Search

On-screen game footage contains rich contextual information that players process when playing and experiencing a game. Learning pixel representations of games can benefit artificial intelligence across several downstream tasks including game-playing agents, procedural content generation, and player modelling. The generalizability of these methods, however, remains a challenge, as learned representations should ideally be shared across games with similar game mechanics. This could allow, for instance, game-playing agents trained on one game to perform well in similar games with no re-training. This paper explores how generalizable pre-trained computer vision encoders can be for such tasks, by decomposing the latent space into content embeddings and style embeddings. The goal is to minimize the domain gap between games of the same genre when it comes to game content critical for downstream tasks, and ignore differences in graphical style. We employ a pre-trained Vision Transformer encoder and a decomposition technique based on game genres to obtain separate content and style embeddings. Our findings show that the decomposed embeddings achieve style invariance across multiple games while still maintaining strong content extraction capabilities. We argue that the proposed decomposition of content and style offers better generalization capacities across game environments independently of the downstream task.

This paper introduces a user-driven evolutionary algorithm based on Quality Diversity (QD) search. During a design session, the user iteratively selects among presented alternatives and their selections affect the upcoming results. We aim to address two major concerns of interactive evolution: (a) the user must be presented with few alternatives, to reduce cognitive load; (b) presented alternatives should be diverse but similar to the previous user selection, to reduce user fatigue. To address these concerns, we implement a variation of the MAP-Elites algorithm where the presented alternatives are sampled from a small region (window) of the behavioral space. After a user selection, the window is centered on the selected individual's behavior characterization, evolution selects parents from within this window to produce offspring, and new alternatives are sampled. Essentially we define an adaptive system of local QD, where the user's selections guide the search towards specific regions of the behavioral space. The system is tested on the generation of architectural layouts, a constrained optimization task, leveraging QD through a two-archive approach. Results show that while global exploration is not as pronounced as in MAP-Elites, the system finds more appropriate solutions to the user's taste, based on experiments with controllable artificial users., Parts of this manuscripts are published at The Genetic and Evolutionary Computation Conference (GECCO) 2023

Architectural design is a highly complex practice that involves a wide diversity of disciplines, technologies, proprietary design software, expertise, and an almost infinite number of constraints, across a vast array of design tasks. Enabling intuitive, accessible, and scalable design processes is an important step towards performance-driven and sustainable design for all. To that end, we introduce Architext, a novel semantic generation assistive tool. Architext enables design generation with only natural language prompts, given to large-scale Language Models, as input. We conduct a thorough quantitative evaluation of Architext's downstream task performance, focusing on semantic accuracy and diversity for a number of pre-trained language models ranging from 120 million to 6 billion parameters. Architext models are able to learn the specific design task, generating valid residential layouts at a near 100% rate. Accuracy shows great improvement when scaling the models, with the largest model (GPT-J) yielding impressive accuracy ranging between 25% to over 80% for different prompt categories. We open source the finetuned Architext models and our synthetic dataset, hoping to inspire experimentation in this exciting area of design research., Comment: 21 pages

Affective computing strives to unveil the unknown relationship between affect elicitation, manifestation of affect and affect annotations. The ground truth of affect, however, is predominately attributed to the affect labels which inadvertently include biases inherent to the subjective nature of emotion and its labeling. The response to such limitations is usually augmenting the dataset with more annotations per data point; however, this is not possible when we are interested in self-reports via first-person annotation. Moreover, outlier detection methods based on inter-annotator agreement only consider the annotations themselves and ignore the context and the corresponding affect manifestation. This paper reframes the ways one may obtain a reliable ground truth of affect by transferring aspects of causation theory to affective computing. In particular, we assume that the ground truth of affect can be found in the causal relationships between elicitation, manifestation and annotation that remain \emph{invariant} across tasks and participants. To test our assumption we employ causation inspired methods for detecting outliers in affective corpora and building affect models that are robust across participants and tasks. We validate our methodology within the domain of digital games, with experimental results showing that it can successfully detect outliers and boost the accuracy of affect models. To the best of our knowledge, this study presents the first attempt to integrate causation tools in affective computing, making a crucial and decisive step towards general affect modeling., 8 pages, accepted for publication and presentation at 2022 10th International Conference on Affective Computing and Intelligent Interaction Workshops and Demos (ACIIW)

Using artificial intelligence (AI) to automatically test a game remains a critical challenge for the development of richer and more complex game worlds and for the advancement of AI at large. One of the most promising methods for achieving that long-standing goal is the use of generative AI agents, namely procedural personas, that attempt to imitate particular playing behaviors which are represented as rules, rewards, or human demonstrations. All research efforts for building those generative agents, however, have focused solely on playing behavior which is arguably a narrow perspective of what a player actually does in a game. Motivated by this gap in the existing state of the art, in this paper we extend the notion of behavioral procedural personas to cater for player experience, thus examining generative agents that can both behave and experience their game as humans would. For that purpose, we employ the Go- Explore reinforcement learning paradigm for training human-like procedural personas, and we test our method on behavior and experience demonstrations of more than 100 players of a racing game. Our findings suggest that the generated agents exhibit distinctive play styles and experience responses of the human personas they were designed to imitate. Importantly, it also appears that experience, which is tied to playing behavior, can be a highly informative driver for better behavioral exploration., peer-reviewed

