and is affected by assets buried under the ground or on the ground surface. The ontology is written in OWL 2 and openly available from the University of Leeds data repository: http://doi.org/10.5518/54 .

Intercultural competence is an essential 21st Century skill. A key issue for developers of cross-cultural training simulators is the need to provide relevant learning experience adapted to the learnerfis abilities. This paper presents a dialogic approach for a quick assessment of the depth of a learner's current intercultural awareness as part of the EU ImREAL project. To support the dialogue, Linked Data is seen as a rich knowledge base for a diverse range of resources on cultural aspects. This paper investigates how semantic technologies could be used to: (a) extract a pool of concrete culturally-relevant facts from DBpedia that can be linked to various cultural groups and to the learner, (b) model a learner's knowledge on a selected set of cultural themes and (c) provide a novel, adaptive and user-friendly, user modelling dialogue for cultural awareness. The usability and usefulness of the approach is evaluated by CrowdFlower and Expert Inspection.

Social web content such as blogs, videos, and other user‐generated content present a vast source of rich “digital‐traces” of individuals' experiences. The use of digital traces to provide insight into human behavior remains underdeveloped. Recently, ontological approaches have been exploited for tagging and linking digital traces, with progress made in ontology models for well‐defined domains. However, the process of conceptualization for ill‐defined domains remains challenging, requiring interdisciplinary efforts to understand the main aspects and capture them in a computer processable form. The primary contribution of this article is a theory‐driven approach to ontology development that supports semantic augmentation of digital traces. Specifically, we argue that (a) activity theory can be used to develop more insightful conceptual models of ill‐defined activities, which (b) can be used to inform the development of an ontology, and (c) that this ontology can be used to guide the semantic augmentation of digital traces for making sense of phenomena. A case study of interpersonal communication is chosen to illustrate the applicability of the proposed multidisciplinary approach. The benefits of the approach are illustrated through an example application, demonstrating how it may be used to assemble and make sense of digital traces.

This paper presents a model-driven evaluation approach to elicit requirements. This approach has been applied to evaluate a technology-driven prototype designed to help analysts make sense of Internet forums. Due to the complexity of sensemaking and the immature nature of the prototype, we believe a user evaluation informed by a sensemaking model is beneficial for requirement elicitation. From the evaluation, we have revealed how this tool can be improved by providing a set of requirements in the design space of people, activity in context and technology. Our case study illustrates how a cognitive model can be used in an evaluation to elicit requirements for issues with cognitive complexity.

The complexity of ontology authoring and the difficulty to master the use of existing ontology authoring tools, put significant constraints on the involvement of both domain experts and knowledge engineers in ontology authoring. This often requires substantial effort for fixing ontologies defects (e.g. inconsistency, unsatisfiability, missing or unintended implications, redundancy, isolated entities). The paper argues that ontology authoring tools should provide immediate semantic feedback upon entering ontological constructs. We present a framework to analyse input axioms and provide meaningful feedback at a semantic level. The framework has been used to augment an existing Controlled Natural Language-based ontology authoring tool – ROO. An experimental study with ROO has been conducted to examine users' reactions to the semantic feedback and the effect on their ontology authoring behaviour. The study strongly supported responsive intuitive ontology authoring tools, and identified future directions to extend and integrate semantic feedback.

Simulated environments for learning are becoming increasingly popular to support experiential learning in complex domains. A key challenge when designing simulated learning environments is how to align the experience in the simulated world with real world experiences. Social media resources provide user-generated content that is rich in digital traces of real world experiences. People comments, tweets, and blog posts in social spaces can reveal interesting aspects of real world situations or can show what particular group of users is interested in or aware of. This paper examines a systematic way to analyze user-generated content in social media resources to provide useful information for learning simulator design. A hybrid framework exploiting Machine Learning and Semantics for social group profiling is presented. The framework has five stages: (1) Retrieval of user-generated content from the social resource (2) Content noise filtration, removing spam, abuse, and content irrelevant to the learning domain; (3) Deriving individual social profiles for the content authors; (4) Clustering of individuals into groups of similar authors; and (5) Deriving group profiles, where interesting concepts suitable for the use in simulated learning systems are extracted from the aggregated content authored by each group. The framework is applied to derive group profiles by mining user comments on YouTube videos. The application is evaluated in an experimental study within the context of learning interpersonal skills in job interviews. The paper discusses how the YouTube-based group profiles can be used to facilitate the design of a job interview skills learning simulator, considering: (1) identifying learning needs based on digital traces of real world experiences; and (2) augmenting learner models in simulators based on group characteristics derived from social media.

Writing a dissertation is a critical aspect of the learning experience for most university students in all disciplines. It is often accompanied by anxiety and uncertainty, which supervisors often struggle to understand and address. A community-driven approach to support dissertation writing based on semantic social scaffolding is presented here. The paper describes how a semantic wiki was tailored to develop a social writing environment to provide holistic support throughout the whole dissertation process. Based on initial evaluation studies, we discuss the benefits and pitfalls of semantics. The work contributes to a recent research strand that examines how to exploit new social computing technologies to develop effective learning environments.

There is an establishing trend towards the socialization of the web. Virtual communities are becoming very popular web spaces for collaboration and knowledge sharing. However, studies have shown that virtual communities may often be ineffective and may not sustain. We propose a novel approach for community-tailored support which is aimed at facilitating processes important for the community’s effectiveness and sustainability. The paper presents algorithms for detecting evolution patterns of community knowledge sharing behavior. The algorithms are applied to a model of an existing closely-knit community, and used to identify when and what intelligent interventions may be needed to support the functioning of the community as an entity.

An approach of employing information visualisation to develop systems that facilitate instructors in Web-based distance learning is presented here. The paper describes a tool, called CourseVis, that uses multidimensional student tracking data collected by CMS and generates graphical representations that can be used by instructors to gain an understanding of what is happening in distance learning classes. The work followed a systematic approach that started from collecting the instructors' needs, produced some appropriate graphical representations of student tracking data, and evaluated the effectiveness, efficiency and usefulness of the proposed representations. The evaluation has shown that with CourseVis the instructors can identify tendencies in their classes, quickly discover individuals that need special attention, and are able to provide better support to their students.

Diversity of students enrolling on Computing degrees is becoming increasingly important in higher education with the number of mature students noticeably increasing and the expectations for learning and teaching gradually changing. This year, the UK government has issued two policy documents; the first will in uence the make-up of the student body in the future, the second is pushing for a unified e-learning strategy within all education sectors which is driven by user needs and not by the technologies. This paper presents a study of two focus groups in a Computing department in a UK university, and discusses the needs of two diverse student groups, traditional and mature students. It is argued that if e-learning is to be driven by the needs of the users, then diversity should be a driving force behind the use of e-learning technology. Further, we suggest that participatory design would be extremely beneficial in developing effective e-learning.