4. Knowledge Representation

To find the Turtle file of our ontology, navigate to the Github repository.

4.1 The Ontology

Our ontology formalizes aesthetics as complex cultural entities that emerge through community-driven creative processes on digital platforms. By modeling their characteristic elements and behavioral patterns, we can better understand how contemporary aesthetics differ from traditional subcultures in their formation and dissemination, and how they offer new popular ways of being creative through the reuse and recontextualization of past creative products.

Aesthetics are abstract, dynamic, and modern concepts. With our ontology, we aim to map their characteristic elements, relevant behaviors, and the creative processes involved in their creation and dissemination. The previous chapters introduced three theoretical frameworks that helped us create these focus points and position these intangible cultural commodities within creative theory.

There have been numerous resources classifying what belongs to an aesthetic and what does not. For example, we made deliberate use of the independent, crowd-sourced Aesthetics Wiki for identifying features of our selected aesthetics. Essentially, this classification task represents the genesis of aesthetics themselves. Aesthetics inherit from their 'parent class' of Subcultures the practice of inclusion and exclusion. This practice operates through various cultural forms such as art, music, and clothing, often based on ethnicity, interests, and social background.

Undeniably, these types of social groupings are extremely similar. Yet recent aesthetics such as Clean Girl, Y2K, and Dark Academia demonstrate distinctive characteristics through Metaconvergence. This term is coined in the earlier mentioned CreOn ontology and has become an increasingly popular method of expressing creativity on digital platforms. To clarify this concept within our framework, we interpret the term in our ontology core as follows:

"Metaconvergence is the process by which individuals or groups combine elements from multiple subcultures or aesthetics to create new, hybrid forms of expression. This phenomenon is particularly prevalent in digital spaces, where the rapid exchange of ideas and cultural artifacts facilitates the blending of diverse influences. Metaconvergence allows for greater creativity and innovation, as individuals draw upon a wider range of sources to inform their aesthetic choices. It also reflects the fluid and dynamic nature of contemporary culture, where traditional boundaries between subcultures are increasingly blurred."

Bricolage, a creative technique within our CreativeProcess class, appears similar to the Montage and Collage art techniques that gained significant popularity during the Dada art movement in the first half of the twentieth century. Reusing existing visual elements is not an uncommon phenomenon in art and creativity. However, contemporary aesthetics like Y2K operate differently. Y2K's reproduction of a 2000s 'aesthetic' evokes particular emotions such as nostalgia or a sense of belonging to a period that users may have never even experienced. Creative products from the past are recontextualized at rapid speed, not just by one dominant artist but by an anonymous web community. Often, the motive is not protest against dominant culture but rather a means to gain acknowledgment for stylistic efforts and to shape identity.

From here, with our ontology, we aim not only bring aesthetics into existence but also formalize the specific creative processes present in their emergence. It is valuable to move beyond understanding aesthetics as predominantly formalistic subcultures. Instead, we can observe how they facilitate a modern, grassroots, community-driven approach to identity formation and art making.

We initially selected six aesthetics and adapted our model to their characteristics: Y2K, Cottagecore, Dark Academia, Grunge, Old Money and Clean Girl. Our selection aimed to cover a variety of aesthetics that address different temporal contexts and demonstrate distinct creative phases. Notwithstanding, we recognize that there is considerable overlap since their usage and emergence occur around the same time period. Therefore, our analysis will show that most of them use the same types of media channels to disseminate the aesthetic. This shared infrastructure highlights the interconnected nature of contemporary aesthetic movements and their collective reliance on digital platforms for expression and growth.

4.2 Modeling

In this section we will describe how we used LLMs to model our ontology using a bottom-up extreme design methodolgy and CLOWN. During this phase, we systematically generated definitions, descriptions, and user scenarios for each aesthetic. This generated content informed the development of competency questions, which provided the structural foundation for constructing the ontology.

4.2.1 User story & LLMs

First stop: Context and Literature State of the Art (CLOWN).

Our search for a model started by searching for interesting concepts that should be defined in the model. Here are some examples of the prompts we fed to the LLM and the output it provided us.

💁‍♀️Please give a definition of an "aesthetic."

💁‍♀️Please give me 10 examples of aesthetics.

When we arrived at our selection of six aesthetics (Y2K, Cottagecore, Dark Academia, Grunge, Old Money and Clean Girl), we aimed to explore its interpretation of these phenomena further.

💁‍♀️Please give me a description and user scenario for our selection of aesthetics.

4.2.2 Classes and properties

During the development of our ontology, we repeatedly refined classes and properties to create an organic structure while maximizing the ontology's informativeness. We began by identifying all possible properties that could connect aesthetics with their elements, resources, influencers, and other related concepts. In the initial "draft" phase, we created over 100 properties. Step by step, we systematically refined them, removing or replacing properties with the most relevant ones. For instance, properties like hasVisualElement/hasAudioElement and similar were replaced with the more concise characterizedBy. Below you can explore the complete list of classes, subclasses, properties and their descriptions.