Serious games often employ a scripted dialogue for player interaction with a virtual character. In our serious game Communicate, a domain expert develops a structured, scripted scenario as a sequence of potential interactions in an authoring tool. A player is often a student learning communication skills and a virtual character represents a person that a student talks to. In the original version of Communicate, a player `converses' with a virtual character by clicking on one of the multiple statement options. Since 2018, we perform blended learning sessions for final year computer science students using Communicate. Our goal is to improve these sessions and in this paper, we apply the action research method over three semesters to iteratively improve these blended learning sessions. In the first semester, our baseline, we conduct sessions where students play a scenario in multiple choice format. In the second semester, we enhance Communicate by enabling a student to enter open text input in an improved scenario. In the third semester, we enhance a session by incorporating peer teaching. Students fill in an evaluation survey after a session and we compare the evaluation of students from the three semesters. Results show that student ratings are significantly higher in sessions incorporating peer teaching compared to the baseline.

Stealth assessment is a principled assessment methodology proposed for serious games that uses statistical models and machine learning technology to infer players’ mastery levels from logged gameplay data. Although stealth assessment has been proven to be valid and reliable, its application is complex, laborious, and time-consuming. A generic stealth assessment tool (GSAT), proven for its robustness with simulation data, has been proposed to resolve these issues. In this study, GSAT’s robustness is further investigated by using real-world data collected from a serious game on personality traits and validated with an associated personality questionnaire (NEO PI-R). To achieve this, (a) a stepwise regression approach was followed for generating statistical models from logged data for the big five personality traits (OCEAN model), (b) the statistical models are then used with GSAT to produce inferences regarding learners’ mastery level on these personality traits, and (c) the validity of GSAT’s outcomes are examined through a correlation analysis using the results of the NEO PI-R questionnaire. Despite the small dataset GSAT was capable of making inferences on players’ personality traits. This study has demonstrated the practicable feasibility of the SA methodology with GSAT and provides a showcase for its wider application in serious games.

This paper proposes a paradigm shift for affective computing by viewing the affect modeling task as a reinforcement learning process. According to our proposed framework the context (environment) and the actions of an agent define the common representation that interweaves behavior and affect. To realise this framework we build on recent advances in reinforcement learning and use a modified version of the Go-Explore algorithm which has showcased supreme performance in hard exploration tasks. In this initial study, we test our framework in an arcade game by training Go-Explore agents to both play optimally and attempt to mimic human demonstrations of arousal. We vary the degree of importance between optimal play and arousal imitation and create agents that can effectively display a palette of affect and behavioral patterns. Our Go-Explore implementation not only introduces a new paradigm for affect modeling; it empowers believable AI-based game testing by providing agents that can blend and express a multitude of behavioral and affective patterns., Comment: 8 pages

This paper argues that generative art driven by conformance to a visual and/or semantic corpus lacks the necessary criteria to be considered creative. Among several issues identified in the literature, we focus on the fact that generative adversarial networks (GANs) that create a single image, in a vacuum, lack a concept of novelty regarding how their product differs from previously created ones. We envision that an algorithm that combines the novelty preservation mechanisms in evolutionary algorithms with the power of GANs can deliberately guide its creative process towards output that is both good and novel. In this paper, we use recent advances in image generation based on semantic prompts using OpenAI’s CLIP model, interrupting the GAN’s iterative process with short cycles of evolutionary divergent search. The results of evolution are then used to continue the GAN’s iterative process; we hypothesise that this intervention will lead to more novel outputs. Testing our hypothesis using novelty search with local competition, a quality-diversity evolutionary algorithm that can increase visual diversity while maintaining quality in the form of adherence to the semantic prompt, we explore how different notions of visual diversity can affect both the process and the product of the algorithm. Results show that even a simplistic measure of visual diversity can help counter a drift towards similar images caused by the GAN. This first experiment opens a new direction for introducing higher intentionality and a more nuanced drive for GANs., peer-reviewed

Having access to accurate game state information is of utmost importance for any artificial intelligence task including game-playing, testing, player modeling, and procedural content generation. Self- Supervised Learning (SSL) techniques have shown to be capable of inferring accurate game state information from the high dimensional pixel input of game footage into compressed latent representations. Contrastive Learning is a popular SSL paradigm where the visual understanding of the game’s images comes from contrasting dissimilar and similar game states defined by simple image augmentation methods. In this study, we introduce a new game scene augmentation technique—named GameCLR—that takes advantage of the game-engine to define and synthesize specific, highly controlled renderings of different game states, thereby, boosting contrastive learning performance. We test our GameCLR technique on images of the CARLA driving simulator environment and compare it against the popular SimCLR baseline SSL method. Our results suggest that GameCLR can infer the game’s state information from game footage more accurately compared to the baseline. Our proposed approach allows us to conduct game artificial intelligence research by directly utilizing screen pixels as input., peer-reviewed

Normalization is a vital process for any machine learning task as it controls the properties of data and affects model performance at large. The impact of particular forms of normalization, however, has so far been investigated in limited domain-specific classification tasks and not in a general fashion. Motivated by the lack of such a comprehensive study, in this paper we investigate the performance of lp-constrained softmax loss classifiers across different norm orders, magnitudes, and data dimensions in both proof-of-concept classification problems and real-world popular image classification tasks. Experimental results suggest collectively that lp-constrained softmax loss classifiers not only can achieve more accurate classification results but, at the same time, appear to be less prone to overfitting. The core findings hold across the three popular deep learning architectures tested and eight datasets examined, and suggest that lp normalization is a recommended data representation practice for image classification in terms of performance and convergence, and against overfitting.

