HT '24: Proceedings of the 35th ACM Conference on Hypertext and Social Media

LLMs offer valuable capabilities, yet they can be utilized by malicious users to disseminate deceptive information and generate fake news. The growing prevalence of LLMs poses difficulties in crafting detection approaches that remain effective across various text domains. Additionally, the absence of precautionary measures for AI-generated news on online social platforms is concerning. Therefore, there is an urgent need to improve people’s ability to differentiate between news articles written by humans and those produced by LLMs. By providing cues in human-written and LLM-generated news, we can help individuals increase their skepticism towards fake LLM-generated news. This paper aims to elucidate simple markers that help individuals distinguish between articles penned by humans and those created by LLMs. To achieve this, we initially collected a dataset comprising 39k news articles authored by humans or generated by four distinct LLMs with varying degrees of fake. We then devise a metric named Entropy-Shift Authorship Signature (ESAS) based on the information theory and entropy principles. The proposed ESAS ranks terms or entities, like POS tagging, within news articles based on their relevance in discerning article authorship. We demonstrate the effectiveness of our metric by showing the high accuracy attained by a basic method, i.e., TF-IDF combined with logistic regression classifier, using a small set of terms with the highest ESAS score. Consequently, we introduce and scrutinize these top ESAS-ranked terms to aid individuals in strengthening their skepticism towards LLM-generated fake news.

This paper explores the intersection of emotional hermeneutics and artificial intelligence (AI), examining the challenges and potential of integrating deep emotional understanding into AI systems. Drawing on Wilhelm Dilthey’s distinction between "explanation" and "understanding", the study highlights the current limitations of AI, particularly large language models, in achieving a genuine interpretative understanding of human emotions. We argue that while AI excels at data-driven explanations, it lacks the capacity for true emotional comprehension due to its inability to have personal experiences and self-awareness. The paper proposes incorporating humanities and social sciences insights to enhance AI’s ability to contextualize and interpret emotions. However, it acknowledges that replicating human emotional intelligence in AI may be fundamentally limited by the nature of artificial systems. The study concludes by calling for interdisciplinary collaboration to advance emotional AI research while recognizing the ongoing philosophical questions about the nature of intelligence, and emotional understanding.

Recently, the issues of AI fairness and algorithmic oppression have received much interest due to the rising awareness of the dangers that AI poses to minority communities. While new methods of managing bias in AI are being developed, it may be impossible to overcome the normativity that these systems reinforce, as they rely on categorizing already biased data. Queer artists and theorists who recognize these constraints explore speculative approaches for conceptualizing future inclusive technology by means of queering AI. This paper analyzes some artistic instances of queering AI that promote experimental strategies of reconceptualizing AI not bound by what is currently possible. Specifically, I aim to frame queering AI as a speculative practice that draws upon those concepts of queer spatialities and temporalities that serve as a framework for artists trying to overcome AI normativity.

This paper considers hypertext in its various forms as a paradigm that has the potential to reduce a number of ethical concerns that come with (generative) AI. Based on a user scenario, the paper points out some ethical issues and explains how they can be addressed by hypertext. To do so, it distinguishes between System 1 (fast automation of simple tasks) and System 2 (critical thinking) tasks. Drawing on existing publications in philosophy, the paper argues that AI systems cannot be moral agents; they cannot be trustworthy or truly intelligent. This breaks with some of the wording, partly used for marketing purposes, that currently makes “artificial intelligence” a hype. The analysis follows the three most important ethical theories: deontology, consequentialism, and virtue ethics. The paper concludes that hypertext, although a niche topic, is already prepared to solve some of the most prominent and urgent ethical issues in AI.

We present a case study using conversational media videos for scalable and inclusive mental health interventions for schoolchildren. The system was designed to complement the existing “Verbal Wellbeing” programme in Northern Ireland, which provides wellbeing resources to teachers. The Verbal Wellbeing programme equips school teachers with tools, such as storytelling and mental health interventions, to support their pupils’ mental health. The presented system allows participants and users to pose questions about psychology theories, the ‘Verbal Wellbeing’ programme, and evidence of its effectiveness. Users’ questions are answered in a real-time conversational mode from a pool of answers pre-recorded by an expert, complementing face-to-face training, supporting taking new teachers on board, or as a follow-up.

Knowledge Graphs are becoming widely used as a method to capture and integrate diverse sources of data into a unified structure of nodes and links. Cultural heritage is an active domain for Knowledge Graph research, bringing together the metadata of cultural objects with associated information about their use and history. Tools exist for the searching and browsing of Knowledge Graphs but on their own they often do not offer the interpretative support required by a more general audience. This paper describes an approach to creating a layer of interpretation over a Knowledge Graph. Experts in the cultural domain, without expertise in the underlying technology, can curate paths through the Knowledge Graph, selecting and associating cultural objects, which are automatically displayed in the path with relevant content from the Knowledge Graph. Path authors can also provide additional interpretation as well invite responses from followers of the paths. A case study is described in the domain of European pipe organs in which domain experts can curate paths through a Knowledge Graph of currently approximately 2000 objects. The potential of the approach as a way of incrementally formalizing changes or additions to the Knowledge Graph emerged as a theme with domain experts. The applicability of the approach to cultural Knowledge Graphs in general is discussed.

