Digital Spaces in Ontology Standards

Here’s a comprehensive structured summary of how existing ontologies model digital/virtual spaces:

---

1. Schema.org — VirtualLocation and Place

URL: https://schema.org/VirtualLocation | https://schema.org/Place

The Key Design Decision: Separate Branches

Schema.org made a deliberate architectural choice: physical and virtual locations are NOT unified. They live in completely separate branches of the type hierarchy:

Thing
├── Place                    (physical — "entities with fixed, physical extension")
│   ├── Accommodation
│   ├── CivicStructure       (Airport, Museum, Park, Library...)
│   ├── LandmarksOrHistoricalBuildings
│   ├── LocalBusiness         (Restaurant, Store, etc.)
│   ├── Residence
│   └── TouristAttraction
│
└── Intangible
    └── VirtualLocation      (digital — "an online or virtual location")

VirtualLocation

• Type: schema:VirtualLocation
• Parent: Thing > Intangible (NOT Place)
• Status: “new” area, still seeking implementation feedback
• Definition: “An online or virtual location for attending events”
• Properties: Inherits only from Thing — name, url, description, identifier, image. No geo properties, no containment relationships, no address.
• Source issue: https://github.com/schemaorg/schemaorg/issues/1842

Place

• Type: schema:Place
• Definition: “Entities that have a somewhat fixed, physical extension”
• Rich properties: geo, latitude, longitude, address, containedInPlace / containsPlace (nested containment), maximumAttendeeCapacity, openingHoursSpecification, publicAccess
• Equivalent to: cmns-loc:Location

Hybrid Events

Schema.org bridges the physical/virtual gap only through Events, not through a unified location model:

• Event.location accepts both Place AND VirtualLocation (as an array for hybrid events)
• Event.eventAttendanceMode has three values:
  • OnlineEventAttendanceMode
  • OfflineEventAttendanceMode
  • MixedEventAttendanceMode

Assessment

VirtualLocation is extremely thin — basically a URL with a name. It has no containment model, no notion of “spaces within spaces,” no capacity, no access control. It was designed for “where do I tune in to this webinar?” not for modeling digital community spaces. The asymmetry is stark: Place has ~30 dedicated properties; VirtualLocation has zero dedicated properties beyond what Thing provides.

---

2. ActivityPub / ActivityStreams 2.0

URL: https://www.w3.org/TR/activitystreams-vocabulary/#dfn-place

Place Object Type (Physical Only)

ActivityStreams defines a Place object type with geographic properties:

• Type: as:Place
• Extends: Object
• Properties: accuracy, altitude, latitude, longitude, radius, units
• Used in activities: Arrive, Leave, Travel (geo-social activities)

This is purely physical. There is no VirtualPlace or DigitalSpace equivalent in the core spec.

How Digital Spaces Are Actually Modeled

The fediverse models digital spaces not as “places” but as Actors and Collections:

Concept	AS2 Type	How It Works
Instance/Server	Not typed — implied by domain	The server itself is the “space”
User	Person actor	Has inbox, outbox, followers, following
Group/Community	Group actor	Receives posts, re-announces to followers
Channel/Feed	Collection / OrderedCollection	Ordered sets of objects

Fediverse Enhancement Proposals (FEPs)

The community is actively extending the vocabulary for community spaces:

• FEP-1b12: Group Federation — Formalizes Group actor type for forum categories. Groups maintain follow relationships and Announce content to followers. (https://docs.nodebb.org/activitypub/fep/1b12/)
• FEP-1d80: Feed Actor — New actor type for collections of Groups (developed by PieFed, Lemmy, Mbin). Needed because Group wasn’t sufficient for hierarchical community organization.
• FEP-400e — Publicly-appended content (how posts get added to Group collections)

places.pub (Social Web Foundation)

URL: https://places.pub/

A project that bridges physical places into ActivityPub by exposing OpenStreetMap data as AP objects. Each OSM location gets an HTTPS URL (https://places.pub/{node|way|relation}/{id}) and becomes a followable ActivityPub Place object. Uses a mix of vocabularies: ActivityStreams, GeoJSON, Dublin Core Terms, vCard.

Assessment

ActivityPub has no concept of digital space as a place. Digital spaces are modeled as actors (entities that can send/receive) or collections (bags of objects). The “where content lives” question is answered by “which actor’s outbox contains it” rather than “which place is it located in.” This is a fundamentally different metaphor — the space IS the actor.

---

3. SIOC (Semantically-Interlinked Online Communities)

URL: http://rdfs.org/sioc/spec/ | http://rdfs.org/sioc/ns#

The Most Relevant Ontology for Your Question

SIOC is the only major ontology specifically designed to model online community spaces. It has no physical location concept at all — it’s digital-only.