This paper presents a generative architecture for general-purpose room layouts that can be treated as geometric definitions of dungeons, mansions, shooter levels and more. The motivation behind this work is to provide a design tool for virtual environments that combines aspects of controllability, expressivity and generality. Towards that end, a two-tier level representation is realized, with a graph-based design specification constraining and guiding the generated geometries, facilitated by constrained evolutionary search. Expressivity is secured through quality-diversity search which can provide the designer with a broad variety of level layouts to choose from. Finally, the generator is general-purpose as it can produce layouts based on different types of static grid structures or as freeform, curved structures through an adaptive Voronoi diagram that is evolved along with the level itself. The method is tested on a variety of design specifications and grid types, and results show that even with complex design constraints or malleable grids the algorithm can produce a broad variety of levels., peer-reviewed

Affect modeling is viewed, traditionally, as the process of mapping measurable affect manifestations from multiple modalities of user input to affect labels. That mapping is usually inferred through endto- end (manifestation-to-affect) machine learning processes. What if, instead, one trains general, subject-invariant representations that consider affect information and then uses such representations to model affect? In this paper we assume that affect labels form an integral part, and not just the training signal, of an affect representation and we explore how the recent paradigm of contrastive learning can be employed to discover general high-level affect-infused representations for the purpose of modeling affect.We introduce three different supervised contrastive learning approaches for training representations that consider affect information. In this initial study we test the proposed methods for arousal prediction in the RECOLA dataset based on user information from multiple modalities. Results demonstrate the representation capacity of contrastive learning and its efficiency in boosting the accuracy of affect models. Beyond their evidenced higher performance compared to end-to-end arousal classification, the resulting representations are general purpose and subject-agnostic, as training is guided though general affect information available in any multimodal corpus., peer-reviewed

This paper introduces a paradigm shift by viewing the task of affect modeling as a reinforcement learning (RL) process. According to the proposed paradigm, RL agents learn a policy (i.e. affective interaction) by attempting to maximize a set of rewards (i.e. behavioral and affective patterns) via their experience with their environment (i.e. context). Our hypothesis is that RL is an effective paradigm for interweaving affect elicitation and manifestation with behavioral and affective demonstrations. Importantly, our second hypothesis—building on Damasio’s somatic marker hypothesis—is that emotion can be the facilitator of decision-making. We test our hypotheses in a racing game by training Go-Blend agents to model human demonstrations of arousal and behavior; Go-Blend is a modified version of the Go-Explore algorithm which has recently showcased supreme performance in hard exploration tasks. We first vary the arousalbased reward function and observe agents that can effectively display a palette of affect and behavioral patterns according to the specified reward. Then we use arousal-based state selection mechanisms in order to bias the strategies that Go-Blend explores. Our findings suggest that Go-Blend not only is an efficient affect modeling paradigm but, more importantly, affect-driven RL improves exploration and yields higher performing agents, validating Damasio’s hypothesis in the domain of games., peer-reviewed

To which degree can abstract gameplay metrics capture the player experience in a general fashion within a game genre? In this comprehensive study we address this question across three different videogame genres: racing, shooter, and platformer games. Using high-level gameplay features that feed preference learning models we are able to predict arousal accurately across different games of the same genre in a large-scale dataset of over 1,000 arousal-annotated play sessions. Our genre models predict changes in arousal with up to 74% accuracy on average across all genres and 86% in the best cases. We also examine the feature importance during the modelling process and find that time-related features largely contribute to the performance of both game and genre models. The prominence of these game-agnostic features show the importance of the temporal dynamics of the play experience in modelling, but also highlight some of the challenges for the future of general affect modelling in games and beyond., Version accepted for IEEE Conference on Games 2021

This paper introduces a novel method for generating artistic images that express particular affective states. Leveraging state-of-the-art deep learning methods for visual generation (through generative adversarial networks), semantic models from OpenAI, and the annotated dataset of the visual art encyclopedia WikiArt, our AffectGAN model is able to generate images based on specific or broad semantic prompts and intended affective outcomes. A small dataset of 32 images generated by AffectGAN is annotated by 50 participants in terms of the particular emotion they elicit, as well as their quality and novelty. Results show that for most instances the intended emotion used as a prompt for image generation matches the participants' responses. This small-scale study brings forth a new vision towards blending affective computing with computational creativity, enabling generative systems with intentionality in terms of the emotions they wish their output to elicit., Comment: Published in the "What's Next in Affect Modeling?" workshop at the Affective Computing & Intelligent Interaction (ACII) 2021 conference, 7 pages, 3 figures