The scholarly edition canonizes and centralizes the work of any writer and provides an academic touchstone for further research. Drawing upon a model of scholarly editions proposed by Christopher Ohge based in philosophical pragmatism, the authors of this paper consider the methodological implications of a hypertextual environment for the development of editing work on the writer Vernon Lee (1856-1935). We simultaneously propose a frame for the development of a prototype edition of Lee's Proteus; Or, The Future of Intelligence (1925) and suggest how hypertextual methodology might be deployed in the preparation of a print edition. Using Lee's literary scholarship to provide shape to our approach, we suggest both a Protean ‘edition to be’ and consider how we might embed these Protean principles in a more traditional print edition.

A defining characteristic of hypermedia systems is the presence of hypermedia controls. In this paper we examine hypermedia controls as found "in the wild", in particular in the World Wide Web. These hypermedia controls are analyzed to derive a functional hypermedia mechanic that can be used to characterize them. This functional mechanic is used to create first an informal and then formal definition of the term "hypermedia control". Using this formal definition we then derive a generalization of the concept, referring contextually to the World Wide Web. We then examine two hypermedia technologies that implement this concept of generalized hypermedia controls: htmx, which does so in the context of the WWW and Hyperview, which does so in a mobile context.

Hypertext visualizations embed interesting assumptions about the underlying structure of ideas. Few novel approaches to hypertext maps have been presented in recent years. The Gaudí view tessellates the idea plane, exploring an approach to presenting a greater number of notes on the screen, at the cost of restricting the fixity of the visualization: you can move notes, but notes can move themselves.

A legacy of print is that much of our process and tooling is predicated on using text in paginated form, such as was required for (paper) printed media. Increasingly, digitally-created (‘born-digital’) documents will never be used non-digitally and yet their internal structure is still optimised for paginated presentation. As modern displays now offer the affordance of decomposition and remediation of complex documents, this article explores the limitations of this legacy design for post-paper-print era use and considers the design implications for writing tools intended for the new digital era.

The Authoring Burden describes the overhead of writing an Interactive Digital Narrative (IDN) as opposed to a linear story. There are many methods and tools that have been proposed for writing IDNs with the implicit goal of reducing this burden, but because there is no comprehensive model of the Authoring Burden assessing the impact of these approaches is difficult. We have undertaken interviews with IDN authors (n=14) to understand how they manage the authoring burden within their own projects. Based on these interviews, and drawing on the existing literature, we propose a model of the Authoring Burden comprised of three parts: Content Creation, Dynamic Authoring, and Programming/Tool Creation. The initial size of this burden is set by the Author’s Goals informed by their Capability and the Audience/Publishing Context. We also find 29 strategies employed by authors to manage the burden. There are five distinct types. Embracing and Reducing strategies impact the overall scale of the challenge, whereas Generative, Reuse, and Decoupling strategies move work between the three parts of the model. We validate our model with focus groups comprising different sets of experts (n=8). Our model shows that many strategies for managing the burden transform rather than reduce work, and that the most appropriate strategy for a given author will be highly dependent on their personal goals and capabilities. It therefore highlights the heterogeneity of IDN as both a strength, but also a challenge to theorists and tool designers.

Interactive stories can be an effective approach for teaching purposes. One shortcoming is the effort necessary to author and create these stories, especially complex storylines with choices for the readers. Based on recent advances in Natural Language Processing (NLP), new opportunities arise for assistance systems in the context of interactive stories. In our work, we present an authoring approach and prototypical tool for the creation of visual comic-strip like interactive stories, a type of hypercomics, that integrate an Artificial Intelligence (AI) assistance. Such comics are already used in our Gekonnt hanDeln web platform. The AI assistance provides suggestions for the overall story outline as well as how to design and write individual story frames. We provide a detailed description about the approach and its prototypical implementation. Furthermore, we present a study evaluating the prototype with student groups and how the prototype evolved in an iterative style based on the students’ feedback.

Powerful predictive AI systems have demonstrated great potential in augmenting human decision making. Recent empirical work has argued that the vision for optimal human-AI collaboration requires ‘appropriate reliance’ on AI systems. However, accurately estimating the trustworthiness of AI advice at the instance level is quite challenging, especially in the absence of performance feedback pertaining to the AI system. In practice, the performance disparity of machine learning models on out-of-distribution data makes the dataset-specific performance feedback unreliable in human-AI collaboration. Inspired by existing literature on critical thinking and mindsets, we propose debugging an AI system as an intervention to foster appropriate reliance. This paper explores whether a critical evaluation of AI performance within a debugging setting can better calibrate users’ assessment of an AI system. Through a quantitative empirical study (N = 234), we found that our proposed debugging intervention does not work as expected in facilitating appropriate reliance. Instead, we observe a decrease in reliance on the AI system — potentially resulting from an early exposure to the AI system’s weakness. Our findings have important implications for designing effective interventions to facilitate appropriate reliance and better human-AI collaboration.

Multilingual countries are made of communities with their own history and traditions that maintain their identity but, at the same time, have strong social interactions and exchanges. This paper proposes a particular perspective to look at these communities: news circulation and translation. The hypothesis is that understanding how the same news is treated by different communities, and how each community focuses on some news instead of others, can be a proxy for their cultural similarities and differences. The paper presents a pilot analysis run on Switzerland news and exploits a fully quantitative approach. The experiments confirmed the feasibility of such an automatic approach and gave valuable insights about possible improvements and applications.

