Information architecture

How information is organised, structured, and labelled to enable discovery, navigation, and use. Information architecture establishes the structural foundation that determines what can be found, how it connects, and which navigation behaviours are possible.

Scope

• Organization schemes — how information is grouped and categorized
• Labeling systems — how information is represented and named
• Navigation systems — how actors browse, move through, and retrieve information (see structure compositions)

Organization schemes

How to group information. The fundamental choice is between closed, predetermined categories and open, multi-dimensional classification.

Enumerative schemes establish complete, predefined category sets where every concept has a designated place. These closed systems work well for stable, bounded domains—alphabetical listings, chronological timelines, geographical hierarchies, topical taxonomies. The structure is rigid but consistent, with single correct placement for each item. Use enumerative schemes when content is stable, categories are well-understood, and consistency matters more than flexibility.

Faceted schemes build categories from multiple independent dimensions rather than single hierarchies. Items are described by combining semantic facets—time, place, topic, task, audience, format, etc. This openness expresses emerging concepts through combination. Use faceted schemes when content evolves rapidly, or actors approach information from different angles.

Real systems typically combine both approaches: enumerative structure provides stable backbone, faceted attributes enable flexible filtering and discovery.

Structure and relationships

How information elements connect determines what paths through content exist. Four common structural patterns shape how information relates and how actors navigate through it.

Hierarchies

Parent-child relationships where items nest within categories. Use hierarchies when content naturally fits into containment structures—product catalogs, file systems, organizational charts. Hierarchies support multilevel tree navigation and enable top-down browsing from general to specific.

Clear hierarchies reduce search depth and provide familiar mental models. Strict hierarchies assign single parents to each item, while polyhierarchies allow multiple parents when items legitimately belong in several categories.

Design challenges emerge at boundaries: where does something belong when it fits multiple categories? How deep should the structure go before actors get lost? What granularity makes sense at each level? Items that don’t fit neatly often get forced into inappropriate categories or hidden in “miscellaneous” sections.

Most hierarchies use controlled vocabularies—defined term sets that reduce ambiguity and support precise filtering. As domains evolve, these vocabularies require maintenance to stay current.

Relational structures

Items are defined through relationships between them. Use relational structures when items connect through multiple types of relationships beyond simple containment.

One-to-many relationships create natural hierarchies where parents have multiple children but each child has a single parent. These are efficient for query and retrieval, matching how many systems naturally organize.

Many-to-many relationships allow items to relate to multiple others without ownership constraints. This flexibility enables multiple browsing paths and supports faceted navigation, though it requires join tables or link structures to implement.

Entity-relationship thinking provides the conceptual foundation: entities are things with attributes, relationships define how entities connect (authorship, participation, association), and attributes capture properties (dates, names, statuses, quantities). This thinking informs both data models and interface structure, bridging technical architecture and user experience.

Hypertext structures

Associative, non-hierarchical connections based on semantic relationships rather than containment. Use hypertext structures when exploring conceptual connections matters more than following prescribed paths.

Links can be structural (navigation through hierarchy), associative (related content, “see also”), temporal (previous/next, chronological). Multiple paths to the same content enable exploring and serendipitous discovery. Dynamic hyperlinks adapt based on context or actor behaviour. Beyond its structural role, the authored link carries a rhetorical register (see Prose § Link rhetoric).

Social structures

Emergent organization from collective activity rather than top-down design. Use social structures when organic, actor-driven categorisation provides value—social bookmarking, community-tagged content, collaborative filtering, recommendation systems.

Tags and folksonomies enable bottom-up categorisation where many actors create labels reflecting actual usage patterns. The organic vocabulary adapts quickly to new concepts and reveals how people actually think about content. This flexibility comes with inconsistency—different actors use different terms for the same concept. Social structures typically supplement rather than replace controlled vocabularies, adding flexibility to formal organization.

Network effects shape what becomes visible through popularity metrics, recommendations, and collective curation. Emergent hierarchies form from link patterns as frequently referenced items rise in prominence. Social proof influences discovery—actors trust what others have validated. These patterns create feedback loops: popular items become more visible, which makes them more popular.

Labeling systems

Labels create the interface between system structure and actor understanding. Consistency matters—use the same term for the same concept throughout, distinguish different concepts with different terms, align with actor vocabulary rather than internal jargon. Labels appear across contexts—navigation (menus, buttons), structural (headings), contextual (in-content links), and metadata (tags, search terms)—each demanding appropriate specificity for its use.

The central tension is control versus flexibility. Controlled vocabularies (taxonomies, thesauri, authority files) provide consistency, synonym management, and governance, but require maintenance as domains evolve and can’t adapt organically to emerging concepts. Uncontrolled vocabularies (tags, folksonomies) adapt quickly and reveal actual usage patterns, but produce fragmentation (e.g. “JavaScript” becomes “js”, “javascript”, “JS”, “Java Script”).

Labels must match actor mental models shaped by domain expertise, cultural context, and prior systems. Multilingual contexts complicate this—direct translation often fails when concepts don’t map cleanly across languages.

Resources & references

• Rosenfeld, Morville, Arango (2015) Information Architecture: For the Web and Beyond, 4th ed.
• Morville, Peter (2005) Ambient Findability
• NN/g / The Difference Between Information Architecture (IA) and Navigation