Search-based procedural content generation methods have recently been introduced for the autonomous creation of bullet hell games. Search-based methods, however, can hardly model patterns of danmakus -- the bullet hell shooting entity -- explicitly and the resulting levels often look non-realistic. In this paper, we present a novel bullet hell game platform named Keiki, which allows the representation of danmakus as a parametric sequence which, in turn, can model the sequential behaviours of danmakus. We employ three types of generative adversarial networks (GANs) and test Keiki across three metrics designed to quantify the quality of the generated danmakus. The time-series GAN and periodic spatial GAN show different yet competitive performance in terms of the evaluation metrics adopted, their deviation from human-designed danmakus, and the diversity of generated danmakus. The preliminary experimental studies presented here showcase that potential of time-series GANs for sequential content generation in games., This paper is accepted by the 2021 IEEE Conference on Games

Serious games, as well as entertainment games, often employ a scripted dialogue for player interaction with a virtual character. In our serious game Communicate, a domain expert develops a structured, scripted scenario as a sequence of potential interactions in an authoring tool. Communicate is widely used and several domain experts have already developed over a thousand scenarios. In the original version of Communicate, a student `navigates' a dialogue with a virtual character by clicking one of the multiple statement options at a step of a scenario. Open text response often requires more complex thinking from a student. In this paper we explore ways to handle open text input from a student at a step of a scenario. Our goal is to match open text to scripted statements using a Natural Language Processing (NLP) method and explore mechanisms to handle matched and unmatched input.

This deliverable, grounded in a review of the state-of-the-art in the pertinent areas, aims tomap out the parametric space thatartificially intelligent algorithms can explore, as well asmathematical formulas for measuring constraint satisfaction, efficiency, and diversity in the different design disciplines contributing to PrismArch. The document surveys how current approaches in AI for architecture define representations, functional objectives and constraints, and identifies formulas, code libraries, and software that can be used for the purpose of AI-assisted content creation in PrismArch. This document takes input from the user and functional requirements of PrismArch described in D1.1 (“Report on current limitations of AEC software tools, leading to user and functional requirements of PrismArch”).

This paper introduces a surrogate model of gameplay that learns the mapping between different game facets, and applies it to a generative system which designs new content in one of these facets. Focusing on the shooter game genre, the paper explores how deep learning can help build a model which combines the game level structure and the game's character class parameters as input and the gameplay outcomes as output. The model is trained on a large corpus of game data from simulations with artificial agents in random sets of levels and class parameters. The model is then used to generate classes for specific levels and for a desired game outcome, such as balanced matches of short duration. Findings in this paper show that the system can be expressive and can generate classes for both computer generated and human authored levels., Proceedings of the FDG Workshop on Procedural Content Generation, 2018, 10 pages

Representing games through their pixels offers a promising approach for building general-purpose and versatile game models. While games are not merely images, neural network models trained on game pixels often capture differences of the visual style of the image rather than the content of the game. As a result, such models cannot generalize well even within similar games of the same genre. In this paper we build on recent advances in contrastive learning and showcase its benefits for representation learning in games. Learning to contrast images of games not only classifies games in a more efficient manner; it also yields models that separate games in a more meaningful fashion by ignoring the visual style and focusing, instead, on their content. Our results in a large dataset of sports video games containing 100k images across 175 games and 10 game genres suggest that contrastive learning is better suited for learning generalized game representations compared to conventional supervised learning. The findings of this study bring us closer to universal visual encoders for games that can be reused across previously unseen games without requiring retraining or fine-tuning., Comment: 8 pages, 7 figures, CoG

The advancement of artificial intelligence (AI) and its increased use in everyday life have exacerbated the need to understand its underlying processes, and to raise awareness about its shortcomings and faults. Consequently there has been an increased effort to teach basic concepts of AI and machine learning (ML) from an early age. Digital games have been shown to be effective as teaching and learning tools, and there are ongoing efforts to increase AI literacy through educational digital games. To this effect, this work describes the process followed to extrapolate the design of such an educational game directly from the pedagogical needs emerging from stakeholders. Seven focus groups and workshops were conducted in Greece, Malta and Norway, with 55 teachers, researchers, practitioners and policy-makers, and 22 primary and secondary education students in total. These workshops identified seven goals which informed the design of a game for AI literacy called ArtBot. The game design process and the final interface and gameplay loop of ArtBot are explained in relation to these goals. The game was subsequently deployed across a variety of platforms, to enable dissemination to a broad audience in classrooms and elsewhere. The game was part of the tools developed in the framework of the LearnML Erasmus+ project. A preliminary online survey was used to gauge how well the game was received by teachers and students, with an overall positive result. A longer-term data collection of the game usage statistics has been initiated and will be analysed over the course of the game dissemination., peer-reviewed