Latin America has witnessed various initiatives over the past decade aimed at identifying, archiving, and curating digital literature, particularly on a continental or national scale. However, there remains a notable scarcity of archives focusing on a more localized (province/state/city) level within the continent. This deficiency hampers our comprehension of the dynamics within digital literature communities formed by creators and readers who share immediate geographical references, as well as closer personal and institutional connections. The primary aim of this paper is to undertake a comprehensive analysis and description of a digital literature archive that fills this gap: the hypertextual Acervo de Literatura Digital Mato-Grossense, showcasing digital literature produced by artists residing in or hailing from the state of Mato Grosso (Brazil). In contributing to similar endeavors, we document and expound upon the key decisions we made, considering the distinct characteristics of digital literature in a region distant from Brazil’s major urban centers (itself a peripheral country within the technocapitalist world system). We elucidate the decolonial concept of digital literature adopted, the archival and curatorial strategies tailored to accommodate these works’ realities, the selection of the open-source hypertextual repository platform Tainacan to structure our archive, and the exploratory methods employed to identify and select works for archiving. Subsequently, we provide an overview of the materials archived to date and outline the forthcoming phases of the project.

Social media platforms are designed in a way that keeps users engaged and occupied for as long as possible, by means of various user interface design elements and personalization techniques. A known problem with these techniques is that they may cause compulsive behavior and feelings of regret because of the time they wasted. However, as users continue to engage with content that researchers have identified as problematic, the overall experience cannot always be that negative. To shed light on this apparent paradox, this study investigates the subjective perceptions of users and provide a constructivist perspective of what we call ‘digital junkfood’. We identified a rich variety of relevant content elements, evoked feelings and behavioral responses. The often conflicting positive, neutral and negative feelings and responses elicited by digital junkfood call for differentiated, individualized rather than normative approaches towards compulsive social media use and the personalization techniques associated with this behavior.

In this paper, we propose an innovative framework for scientific podcasts that integrates principles of ergodic literature and hypertext theory to enhance user engagement and educational depth. Traditional podcasts follow a linear structure, limiting user interaction and exploration. Our approach disrupts this norm by transforming podcasts into non-linear, interactive experiences. We introduce a prototype that segments expert interviews into concise, interconnected video fragments, each augmented with hyperlinked references to related scientific articles, interactive diagrams, and educational resources. This method allows users to navigate content in a personalized, exploratory manner, facilitating deeper engagement with complex topics in computational neuroscience. Detailed metadata annotations and dynamic hyperlinking create a robust network of content, enabling seamless transitions between topics and revisiting previously explored segments. We incorporate game design elements to sustain interest and employ user tracking to personalize content recommendations, enhancing the overall learning experience. This prototype represents a significant advancement in digital media, combining the accessibility of podcasts with the interactivity of hypertext, thus paving the way for more immersive and informative scientific communication.

International Auxiliary Languages (IALs) are constructed languages designed to facilitate communication among speakers of different native languages while fostering equality, efficiency, and cross-cultural understanding. This study focuses on analyzing the editions of IALs on Wikipedia, including Simple English, Esperanto, Ido, Interlingua, Volapük, Interlingue, and Novial. We compare them with three natural languages: English, Spanish, and Catalan. Our aim is to establish a basis for the use of IALs in Wikipedia as well as showcase a new methodology for categorizing wikis. We found in total there are 1.3 million articles written in these languages and they gather 15.6 million monthly views. Although this is not a negligible amount of content, in comparison with large natural language projects there is still a big room for improvement. We concluded that IAL editions on Wikipedia are similar to other projects, behaving proportionally to their communities’ size. Therefore, the key to their growth is augmenting the amount and quality of the content offered in these languages. To that end, we offer a set of statistics to understand and improve these projects, and we developed a web page that displays our findings to foster knowledge sharing and facilitate the expansion of the IAL communities.

Histories of hypertext have largely upheld a view of its subject as fundamentally interdisciplinary, recognising those phases of outside influence that have waxed and waned over the decades. Film and literary studies, social sciences and book history: ongoing developments in each brought new perspectives to hypertext over the years, leaving its mark on the language we use today.

The recent past has seen a revival in such activity, with hypertext increasingly reaching out to scholars from adjacent communities. This work brings new challenges and fresh opportunities for collaboration; at the same time, established language comes under new interrogation as scholars begin to explore hypertext's history in a new way. Drawing on insights from linguistic theory, cybernetics, and critical discourse analysis, this paper explores the dynamics of interdisciplinary collaboration in hypertext research. It examines how language shapes power relations, influences the negotiation of research agendas, and impacts the evaluation of interdisciplinary work in general.

We present a procedure for assessing group creativity that allows us to compare the contributions of human interlocutors and chatbots based on generative AI such as ChatGPT. We focus on everyday creativity in terms of dialogic communication and test four hypotheses about the difference between human and artificial communication. Our procedure is based on a test that requires interlocutors to cooperatively interpret a sequence of sentences for which we control for coherence gaps with reference to the notion of entailment. Using NLP methods, we automatically evaluate the spoken or written contributions of interlocutors (human or otherwise). The paper develops a routine for automatic transcription based on Whisper, for sampling texts based on their entailment relations, for analyzing dialogic contributions along their semantic embeddings, and for classifying interlocutors and interaction systems based on them. In this way, we highlight differences between human and artificial conversations under conditions that approximate free dialogic communication. We show that despite their obvious classificatory differences, it is difficult to see clear differences even in the domain of dialogic communication given the current instruments of NLP.