4.2.3 Competency questions

To compile a list of the most representative questions for our ontology, we also used an LLM to simulate a potential user's perspective.

💁‍♀️Imagine that you are a social anthropologist doing research about internet aesthetic phenomena, what questions would you like to ask first?

PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
    PREFIX : <http://webprotege.stanford.edu/>

    SELECT ?aestheticName ?timePeriodName
    WHERE {
    ?aesthetic a :Aesthetic ;
                rdfs:label ?aestheticName ;
                :hasTemporalContext/rdfs:label ?timePeriodName .
    }

PREFIX wp: <http://webprotege.stanford.edu/>
    PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
    PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

    SELECT ?aesthetic ?element ?elementLabel ?elementType ?typeLabel
    WHERE {
    ?aesthetic a wp:Aesthetic ;
                rdfs:label "DarkAcademia"@en ;
                wp:characterizedBy ?element .
    
    OPTIONAL { ?element rdfs:label ?elementLabel }
    
    ?element rdf:type ?elementType .
    OPTIONAL { ?elementType rdfs:label ?typeLabel }
    
    FILTER (?elementType != owl:NamedIndividual)
    }

PREFIX wp: <http://webprotege.stanford.edu/>
    PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

    SELECT ?phase ?phaseLabel 
        (GROUP_CONCAT(DISTINCT ?processLabel; SEPARATOR=", ") AS ?processes)
        (GROUP_CONCAT(DISTINCT ?processType; SEPARATOR=", ") AS ?processTypes)
    WHERE {
    ?phase rdfs:label ?phaseLabel .
    ?phase wp:involvesProcess ?process .
    
    OPTIONAL { ?process rdfs:label ?processLabel }
    OPTIONAL { ?process a ?processType }
    
    FILTER regex(?phaseLabel, "Y2K", "i")
    }
    GROUP BY ?phase ?phaseLabel

PREFIX wp: <http://webprotege.stanford.edu/>
    PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

    SELECT ?aesthetic ?aestheticLabel 
        (GROUP_CONCAT(DISTINCT ?emotionLabel; SEPARATOR=", ") AS ?emotions)
    WHERE {
    ?aesthetic a wp:Aesthetic ;
                rdfs:label ?aestheticLabel ;
                wp:involvesEmotion ?emotion .
    ?emotion rdfs:label ?emotionLabel .
    }
    GROUP BY ?aesthetic ?aestheticLabel

PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
    PREFIX : <http://webprotege.stanford.edu/>

    SELECT ?aestheticName (GROUP_CONCAT(?mediumName; SEPARATOR=", ") AS ?platforms)
    WHERE {
    ?aesthetic a :Aesthetic ;
                rdfs:label ?aestheticName ;
                :spreadThrough ?medium .
    ?medium rdfs:label ?mediumName .
    }
    GROUP BY ?aestheticName
    ORDER BY ?aestheticName

PREFIX : <http://webprotege.stanford.edu/>
    PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
    PREFIX base: <urn:webprotege:ontology:04d732ad-5fda-4ee8-89b2-5643bf6d86da#>

    SELECT ?aestheticName ?communityName (STR(?ageRange) AS ?age)
    WHERE {
    ?aesthetic a :Aesthetic ;
                rdfs:label ?aestheticName ;
                :experiencedBy ?community .
    ?community rdfs:label ?communityName .
    OPTIONAL { ?community base:ageRange ?ageRange }
    }

PREFIX : <http://webprotege.stanford.edu/>
    PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

    SELECT ?aestheticName (STR(?productName) AS ?product) 
        (GROUP_CONCAT(DISTINCT ?productTypeLabel; SEPARATOR=", ") AS ?productTypes)
    WHERE {
    ?aesthetic a :Aesthetic ;
                rdfs:label ?aestheticName ;
                :resultsIn ?product .
    ?product rdfs:label ?productName .
    
    ?product a ?productType .
    FILTER (?productType != <http://www.w3.org/2002/07/owl#NamedIndividual>)
    
    OPTIONAL { ?productType rdfs:label ?productTypeLabel }
    }
    GROUP BY ?aestheticName ?productName
    ORDER BY ?aestheticName ?productName

PREFIX webprotege: <http://webprotege.stanford.edu/>
    PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

    SELECT ?inspiredAesthetic 
        (GROUP_CONCAT(DISTINCT ?inspirationSource; SEPARATOR=", ") AS ?inspirationSources)
    WHERE {
    ?aesthetic a webprotege:Aesthetic ;
                rdfs:label ?inspiredAesthetic ;
                webprotege:inspiredBy ?source .
    OPTIONAL { ?source rdfs:label ?inspirationSource }
    FILTER (isIRI(?source))
    }
    GROUP BY ?inspiredAesthetic
    ORDER BY ?inspiredAesthetic