This research is supported by the IEEE CIS Graduate Student Research Grants and the EP/L015846/1 for the Centre for Doctoral Training in Intelligent Games and Game Intelligence(IGGI) from the UK Engineering and Physical Sciences Research Council (EPSRC)., Assessing the believability of agents, characters and simulated actors is a core challenge for human computer interaction. While numerous approaches are suggested in the literature, they are all limited to discrete and low-granularity representations of believable behavior. In this paper we view believability, for the first time, as a time-continuous phenomenon and we explore the suitability of two different affect annotation schemes for its assessment. In particular, we study the degree to which we can predict character believability in a continuous fashion through a two-player game study. The game features various opponent behaviors that are assessed for their believability by 89 participants that played the game and then annotated their recorded playthrough. Random forest models are then trained to predict believability based on ad-hoc designed in-game features. Results suggest that a discrete annotation method leads to a more robust assessment of the ground truth and subsequently better modelling performance. Our best models are able to predict a change in perceived believability with a 72.5% accuracy on average (up to 90% in the best cases) in a time-continuous manner., peer-reviewed

This project has received funding from the EU’s Horizon 2020 programme under grant agreement No 951911, and from the University of Malta internal research grants programme Research Excellence Fund under grant agreement No 202003., To which degree can abstract gameplay metrics capture the player experience in a general fashion within a game genre? In this comprehensive study we address this question across three different videogame genres: racing, shooter, and platformer games. Using high-level gameplay features that feed preference learning models we are able to predict arousal accurately across different games of the same genre in a largescale dataset of over 1, 000 arousal-annotated play sessions. Our genre models predict changes in arousal with up to 74% accuracy on average across all genres and 86% in the best cases. We also examine the feature importance during the modelling process and find that time-related features largely contribute to the performance of both game and genre models. The prominence of these game-agnostic features show the importance of the temporal dynamics of the play experience in modelling, but also highlight some of the challenges for the future of general affect modelling in games and beyond., peer-reviewed

The research leading to these results has received funding from the European Union H2020 Horizon Programme (2014-2020) under grant agreement 952002, project PrismArch: Virtual reality aided design blending crossdisciplinary aspects of architecture in a multi-simulation environment., How does the form of our surroundings impact the ways we feel? This paper extends the body of research on the effects that space and light have on emotion by focusing on critical features of architectural form and illumination colors and their spatiotemporal impact on arousal. For that purpose, we solicited a corpus of spatial transitions in video form, lasting over 60 minutes, annotated by three participants in terms of arousal in a time-continuous and unbounded fashion. We process the annotation traces of that corpus in a relative fashion, focusing on the direction of arousal changes (increasing or decreasing) as affected by changes between consecutive rooms. Results show that properties of the form such as curved or complex spaces align highly with increased arousal. The analysis presented in this paper sheds some initial light in the relationship between arousal and core spatiotemporal features of form that is of particular importance for the affect-driven design of architectural spaces., peer-reviewed

With the ubiquitous role of Artificial Intelligence (AI) in everyday applications such as smartphones and social media, children need digital literacy skills to navigate the digital world, critically view, and reflect on the social and ethical implications of the design and architecture of AI systems. To address this increasing need for AI literacy skills, particularly for younger students, this paper presents the rationale of the LearnML project which aims to develop a framework and game-based educational material for promoting AI literacy among primary and secondary education students. We also describe the design and initial assessment of the game “ArtBot”, developed as part of the LearnML project. We review existing literature, discuss the educational game design and development of “ArtBot” and describe the initial feedback of students and teachers. Our goal is to provide insights and suggest guidelines for the implementation of game-based learning environments for supporting AI literacy skills of students., peer-reviewed

Search-based procedural content generation methods have recently been introduced for the autonomous creation of bullet hell games. Search-based methods, however, can hardly model patterns of danmakus—the bullet hell shooting entity— explicitly and the resulting levels often look non-realistic. In this paper, we present a novel bullet hell game platform named Keiki, which allows the representation of danmakus as a parametric sequence which, in turn, can model the sequential behaviours of danmakus. We employ three types of generative adversarial networks (GANs) and test Keiki across three metrics designed to quantify the quality of the generated danmakus. The time-series GAN and periodic spatial GAN show different yet competitive performance in terms of the evaluation metrics adopted, their deviation from human-designed danmakus, and the diversity of generated danmakus. The preliminary experimental studies presented here showcase that potential of time-series GANs for sequential content generation in games., peer-reviewed

Gamification aims at addressing inherent problems of massive open online courses (MOOC): high dropouts, lack of engagement, isolation, lack of individualization. However, each MOOC platform offers different features and technical interfaces. Also, each platform collects different sets of data about user interaction, learning progress, or completion and success rates. This is an obstacle to the theoretically sound application of gamification in a vendor independent way and to the evaluation of the impact of gamification. We define our understanding of meaningful gamification, introduce requirements for platform-independent gamification, present the resulting Gamifire infrastructure, and describe application cases. We also point out planned development activities.