Combating misinformation on social media is critical, with preemptive strategies like prebunking and nudging gaining prominence. This paper evaluates the effectiveness of nudge and prebunking strategies in enhancing individuals' ability to distinguish between misinformation and factual content and their confidence in their accuracy judgments. Employing a between-subject experimental design, participants (N = 328) were categorised into three conditions: a control condition with no intervention, a nudge condition exposed to behavioural cues to promote critical scrutiny of information and a prebunking condition receiving implicit context about the claim. The results indicate that the prebunking message improves the identification of misinformation and confidence in judgments compared to nudging and control interventions. No significant difference was observed between the nudge and control groups regarding judgment accuracy or confidence. However, individual differences in interventions were noted. The study reveals that deliberate thinkers require some form of intervention to discern false news claims effectively. Moreover, it was found that participants with right-leaning political views were less influenced by prebunking messages, suggesting that nudges might be more effective for this demographic. These findings highlight the necessity of adopting a user-centric approach that considers individual characteristics to tailor interventions that may be required to combat misinformation effectively among diverse user groups on social media platforms.

Due to the increasing ubiquity of social media, compulsive social media use is an increasingly persistent issue that compels users toward excessive social media usage, which can lead to negative consequences. Though reasons for this behaviour are difficult to research, it is theorized that addictive social media design may contribute to compulsive social media use. In this study, we design and evaluate alternatives for the so-called infinite scroll, offering natural moments to leave the feed, and for notifications, providing users with various types of support to decide to respond to them, or to ignore or postpone them. The designs are inspired and informed by the literature and the results of a survey study. The results of this study contribute to understanding addictive design better and can contribute to the improvement of social media to make them less addictive.

Web analytics tools like Google Analytics are widely available, but website owners outside the eCommerce sector struggle to extract actionable insights from their data to curate and optimize content. This difficulty often arises from challenges in identifying and aligning objectives with standard website performance metrics provided by these tools, compounded by a lack of expertise in tool configuration. This study focuses on automated approaches that generate actionable insights for owners of content-driven websites, analyzing visitor attention at the most granular level by focusing on segments of web pages. It considers both the length of the page and different device types used to access these pages.

Existing research is augmented with four major contributions: First, a robust regression model to predict user behaviour based on the scroll behaviour of 850,000 visitors and more than 9 million data points from five diverse websites. Second, a dataset of measurements of web page lengths from a random sample of one million websites, for a better understanding of the relation between scroll behaviour and web page lengths. Third, an actionable knowledge discovery method for web analytics data of non-transactional websites that allows to identify deviations from expected visitor behaviour, enabling content optimization for those web analytics users who find it difficult to leverage their data today. Finally, an indicator for page performance that allows to compare page performance based on in-page visitor engagement. This research exemplifies the intersection of web analytics and intelligent content curation, showcasing a methodological framework that facilitates the generation of automated suggestions for digital content optimization, rooted in comprehensive behavioral data analysis.

By ‘augmenting human intellect’ we mean increasing the capability of a man to approach a complex problem or situation, to gain comprehension to suit his particular needs, and to derive solutions to problems.” So wrote Douglas Engelbart in 1962, initiating a narrative of augmentation that runs through hypertext's history. To augment is to amplify something: our ideas, our activities, our thoughts. This paper explores hypertext's relationship with computers as intentional machines, its relationship with artificial and human intelligence, and what role it can play in negotiating between the two.

This paper explores "digital gardening" as a form of feral hypertext and a network of practice for personal information organization. Through an examination of digital gardens in practice, we investigate the structure of conceptualization and the varied implementations of digital gardening. We aim to contribute to the discussion on the feralization of hypertext, its implications for personal information management in today’s digital age, and the dynamic between the emergence of concepts and practices in the online sphere and the adoption by the public.

Hypertext offers a distanced perspective on technology and communication. On the level of style, for example, the link can be seen as modern punctuation mark [15]. On the level of narrative design, the link functions as a logical engine of interaction in computer games and elsewhere [1]. On the level of social discourse, the link serves as a tool for expressing individual freedom [7]. Hyperlepsis—a multi-level trope introduced in this article, found in interactive fiction (IF), hypertext fiction and ambient literature for smartphones, combines grammar, style, and discourse within a single rhetorical gesture. As such, hyperlepsis can be a useful descriptive category for evaluating the effectiveness of interactive rhetorics across all digital forms. It renders early hypertext fiction as a reservoir of artistic ingenuity and a field for creative intelligence. As a trope from the same group as metalepsis it contributes to the reflection on embodied, material tropes – "sign-tool-objects" – as subordinate tools of immersion, rather than disruption and metafictional intrusion. Finally, for preservationists and translators, the presence of hyperlepsis can be a crucial indicator about components of the original work that cannot be lost during migration from source to target or during reconstruction.

This paper posits that spatially structured concepts can function as a visual representation of knowledge, a notion supported by common methods of eliciting and presenting mental models. Despite an existing gap in understanding the correlation between visual structure and knowledge representation, this study aims to clarify this relationship. To this end, we conducted a study wherein participants rated pairwise relationships between ten concepts on a discrete scale ranging from one to ten. Subsequently, we compared these ratings with weights derived from the distances between concepts in human-generated spatial structures. Our findings unveil a linear relationship between the weights obtained through both methods, indicating that spatial arrangements may systematically reflect and encode knowledge.

Hypertext readership has been placed under extensive scrutiny. At the same time, little to no work has been done on how hypertext systems can contribute to reading scholarship. This contribution reports on the experience of the EU Joint Programme Initiative for Cultural Heritage (JPI-CH) funded READ-IT project and the development of an interactive, multimodal system to crowdsource testimonies of reading experiences. The contribution describes articulating the research infrastructure aimed at federating research case studies across Europe in several languages and about the readership of different kinds in different periods. The infrastructure aims to broaden the understanding of reading in Europe. The infrastructure included a system to collect sources (testimonies of reading experiences) to be studied within and beyond the project. The infrastructure also provided a combination of online forms, postcards, and a chatbot to facilitate reflection on reading by members of the public. This contribution accounts for the experience of using this system in several countries and throughout a wide range of initiatives.

