Computational iconology, embodied spatial interfaces, and the survival of visual patterns

Generative image models are usually described from the outside, by what they output. This paper takes the opposite route, tracing how an AI-generated image comes into being as a movement through latent space, and treating that trajectory as the object of analysis rather than the finished picture. Opacity becomes a question about formation.

Extinction data is hard to hold at the scale it describes. Here the dataset becomes a space in virtual reality that the visitor moves through, so that what is lost registers as spatial memory rather than as a figure on a page. Navigation is the interface to the record.

Moving through a space counts as a way of knowing it, not as a way of getting somewhere. The paper applies that premise to immersive acoustic environments, where locomotion becomes the means by which a listener comes to understand timbre. What the body does with distance and direction is what the argument turns on.

The work builds an imaging pipeline that turns SBBrasil dental records stored as PDFs into classified dental views, keeping human correction inside the labeling loop.

Anomalous events in video are rare by definition, which leaves detectors short of labeled examples for the cases that matter most. The paper approaches detection without task-specific training, aligning visual representations with language so that categories absent from training can still be flagged. That shifts the problem from collecting anomalies to describing them.

Generative art theory supplies the frame for reading what recent technological advances do to the relation between human and machine in artistic production. What counts as the human contribution, and what as the machine's, is the question the study puts to that literature.

An evolving fuzzy neural network built on or-type logic neurons is applied to auction fraud, with the logic layer used to extract readable knowledge from the patterns the network identifies.

Knowledge exchanges between new media artists become the material for a model of how their creative process unfolds.

Intrusion detection in wireless sensor networks is the target: an evolving fuzzy neural network predicts the k barriers required, and the model is built to expose the reasoning behind its predictions.

Null-unineurons form the basis of an evolving fuzzy neural network directed at the identification of coronary artery disease.

Facial animation asks for resources that a small VR production rarely has. The paper sets out a pipeline for producing it under that constraint, aimed at teams working without the budget of a studio. What it offers is a sequence of steps rather than a technique.

Small and medium enterprises accumulate social network data without the analytical apparatus that larger firms bring to it. The paper proposes an opinion mining framework fitted to that situation, taking social network data as its input and the constraints of smaller firms as its design condition.

Opinions that customers post on digital media can be sorted by hand or by algorithm. Setting the two procedures against the same body of customer text, the study compares what each classification actually delivers.

A virtual reality simulator is the vehicle here for heritage education, with the visit to the site replaced by an immersive environment the student moves through. Gamification carries the learning structure, and the classroom is the setting the design assumes.

Curricular content for elementary school is delivered through an interactive augmented reality application designed for classroom use.

Co-creation practices in the game industry are examined through the case of one Brazilian developer.

Cellular automata provide the model through which the paper follows how rumors propagate inside a company.

Engagement is treated here as a mechanic rather than an outcome: something a small or medium enterprise can design into its Web 2.0 presence in order to manage what it knows about its consumers. That mechanic is tied to consumer knowledge management, with Web 2.0 tools as the channel.

Porter's algorithm is combined with the linguistic rules of an existing Portuguese stemmer to produce a stemming algorithm for the language.

Two competency models, T-Shaped and K.S.A., are set against each other in a qualitative comparison.

Market reports on the leading suppliers of enterprise content management systems are set side by side and compared.

Customer knowledge management is laid out as a process supported by Web 2.0 tools.

The congress version, in Portuguese, examines the same market reports on the companies that supply enterprise content management systems.

The companion quantitative study measures the same two competency frameworks, here named by their Portuguese acronym C.H.A.

The deposited package holds the materials that support Beyond the Black Box, released on OSF alongside the article.

Bridges between fields often exist in the literature without anyone having stated them. This paper applies Swanson's approach to that problem in art-and-technology research, searching for interdisciplinary links the published record implies but does not make. What it looks for is not what the field has said, but what its own literature would allow it to say.

Peirce's icon depends on resemblance grounded in something that was there to resemble. Generative images have no such ground, and that gap is where the paper works: what, semiotically, a synthetic image can be said to mean. The argument runs through iconicity rather than through the mechanics of the model.

Digitizing a colonial archive does not undo the conditions under which its photographs were taken; it makes them easier to reach. The paper presents a VR installation in which 196 photographs from French Indochina withdraw as the viewer approaches, each enacting one of five refusal behaviors assigned by a violence index scored by four independent coders. Made in Ho Chi Minh City, formerly colonial Saigon, the work treats computation as a means of declining the archival encounter rather than smoothing it.

Between 2010 and 2024 immersive technology went from laboratory apparatus to consumer product. The study tracks how knowledge moved across that period and where innovation concentrated, treating the asymmetries between contributors as the object of measurement. What the mapping registers is the shape of the field's growth and not its size.

Retrieval across digital museum collections runs on embedding models, so whatever those models under-represent becomes harder to find. This study measures cultural representation bias in CLIP embeddings across five museum collections, at a scale meant to make the measurement empirical rather than illustrative. The question is which cultures the representation space keeps near its center.

Timbre is arranged as a constellation the visitor walks among, giving sonic material a spatial form that can be entered. The paper sets out the virtual reality environment and the mode of navigation it asks of the visitor. At stake is what becomes audible when listening is organized as movement instead of sequence.

Georges Balandier's colonial situation and Frantz Fanon's account of colonial power are brought to bear on The Battle of Algiers. The reading treats the film as material for an anthropology of power rather than as a document of the war it depicts. What the film stages is a structure, not a sequence of events.

Generative systems arrived in creative education before the vocabulary for describing what students do with them. The paper puts forward sympoietic pedagogy as a name for that relation and as a framework for organizing human-AI co-creation in teaching. Its concern is the pedagogical arrangement rather than the tool.