As interactive narratives, by definition, change according to the user's choices (dynamic story), so do the dialogue utterances by the characters. Writing all possible utterances manually faces scaling problems. This motivates the use of natural language generation techniques. We present ExpReal, a surface realizer and templating language that allows authors of interactive narratives to write flexible and enriched templates while maintaining control over their use. Templates are automatically selected based on author-specified conditions relating to the world state (e.g. characters' emotions) or the current task at hand. ExpReal has been developed to support at least three languages (English, French and Dutch).

Is it possible to predict moment-to-moment gameplay engagement based solely on game telemetry? Can we reveal engaging moments of gameplay by observing the way the viewers of the game behave? To address these questions in this paper, we reframe the way gameplay engagement is defined and we view it, instead, through the eyes of a game's live audience. We build prediction models for viewers' engagement based on data collected from the popular battle royale game PlayerUnknown's Battlegrounds as obtained from the Twitch streaming service. In particular, we collect viewers' chat logs and in-game telemetry data from several hundred matches of five popular streamers (containing over 100,000 game events) and machine learn the mapping between gameplay and viewer chat frequency during play, using small neural network architectures. Our key findings showcase that engagement models trained solely on 40 gameplay features can reach accuracies of up to 80% on average and 84% at best. Our models are scalable and generalisable as they perform equally well within- and across-streamers, as well as across streamer play styles., Comment: Version accepted for the Conference on the Foundations of Digital Games 2020 - Malta

In this paper, we study the effect of time pressure on player behaviour during a dilemma-based crisis management game. We employ in-game action tracking, physiological sensor data and self-reporting in order to create multi-modal predictive models of player stress responses during a crisis management scenario. We were able to predict the experimental condition (time pressure vs. no time pressure) with 84.5% accuracy, using a game-only feature set. However, lower accuracy was observed when physiological sensor data was used for the same task. The method presented in this paper can be employed in crisis management training, aiming at assessing players' responses to stressful conditions and manipulating player stress levels to provide personalised training scenarios.

In this paper, we study the effect of time pressure on player behaviour during a dilemma-based crisis management game. We employ in-game action tracking, physiological sensor data and self-reporting in order to create multi-modal predictive models of player stress responses during a crisis management scenario. We were able to predict the experimental condition (time pressure vs. no time pressure) with 84.5% accuracy, using a game-only feature set. However, lower accuracy was observed when physiological sensor data was used for the same task. The method presented in this paper can be employed in crisis management training, aiming at assessing players’ responses to stressful conditions and manipulating player stress levels to provide personalised training scenarios.

It is a great pleasure for us to introduce this issue of the IJSG, which is dedicated to the Games and Learning Alliance Conference (GaLA Conf) that was held in Athens, November 27-29, 2019. Almost 70 participants converged in this beautiful, historic city in order to share knowledge and experiences related to serious games and gamification, their techniques and their application. A number of the best conference papers were selected, and the authors were invited to extend their paper with at least one-third new content, and to submit their paper to IJSG for a regular peer-review process.

Learning to read and write is a fundamental right and a necessary skill for the personal, cultural, and economic development of people and their societies. However, children of developing countries, such as sub-Saharan areas, are currently at a greater risk of illiteracy. The current penetration of mobile technologies and the internet in sub-Saharan rural areas, however, offers a unique opportunity for tackling the challenge of literacy at a large scale. Motivated by the current shortage of preschool teachers for training handwriting in a personalised manner, this paper discusses the design of Djehuty, an educational gamified environment for preschoolers. Djehuty is equipped with an artificial intelligence module which generates a style of handwriting and suggests handwriting paths to the child in a mixed-initiative manner. The paper presents the key elements of the game prototype., peer-reviewed

Performance of newly-formed project teams is often limited, or at least delayed, when team members refrain from sharing their ideas due to unfamiliarity with their peers. A variety of ice-breaking methods can help overcome this cold start, but mostly they need to be deployed and moderated by experienced facilitators. This setup is rarely an option for most undergrad project courses at university level, typically carried out in small teams. In order to help breaking the ice in this context, we developed Maze Maestro, a collaborative tabletop game in which the board is made up by attaching the displays of the team members' mobile phones to form a large maze. Each member controls a character in the maze, and the whole team has the common goal of leaving the maze together; however, this is only possible with timely communication and much cooperation. While playing, team members are encouraged to confer possible plans and share their ideas, which is the fertile ground for breaking the ice. Play testing has shown that Maze Maestro was perceived as a fun and original collaborative game. So far, results of a preliminary user study are optimistic about the ability of Maze Maestro to break the ice within newly-formed teams, without requiring any facilitator.