Individuals can leverage data referencing in collective scenarios, akin to deixis, a common human interaction in which they can physically point to relevant information. While deixis is easily performed in online synchronous interactions, e.g., mouse pointing during screen sharing in video conferences, reproducing it in asynchronous online interactions is not trivial.

A significant step towards finer mimicking in-person co-presence is enhancing the granularity of links to data, enabling users to reference specific pieces of data in their content. This paper reviews 43 digital platforms in 8 sub-categories, organized in 3 main categories, and proposes requirements, opportunities, and challenges of link granularity for asynchronous online interactions.

Discussion forums are one of the favored platforms for knowledge sharing. Given their popularity, copious research exists on understanding linguistic and behavioral characteristics of forum conversations. However, prior investigations have mainly focused on general forums designed primarily for sighted users, and as such the applicability of their findings to the dedicated accessibility discussion forums frequented by blind individuals remains unanswered. To bridge this knowledge gap, and facilitate the development of better-informed assistive technologies for blind people, we investigated language use and identified the key semantic and cognitive characteristics of online accessibility forums. To aid our investigation, we collected a dataset of 1000 accessibility forum threads and a baseline of 1000 general forum threads. These threads were carefully curated to ensure similarity of topics discussed. We found the language in accessibility forum conversations to be more task-oriented and less abstract, with significantly higher number of descriptive action words than in general forum conversations. Results also showed an emphasis on sharing first-hand personal experiences in accessibility forums, relative to the general forums.

Social media influence materialises in the physical world by shaping our urban environment and changing the modality we experience. This under-explored outcome still lacks the conceptual tools to build a discourse around it. We present a case study that showcases how urban planning and development are influenced by social media and a first framework addressing the interrelation between social media, urban design, and the experience of the urban environment. This short contribution provides concepts for critically analysing the physical implications of social media-based ubiquitous hypertext systems.

Given the recent proliferation of fake news online, fact-checking has emerged as a critical defence against misinformation. Several fact-checking organisations are currently employed in the initiative to assess the truthfulness of online claims. Verified claims serve as foundational data for various cross-domain research, including fields of social science and natural language processing, where they are used to study misinformation and several downstream tasks such as automated fact-verification. However, these fact-checking websites inherently harbour biases, posing challenges for academic endeavours aiming to discern truth from misinformation. In this study, we aim to explore the evolving landscape of online claims verified by multiple fact-checking organisations and analyse the underlying biases of individual fact-checking websites. Leveraging ClaimsKG, the largest available corpus of fact-checked claims, we analyse the temporal evolution of claims, focusing on topics, veracity levels, and entities to offer insights into the complex dimensions of online information. We utilise data and dimensions available from ClaimsKG for our analysis and for dimensions such as topics which are not present in ClaimsKG, we create a topic taxonomy and implement a transformer-based model, for multi-label classification of claims. We also observe how similar claims are co-occurant amongst different websites. Our work serves as a standardised framework for categorising claims sourced from diverse fact-checking organisations, laying the foundation for coherent and interpretable fact-checking datasets. The analysis conducted in this work sheds light on the dynamic landscape of online claims verified by several fact-checking organisations and dives into biases and distributions of several fact-checking websites.

Human smuggling is a grave social issue that carries several negative ramifications for many countries across the globe. Human smugglers operating on the United States-Mexico border, known as ‘coyotes’, are increasingly relying on online social media to advertise their services, so automatically detecting such advertisements related to this illicit activity is crucial for devising effective countermeasures. However, identifying human-smuggling advertisements is not straightforward as they are often disguised as innocent travel or tourism-related adverts in regional languages. Moreover, there are no readily available datasets that can train models to recognize such adverts automatically. This paper addresses both these issues as follows. First, we built a novel dataset comprising both coyote ads and legitimate travel/tourism adverts by leveraging different sources, including the Web, NGOs, and regional connections in countries with high human-smuggling activities. Specifically, our dataset contained an equal number (1000 images plus 500 videos) of coyote ads and legitimate travel ads, all of which contained embedded textual information. Using this dataset, we trained and assessed several state-of-the-art baseline models, including GPT-4, Gemini, and CLIP. For the image ads, the highest F1 score achieved by a model was 0.92, and for the video ads, the highest achieved F1 score was 0.86. We also conducted an in-depth analysis of the model performances to gain comprehensive insights into their respective strengths and weaknesses.

One way that YouTube aims to support its users to make sense of problematic content they find on its platform is through the inclusion of information panels (IPs) in its search results or video watch pages. Such panels, when present, offer topical context from third-party sources such as Encyclopedia Britannica or Wikipedia, but also fact-checks or breaking news. What search queries trigger these IPs? Do they appear consistently over time and space? Using 620 query phrases related to conspiracy theories, crowd-sourced from 309 individuals in the United States, we conducted a longitudinal audit of YouTube search pages, simultaneously searching for the query phrases, using computers scattered in 14 locations during four months. We find that only 16.63% of the search pages contained an IP, with 114 unique query phrases triggering a total of 49 unique panels of three different types (context, fact-checks, and news). Qualitative analysis of queries and IPs reveals that keywords in query phrases match with keywords in content labels. Context IPs cover multiple queries related to the same conspiracy theory, while fact-check and news IPs are usually targeted to one particular query. However, many topical queries do not trigger an IP, indicating inconsistent coverage. Since some IPs seem to be sensitive to real-world events, more research into methods for “news-cycle-aware auditing” might be necessary.