Core Class Hierarchy

Space           — "A data Space which is the location of a set of Container data"
├── Site        — "A Site can be the location of an online community or set of communities"
│
Container       — "An area in which content Items are contained"
├── Forum       — "A discussion area on which Posts or entries are made"
│   └── Thread  — "A container for a series of threaded discussion Posts or Items"
│
Item            — "Something which can be in a Container"
├── Post        — "An article or message that can be posted to a Forum"
│
Community       — "A high-level concept that defines an online community"
UserAccount     — "A user account in an online community site"
Usergroup       — "A set of UserAccounts whose members have a common purpose"
Role            — "A function of a UserAccount within a scope"

Key Properties (Space/Container Model)

Property	Domain	Range	Description
has_space	Site	Space	Links a site to its data space
space_of	Space	Site	Inverse
has_host	Container	Site	The Site that hosts this Container
host_of	Site	Container	Inverse
has_container	Item	Container	The Container an Item belongs to
container_of	Container	Item	Inverse
has_parent	Container	Container	Parent Container (nesting!)
parent_of	Container	Container	Inverse
has_member	Community	UserAccount
has_moderator	Forum	UserAccount
has_creator	Item	UserAccount
has_reply	Item	Item	Threading
reply_of	Item	Item	Inverse

SIOC Types Module (Subclasses)

Forum subtypes: ArgumentativeDiscussion, ChatChannel, MailingList, MessageBoard, Weblog

Container subtypes: AddressBook, AnnotationSet, AudioChannel, BookmarkFolder, Briefcase, EventCalendar, ImageGallery, ProjectDirectory, ResumeBank, ReviewArea, SubscriptionList, SurveyCollection, VideoChannel, Wiki

Post subtypes: BlogPost, BoardPost, Comment, InstantMessage, MailMessage, WikiArticle

Relationship to Physical Places

None. SIOC is purely digital. It does not model physical locations at all, nor does it attempt to bridge to them. The Space class in SIOC means “data space” (a namespace/scope for content), not a physical or even metaphorically spatial concept.

Assessment

SIOC is the richest existing model for “digital spaces as containers for community interaction.” Its Container hierarchy with nesting (has_parent/parent_of) is exactly the kind of structure needed for modeling spaces-within-spaces. However, it’s from the Semantic Web era (~2007-2018) and sees limited adoption today. It was designed for interoperability between forums, blogs, and mailing lists — not for modern platforms like Discord, Slack, or Reddit.

---

4. FOAF (Friend of a Friend)

URL: https://xmlns.com/foaf/spec/

Relevant Concepts

FOAF focuses on people and their relationships, not spaces. But it has a few relevant concepts:

Concept	Type	Purpose
foaf:OnlineAccount	Class	An account in an online service
foaf:Document	Class	Any document (web page, etc.)
foaf:Group	Class	A collection of agents
foaf:based_near	Property	”A location that something is based near” (geo, not digital)
foaf:account	Property	Links Agent to OnlineAccount
foaf:accountServiceHomepage	Property	The service where the account exists
foaf:homepage	Property	Links Agent to their web page
foaf:weblog	Property	Links Person to their blog

Relationship to Digital/Physical

foaf:based_near is geographic (uses geo:SpatialThing). There’s no foaf:based_in_digital_space equivalent. FOAF models people who exist in digital spaces but not the spaces themselves. The closest it gets is OnlineAccount + accountServiceHomepage, which says “this person has an account on this service” — the service homepage is the weakest possible proxy for “digital space.”

Assessment

FOAF is complementary to SIOC (they were designed to work together). FOAF describes who; SIOC describes where. Neither models physical location in a way that bridges to digital.

---

5. Dublin Core

URL: https://dublincore.org/specifications/dublin-core/dcmi-terms/

Spatial Concepts

Concept	URI	Purpose
dcterms:spatial	http://purl.org/dc/terms/spatial	Spatial characteristics of a resource
dcterms:coverage	http://purl.org/dc/terms/coverage	Spatial or temporal topic of a resource
dcterms:Location	http://purl.org/dc/terms/Location	”A spatial region or named place”
DCMI Point	Encoding scheme	Lat/long point
DCMI Box	Encoding scheme	Bounding box

Digital Location?

No. Dublin Core’s spatial concepts are explicitly geographic — named places from gazetteers, geographic coordinates, bounding boxes. The dcterms:Location class is defined as “a spatial region or named place” and all encoding schemes (Point, Box) are geographic coordinate systems.

Dublin Core does not model URLs, URIs, or digital addresses as “locations.” A resource can have a dcterms:spatial of “Boston, MA” but there’s no standard pattern for dcterms:spatial of “the #general channel in the Cursor Discord.”