Affective interaction between players of video games can elicit rich and varying patterns of emotions. In multiplayer activities that take place in a common space (such as sports and board games), players are generally aware of the emotions of their teammates or opponents as they can directly observe their behavioral patterns, facial expressions, head pose, body stance and so on. Players of online video games, however, are not generally aware of the other players' emotions given the limited channels of direct interaction among them (e.g. via emojis or chat boxes). It also turns out that the impact of real-time emotionawareness on play is still unexplored in the space of online digital games. Motivated by this lack of empirical knowledge on the role of the affect of others to one's gameplay performance in this paper we investigate the degrees to which the expression of manifested emotions of an opponent can affect the emotions of the player and consequently his gameplay behavior. In this initial study, we test our hypothesis on a two-player adversarial car racing game. We perform a comprehensive user study to evaluate the emotions, behaviors, and attitudes of players in emotion aware versus emotion agnostic game versions. Our findings suggest that expressing the emotional state of the opponent through an emoji in real-time affects the emotional state and behavior of players that can consequently affect their playing experience., Comment: Accepted and to appear in IEEE Transactions on Games

In this paper we explore the possibilities to combine two research methods we regard as being very useful when interacting with stakeholders in complex systems. We discuss a mixed research methods approach, based on the Q methodology and a simulation game. In a game design process, translating the real or reference system into the game design is an intricate process and rather challenging due to the complexity of today’s societal systems. As shown by various studies, different data techniques are proposed in order to translate reality aspects. One of the proposed data gathering techniques in combination with simulation games is Q methodology. Q methodology is a suitable method to retrieve social perspectives of stakeholders on a particular topic. Yet it is still elusive how the results of a Q methodology can be used in a game design process. In this paper, we explore the possibilities how to combine the two methods and how to translate the results of the Q analysis into a game design concept. In the context of a case within the domain of transport and logistics, we discuss how such mixed research methods approach could look like. We conclude with a future outlook on our research.

Persuasive games for societal interventions aim to shape, reinforceor change players’ attitudes and behavior to help solving complex societalissues. In earlier work, we explored how persuasive game mechanics maycontribute to the formation of attitudes in persuasive games. As a follow-up, this paper presents four design recommendations that could increase the chance that these attitudes will actually lead to the desired behavior shown by players after the game: viz., these attitudes require the right conditions to become a predictor of the desired, post-game behavior.In order to arrive at these recommendations we looked at relevant work from the field of social psychology. Next we linked our insights to the context of persuasive game design. This yielded four conceptual design recommendations for maximizing the likelihood for an attitude influenced by a persuasive game to result in the desired behavior in the real world;1. aligning the degree of specification of a game’s message and the desired behavior2. emphasizing the function of the attitude to be influenced3. enabling players to reflect on their internal states4. emphasizing personal relevance of an attitude to a behavioral choiceSo far, these recommendations are still theoretical in nature. We therefore discuss how future work should empirically examine these, including their implications for the effective use of persuasive game mechanics.

Is it possible to predict the motivation of players just by observing their gameplay data? Even if so, how should we measure motivation in the first place? To address the above questions, on the one end, we collect a large dataset of gameplay data from players of the popular game Tom Clancy's The Division. On the other end, we ask them to report their levels of competence, autonomy, relatedness and presence using the Ubisoft Perceived Experience Questionnaire. After processing the survey responses in an ordinal fashion we employ preference learning methods based on support vector machines to infer the mapping between gameplay and the reported four motivation factors. Our key findings suggest that gameplay features are strong predictors of player motivation as the best obtained models reach accuracies of near certainty, from 92% up to 94% on unseen players., Comment: Version accepted for IEEE Conference on Games, 2019

This book constitutes the refereed proceedings of the 8th International Conference on Games and Learning Alliance, GALA 2019, held in Athens, Greece, in November 2019. The 38 regular papers presented together with 19 poster papers were carefully reviewed and selected from 76 submissions. The papers cover the following topics: serious game design and pedagogical foundations; AI and technology for SG; gamification; applications and case studies; and posters.

Unlike classical physics, quantum physics is harder to explain, as it involves very small scales and phenomena that are not visible to the naked eye. Understanding the differences between classical and quantum physics is difficult, especially for children, who cannot grasp the subtleties conveyed in complicated formulae.We propose to achieve this in a playful and immersive manner, which is a more familiar and convenient way to introduce children to new concepts. For this we developed Save Schrödinger’s Cat, a puzzle game in virtual reality featuring a classical physics mode and a quantum physics mode. As virtual objects and phenomena behave differently in each mode, this mechanic encourages players to toggle between modes, in order to explore the differences between quantum and classical physics in an immersive, entertaining and challenging way. A preliminary evaluation showed that players could better identify various distinguishing features of either mode.

Serious games, as well as entertainment games, often employ a scripted dialogue for player interaction with a virtual character. In our serious game Communicate, a domain expert develops a structured, scripted scenario as a sequence of potential interactions in an authoring tool. Communicate is widely used and several domain experts have already developed over a thousand scenarios. In the original version of Communicate, a student `navigates' a dialogue with a virtual character by clicking one of the multiple statement options at a step of a scenario. Open text response often requires more complex thinking from a student. In this paper we explore ways to handle open text input from a student at a step of a scenario. Our goal is to match open text to scripted statements using a Natural Language Processing (NLP) method and explore mechanisms to handle matched and unmatched input.