The exhibition shows the coexistence of humans and technology in everyday life, exploring the topic from various perspectives. It consists of four sections through which participants will navigate using an application that reads QR codes. The application will guide participants to subsequent parts of the exhibition and provide basic information about the thematic context of each section, as well as suggest interpretative clues and insights. Each section presents different aspects of everyday life dominated by technology, bringing out the artistic potential of everyday human activity. The key focus will be the interaction between humans and digital technology, especially generative artificial intelligence, as the basis for considering creative cooperation. The individual sections of the exhibition are titled as follows: [1.1] The Boundaries of Creativity: Dimensions of Human and Non-Human Creativity, [1.2] Humans as Objects of Technological Experiences, [1.3] Technology as a Capsule of Memories, and [1.4] Technology as a Safety Valve.

This article describes the theoretical matrix from which this particular work of interactive media emerges while also expounding on how the subject matter and content of the piece relates to the ACM conference theme of “Creative Intelligence.” It is suggested that Creative Intelligence is shaped and made manifest by the medium of an artistic creation – in this case, the particular hypertext tool employed to compose the work (Twine). Comparison is made to the author's previous work in Storyspace, a late 20th century hypertext authoring software available at the dawn of digital literature. The author suggests that these different modes of composition allow for a particular assemblage of self, a result of medium-specific peculiarities of the authoring tools themselves.

The submission is for the first-ever exhibit of the interactive, hypertextual radio show "The Puffball" (Purchawka) by Artur Sosen Klimaszewski and Przemysław Frankowski (aka DJ Jah Jah) that aired in Poznań’s Radio S for 12 episodes in 1999 and 2000. The show represents a genre of radio hypertext that is non-digital yet not fully feral.

This paper investigates the application of large language models (LLMs), specifically GPT-4, in translating and transforming Alfred Jarry’s avant-garde play Ubu Roi (Ubu King) through the lens of "uncreative writing." This concept involves repurposing existing texts to generate new forms of literary expression that challenge traditional notions of originality and authorship. We conducted a case study by first translating Ubu Roi from French to Polish using GPT-4. Subsequently, the translated text was reinterpreted into various genres and styles, showcasing the model’s capability to navigate and creatively reshape classical literature. Generative Adversarial Networks (GANs), particularly Dall-E 3, were employed to produce illustrative content complementing these textual adaptations. This experiment highlights the transformative potential of AI in literature, emphasizing the convergence of digital technology and classical literary forms to generate novel insights and reinterpretations. The culmination of this project is the publication of a book that amalgamates these AI-driven reinterpretations, bridging the digital and analog realms.

Exploration of the unique confluence of literary and programming elements, exemplified by "In nihilum reverteris," a collaborative creation by Yerzmyey and Hellboj. This work transcends conventional digital genres—blending aspects of electronic literature, computer games, and demoscene creations—to produce a hybrid form that immerses the viewer in a world where familiar reality is annihilated. The paper examines how such digital works use complex interplays of narrative and code to foster new forms of creative expression. It argues that these hybrid forms, despite their rootedness in distinct aesthetic and technical traditions, offer a new immersive potential that challenges the boundaries of genre purity. This study asserts that the piece’s placement within demoscene and gaming communities, rather than purely literary circles, highlights the inter-genre dialogue that is crucial for understanding and advancing digital artistic expressions.

The early years of hypertext research saw a profusion of strange new hypertext models, formalisms, and systems. The exuberant abundance of this era was overwhelmed by the emergence of the Web and, later, the field’s fascination with social media. Elegies to the freedom of the classical era of hypertext continue to be heard, yet novel systems are rare. This paper proposes a simple architectural twist that embeds web components — which may include hypertext systems — inside a classical hypertext framework. The goal of this work is to enable researchers to implement useful, testable prototype systems very quickly: an exotic hypertext system in an afternoon.

The history-based content present on Wikipedia has been used for various NLP-related research. For large-scale analysis, efficient retrieval of past states of Wikipedia is a prerequisite. However, the lack of efficient tools for managing the massive amount of provided data acts as a bottleneck. We present a detailed analysis of online algorithms to efficiently compress and retrieve the revision history of Wikipedia articles. We give theoretical evidence that our methods perform efficient compression and extraction while optimizing time and space complexity.

Moreover, the experiments on sampled Wikipedia articles using the online parameters extraction method show that our algorithm can compress the dataset up to 94% of its original size. To the best of our knowledge, this is the first attempt to show a detailed analysis of Wikipedia’s full revision history dataset compression.

This demonstration is derived from research currently being undertaken by an international team of scientists, artists, and digital humanities scholars about the future of text in XR. In its current stage of development, the project uses WebXR to access a library of documents derived from ACM Hypertext proceedings; select a document or documents to read; manipulate the documents by moving them around in the virtual space; navigate among various linked outputs of the documents, such as abstracts and references; and save the state of the environment.