Assessment

Dublin Core treats location as metadata about resources, not as a first-class entity. It answers “where is this resource relevant?” not “what is this digital space?” No bridge to digital.

---

6. W3C / Spatial Data on the Web

URL: https://www.w3.org/2021/sdw/

Working Group Status

The Spatio-temporal Data on the Web Working Group (joint W3C/OGC) produces:

• Semantic Sensor Network Ontology (W3C Recommendation)
• Time Ontology in OWL (W3C Recommendation)
• Spatial Data on the Web Best Practices
• GeoSPARQL (from OGC)

Digital Location Work

Essentially none. The entire working group is focused on physical geography — sensors, earth observation, geospatial data. The Locations and Addresses Community Group reviews location vocabularies (GeoSPARQL, NeoGeo, schema.org) but is focused on physical addresses and geocoding.

There is no W3C working group or specification for modeling virtual/digital locations as spatial entities.

Assessment

The W3C spatial work is entirely about physical space. Digital location is simply not in scope.

---

7. Academic/Applied Research — Digital Public Spaces

New Public’s Digital Spaces Directory

URL: https://newpublic.substack.com/p/introducing-our-digital-spaces-directory

A catalog of 200+ digital social products, classified by:

Dimension	Values
Entity Type	Digital Space, Infrastructure for Spaces, Federated Infrastructure
Interaction Format	(not enumerated in public docs)
Business Model	(various)
Stage	Maturity level
Open Source	Yes/No
Civic Signals	Which of the 14 apply

The three-way Entity Type split is interesting: it distinguishes the space itself from infrastructure that hosts spaces from federated infrastructure. This is a meta-taxonomy — it classifies the platform architecture, not the space within it.

Civic Signals Framework (New_Public + Center for Media Engagement, UT Austin)

URL: https://mediaengagement.org/research/civic-signals-the-qualities-of-flourishing-digital-spaces/

14 signals in 4 building blocks — these describe qualities of spaces, not spatial structure:

Building Block	Signals
Welcome	1. Invite everyone to participate, 2. Ensure safety, 3. Humanize others, 4. Secure information
Connect	5. Cultivate belonging, 6. Bridge groups, 7. Strengthen local ties, 8. Make power accessible
Understand	9. Elevate shared concerns, 10. Show reliable information, 11. Build civic competence, 12. Promote thoughtful conversation
Act	13. Boost community resilience, 14. Support civic action

All Tech Is Human — “Healthy Digital Public Spaces”

URL: https://alltechishuman.org/healthyspaces

Collaborated with New Public on the Digital Spaces Directory. Uses the same taxonomy. Their contribution is more on the policy/practice side than on formal ontology.

Assessment

The academic/applied work treats digital spaces as sociological phenomena to be evaluated, not as entities to be modeled in a graph. The taxonomies are about qualities (Civic Signals) and platform categories (Directory), not about spatial structure or containment. There’s no ontology here — it’s a classification framework.

---

Summary Table

Standard	Digital Space Concept	Relation to Physical	Containment Model	Status
Schema.org	VirtualLocation (under Intangible)	Separate branch from Place. Bridged only via Event.location	None	Thin; URL + name only
ActivityStreams	None (spaces = Actors)	Place is physical only; digital = Group/Collection actors	Actor has inbox/outbox/collections	Active; FEPs extending
SIOC	Space > Site; Container > Forum > Thread	No physical model at all	Yes! has_parent/parent_of nesting	Rich but dated (2007-2018)
FOAF	OnlineAccount (person-centric)	based_near is geographic only	None	Complementary to SIOC
Dublin Core	None	dcterms:spatial = geographic only	None	Metadata about resources
W3C SDW	None	Physical geography only	GeoSPARQL spatial relations	No digital scope
Civic Signals	”Digital Space” (entity type in directory)	Separate concept; Signal 7 bridges to “local ties”	None (quality framework)	Applied research

---

Key Insight for AgentOS

No existing standard unifies physical and digital spaces under a common model. Everyone who has tried has either:

1. Kept them separate (Schema.org: Place vs VirtualLocation in different branches)
2. Modeled only digital (SIOC: rich digital-only model)
3. Modeled only physical (Dublin Core, W3C SDW, GeoSPARQL)
4. Used a completely different metaphor (ActivityPub: spaces are actors, not locations)

The closest to a unified model is SIOC’s Container concept — an abstract “area in which content Items are contained” — which could theoretically apply to a physical room or a digital channel. But SIOC never actually models physical containers; it just uses spatial metaphors for digital ones.

If you’re thinking about a unified Place entity that can represent both a coffee shop AND a Discord channel, that would be genuinely novel in the ontology space.