While visual impairment is relatively common, most sighted people have no idea of what it is like to live without one of the most heavily utilised senses. We developed the game Loud and Clear in order to have them experience the difficulties of being visually impaired, as well as to put in evidence the abilities blind people have developed, which sighted people mostly lack. In this game without visuals, the player has to rely solely on audio to complete objectives within the game. The game consists of a number of puzzle rooms the player has to solve. These puzzles illustrate the challenges of being blind in a playful setting, and challenge the player to use different auditory skills that are key to achieving objectives without vision, such as sound localisation, sound recognition and spatial orientation. The game uses audio spatialisation techniques to give the player a realistic and immersive auditive experience. Preliminary tests of this game show that players acknowledge the initial high difficulty of ‘living’ as a blind person, to which eventually they were able to somehow adapt. In addition, players reported feeling both immersed and educated by the experience.

Burning fossil fuels is a big part of our heat production. Since this process is both non-renewable and polluting, finding other options is important. A clean and underutilized alternative is geothermal energy. However, it is often not considered due to sheer ignorance or misconceptions. HotPipe is a serious game designed to alleviate these problems, particularly among youth populations. Players control a drill to create geothermal wells solving a variety of puzzles, which introduce relevant cases for geothermal heating and show what geothermal wells are made of. The game focuses primarily on conveying the concepts of water circulation, relation between temperature and depth, androck type proprieties. From our game evaluation, players revealed a solid improvement on their geothermal energy knowledge.

In order to support the creation of sustainable and healthy ecosystems, citizens should have knowledge of the necessary maintenance needed. For example, they should be aware of the challenges of maintaining proper urban surface waters, so that they can take on a responsible and proactive role. Ideally, citizens should acquire this knowledge from an early age. We describe the design, implementation, and evaluation of the serious game Hydro Hero, aimed at forming and reinforcing this awareness and reasoning. Hydro Hero is an infinite runner game with extended minigames, which show both what should not be on the canals as well as why they should be removed. We assessed the ability of Hydro Hero to teach young children about canal maintenance by combining a questionnaire with open-ended play-testing sessions at a science museum. On average, players were able to improve their ability to justify why certain items should be removed or left in a canal. It has not been fully confirmed that Hydro Hero conveyed this specific knowledge, due to the absence of quizzing beforehand. However, given the high participants’ engagement, we consider the game to have contributed to their awareness about the importance of canal cleaning for a sustainable urban environment.

The question of representing emotion computationally remains largely unanswered: popular approaches require annotators to assign a magnitude (or a class) of some emotional dimension, while an alternative is to focus on the relationship between two or more options.Recent evidence in affective computing suggests that following a methodology of ordinal annotations and processing leads to better reliability and validity of the model. This pa-per compares the generality of classification methods versus preference learning methods in predicting the levels of arousal in two widely used affective datasets. Findings of this initial study further validate the hypothesis that approaching affect labels as ordinal data and building models via preference learning yields models of better validity.

This paper focuses on the design and implementation of a tool that allows educators to author 3D virtual labs. The methodology followed is based on web 3D frameworks such as three.js and WordPress that allowed us to develop simplified interfaces for modifying Unity3D templates. Two types of templates namely one for Chemistry and one for Wind Energy labs were developed that allow to test the generalization, user-friendliness and usefulness of such an approach. Results have shown that educators are much interested on the general concept, but several improvements should be made towards the user-friendliness and the intuitiveness of the interfaces in order to allow the inexperienced educators in 3D gaming to make such an attempt., peer-reviewed

Traditional evolutionary algorithms tend to converge to a single good solution, which can limit their chance of discovering more diverse and creative outcomes. Divergent search, on the other hand, aims to counter convergence to local optima by avoiding selection pressure towards the objective. Forms of divergent search such as novelty or surprise search have proven to be beneficial for both the efficiency and the variety of the solutions obtained in deceptive tasks. Importantly for this paper, early results in maze navigation have shown that combining novelty and surprise search yields an even more effective search strategy due to their orthogonal nature. Motivated by the largely unexplored potential of coupling novelty and surprise as a search strategy, in this paper we investigate how fusing the two can affect the evolution of soft robot morphologies. We test the capacity of the combined search strategy against objective, novelty, and surprise search, by comparing their efficiency and robustness, and the variety of robots they evolve. Our key results demonstrate that novelty-surprise search is generally more efficient and robust across eight different resolutions. Further, surprise search explores the space of robot morphologies more broadly than any other algorithm examined., peer-reviewed

Affective computing is a multidisciplinary field that studies the various ways by which computational processes are able to elicit, sense, and detect manifestations of human emotion. While the methods and technology delivered by affective computing have demonstrated very promising results across several domains, their adoption by healthcare is still at its initial stages. With that aim in mind, this commentary paper introduces affective computing to the readership of the journal and praises for the benefits of affect-enabled systems for prognostic, diagnostic and therapeutic purposes., peer-reviewed

The question of representing emotion computationally remains largely unanswered: popular approaches require annotators to assign a magnitude (or a class) of some emotional dimension, while an alternative is to focus on the relationship between two or more options. Recent evidence in affective computing suggests that following a methodology of ordinal annotations and processing leads to better reliability and validity of the model. This paper compares the generality of classification methods versus preference learning methods in predicting the levels of arousal in two widely used affective datasets. Findings of this initial study further validate the hypothesis that approaching affect labels as ordinal data and building models via preference learning yields models of better validity., peer-reviewed