This paper presents a customized and sustainable workflow for digital scholarly publications in the humanities that aims to overcome the difficulties of transitioning from traditional fixed formats to those that require fluid layout and structure, such as ePub, (X)HTML or XML. While the process fits well within the humanities tradition and takes into account the specificities of the humanities that distinguish them from the hard sciences, such as extended footnotes and non-DOI references, it can be easily adapted to publications in different fields. The paper outlines the various challenges encountered in the digital publication process, including the transformation of text documents into JATS XML and the process of uploading to OJS (Open Journal System). It describes our attempts to streamline the steps and provide solutions that fit within the capacity of limited budgets without compromising the content and structure of the publications. The paper concludes with an evaluation of the proposed workflow and outlines future improvements to increase its efficiency.

Short videos have become one of the dominant formats of social media content. Such short videos are often boosted by advertising to increase their reach and visibility. We investigate user churn (i.e., drop or loss of consumers during the video viewing) of short video ads on TikTok. Data from the TikTok Ads account of an e-commerce company was analyzed using logistic regression to examine how gender, age groups, and video duration affect users’ video viewing behaviors. The findings indicate that (1) most customers leave the ads within the first quarter of the video; (2) males and females exhibit similar viewing behaviors, but (3) age groups vary by their viewing behaviors. Somewhat surprisingly, we discovered that (4) there is no significant correlation between user churn and ads performance metrics (i.e., cost-per-mille, cost-per-click, and click-through rate).

We present HyperCausal, a 3D hypertext visualization framework for exploring causal inference in generative Large Language Models (LLMs). HyperCausal maps the generative processes of LLMs into spatial hypertexts, where tokens are represented as nodes connected by probability-weighted edges. The edges are weighted by the prediction scores of next tokens, depending on the underlying language model. HyperCausal facilitates navigation through the causal space of the underlying LLM, allowing users to explore predicted word sequences and their branching. Through comparative analysis of LLM parameters such as token probabilities and search algorithms, HyperCausal provides insight into model behavior and performance. Implemented using the Hugging Face transformers library and Three.js, HyperCausal ensures cross-platform accessibility to advance research in natural language processing using concepts from hypertext research. We demonstrate several use cases of HyperCausal and highlight the potential for detecting hallucinations generated by LLMs using this framework. The connection with hypertext research arises from the fact that HyperCausal relies on user interaction to unfold graphs with hierarchically appearing branching alternatives in 3D space. This approach refers to spatial hypertexts and early concepts of hierarchical hypertext structures. A third connection concerns hypertext fiction, since the branching alternatives mediated by HyperCausal manifest non-linearly organized reading threads along artificially generated texts that the user decides to follow optionally depending on the reading context.

Social media datasets are essential for research on disinformation, influence operations, social sensing, hate speech detection, cyberbullying, and other significant topics. However, access to these datasets is often restricted due to costs and platform regulations. As such, acquiring datasets that span multiple platforms which are crucial for a comprehensive understanding of the digital ecosystem is particularly challenging. This paper explores the potential of large language models to create lexically and semantically relevant social media datasets across multiple platforms, aiming to match the quality of real datasets. We employ ChatGPT to generate synthetic data from a real dataset consisting of posts from three different social media platforms. We assess the lexical and semantic properties of the synthetic data and compare them with those of the real data. Our empirical findings suggest that using large language models to generate synthetic multi-platform social media data is promising. However, further enhancements are necessary to improve the fidelity of the outputs.

Establishing rapport between authors and readers of scientific texts is essential for supporting readers in understanding texts as intended, facilitating socio-discursive practices within disciplinary communities, and helping in identifying interdisciplinary links among scientific writings. We propose a Reader-aware Congruence Assistant (RaCA), which supports writers to create texts that are adapted to target readers. Similar to user-centered design which is based on user profiles, RaCA features reader-centered writing through reader profiles that are dynamically computed from information discovered through academic search engines. Our assistant then leverages large language models to measure the congruence of a written text with a given reader profile, and provides feedback to the writer. We demonstrate our approach with an implemented prototype that illustrates how RaCA exploits information available on the Web to construct reader profiles, assesses writer-reader congruence and offers writers color-coded visual feedback accordingly. We argue that our approach to reader-oriented scientific writing paves the way towards the more personalized interaction of readers and writers with scientific content, and discuss how integration with Semantic Web technologies and Adaptive User Interface design can help materialize this vision within an ever-growing Web of scientific ideas, proof, and discourse.

This paper presents "Seed Hypermedia," an innovative architecture for scalable and decentralized collaboration on the web. It integrates elements such as digital signatures, peer-to-peer (P2P) gossip protocols, and distributed hash tables (DHT) to create a robust, secure, and user-managed content system. The core of Seed Hypermedia is a decentralized, indexable network designed to support deep-linking and fine-grained access to content, leveraging IPFS for immutable data storage and CRDTs for managing version-controlled documents. This system enables collaborative editing with a high degree of data integrity through a directed acyclic graph (DAG) of changes, allowing users to maintain ownership and control over their contributions. The architecture promotes a more democratic web environment by decentralizing the indexing and retrieval of content, significantly reducing the reliance on central servers. By integrating advanced cryptographic techniques and a peer-to-peer networking model, Seed Hypermedia aims to empower creators and communities, ensuring permanence, traceability, and scalability in digital content management.

Flexibility, adaptability, modularity, and extensibility in the context of a collaborative system are critical features for multi-user hypertext systems. In addition to facilitating acceptance and increasing reusability, these features simplify development cycles and enable a larger range of application areas. However, especially in virtual 3D hypertext systems, many of the features are only partially available or not available at all. To fill this gap, we present an approach to virtual hypertext systems for the realization of dynamic event systems. Such an event system can be created and serialized simultaneously at run time regarding the modification of situational, environmental parameters. This includes informing users and allowing them to participate in the environmental dynamics of the system. We present Va.Si.Li-ES as a module of Va.Si.Li-Lab, describe several environmental scenarios that can be adapted, and provide use cases in the context of 3D hypertext systems.

Within the realm of peer production, this study explores whether individuals are inclined to generate entertainment content or serious information, and whether these forms of content creation are in competition. The analysis encompassed 315,916 articles edited by 8,551 editors on Baidu Baike, a platform analogous to Wikipedia. Utilizing a pre-trained neural network language model, the articles were categorized to discern editorial preferences in content modification. Social network analysis and fixed effects models illuminated consistent patterns in individuals’ propensities toward diverse content editing behaviors, underscoring the interplay between entertainment and serious content production. These findings deepen our understanding of users’ preferences and motivations in content creation and editing, while also contributing to our insights into the actors involved in science and health communication initiatives.

As the integration of artificial intelligence into social media continues to attract attention, the key impacts on content marketing are still undefined. Initial studies have shown that language models are capable of producing content that is competitive with content created by humans. However, how can such content be tailored for different social media platforms as part of an organization’s content marketing strategy? To address this question, we evaluate the effectiveness of using GPT-4, to generate cross-platform content for Facebook, Instagram, and Twitter (currently X). Participants (N = 892) evaluated 30 AI-created content (ACC) and human-created content (HCC). Findings show that ACC scored higher on preference by users, call-to-action, and emotional responses than HCC for Facebook. However, AI’s advantage wanes on Twitter and Instagram, where posts are terser. The results imply that GPT-4 comprehends what type of content to create for different platforms, making it a useful tool for cross-platform content creation.

HUMAN 2024 is the 7th workshop of a series for the ACM Hypertext conferences. The HUMAN workshop has a strong focus on the user and thus is complementary to the strong machine analytics research direction that could be experienced in previous conferences.

The user-centric view on hypertext not only includes user interfaces and interaction, but also discussions about hypertext application domains as well as human-centered AI. Furthermore, the workshop raises the question of how original hypertext ideas (e.g., Doug Engelbart’s “augmenting human intellect” [6] or Jeff Conklin’s “hypertext as a computer-based medium for thinking and communication” [5]) can improve today’s hypertext systems.

Web/Comics 2024 continues the workshop series commenced at last year’s ACM Hypertext conference. It focuses on the transformation of the comics medium enacted by hypertext through the emergence of webcomics, or “graphic sequential narratives that are created, published, and read on-line” [6]. The Web/Comics workshop brings together interdisciplinary perspectives from the humanities and technological communities to share work and discuss the latest research on webcomics from the perspectives of multiple disciplines. It wants to act as a bridge to increase collaboration between the comics and hypertext research communities. Researchers and practitioners working with webcomics or hypertext are invited to attend this workshop. The planned event is a half day hybrid workshop with emphasis on opportunities for dialogue and discussion in a roundtable session.

In keeping with this year’s conference theme, we particularly welcome discussions that address topics such as

Online Social Networks (OSNs) have become part of everyday life for many people around the world. They are one of the main channels through which information can spread at lightning speed. Thanks to this fact, people use them for the most disparate reasons, such as sources of information in place of newspapers, to receive emotional or technical support, or to share their ideas and opinions to satisfy their need for sociality.

Since their introduction, people questioned these services because they are affected by several problems. These problems include the preservation of the users’ privacy, fake news diffusion, diffusion of illegal pieces of content, censorship vs. free speech, economic value redistribution, security vs trust, and so on. This workshop aims to partially overcome these problems by setting up a platform for researchers to publish their contributions.

The contributions can point to innovative methods and algorithms for social graph mining, which can help develop more efficient information diffusion techniques; the problem of privacy, and how it can be enforced in these systems is current, and in particular the relation between security, trust and privacy is crucial in the scenario of OSNs; the decentralisation and its impact on the implementation of social services; how Artificial Intelligence techniques that respect the privacy of the users can be implemented; technologies that enable the metaverse.

LIRAI is a workshop series on Legal Information Retrieval and Legal Artificial Intelligence. It provides a forum for discussing current trends and challenges in legal artificial intelligence, specifically related to the hypertext nature of legal documents and retrieval tasks. The second edition of LIRAI focuses on three main directions: explainable / justifiable artificial intelligence, hybrid systems that combine formal approaches and machine learning-based methods, including deep learning-based methods, and finally generative artificial intelligence. We call for contributions on these topics in the form of short and long papers, and we aim to publish them as open-access proceedings on CEUR-WS.org once again.

NHT is a continuing workshop series associated with the ACM Hypertext conference for over a decade. The workshop acts as forum of discussion for the narrative systems community within the wider audience of the Hypertext conference. The workshop runs both presentations from authors of accepted short research papers, and unstructured unconference sessions to provide a venue for important discussions of issues facing and opportunities for members of the narrative and hypertext community. This year the workshop aims to build on last years discussed of “Mixed Reality Hypertext” with a targeted discussion of the ethics confronting narrative hypertext and specifically mixed reality hypertext as part of a wider discussion on the ethics of mixed reality games.

The first edition of the Positive Intelligent Media (PiM) workshop brings together scholars and practitioners with a shared interest in interactive intelligent hypertext systems for health, wellbeing and resilience. The PiM workshop will provide the opportunity to develop a common agenda for this new field and define the research objectives around computational methods for steering positive use, evaluation methods, and applications of positive systems. The workshop hosts an open forum to identify common challenges and themes to inform a white paper and a community manifesto.