Semantic Web: The Story So Far
Ian Horrocks · 2007 · Proceedings of the 2007 International Cross-Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/1243441.1243469
Summary
This W4A 2007 keynote paper by Ian Horrocks of the University of Manchester provides an overview of the Semantic Web vision and the progress made toward realizing it, with particular focus on the Web Ontology Language (OWL). The paper opens by identifying fundamental limitations of the current Web: content is primarily designed for human consumption using presentational HTML markup, making it difficult for automated processes to understand the meaning of web resources. This problem is analogous to challenges faced by users with cognitive or sensory impairments who struggle to interpret visually-oriented content. The Semantic Web aims to address this by adding semantic annotations — machine-readable descriptions of meaning — to web content, transforming the Web from a linked document repository into a distributed knowledge base. The core enabling technology is ontologies: formal vocabularies with precisely defined terms and relationships that allow applications to understand what web content means rather than merely what it looks like. The paper traces the philosophical roots of ontology from Porphyry's classical tree of categories through to modern computational ontologies, then details how OWL was developed by the W3C as a standard ontology language based on Description Logic (DL). OWL uses an RDF-based syntax and provides constructors for building complex concept descriptions including intersection, union, complement, existential and universal quantification, cardinality restrictions, and enumerated classes.
Key findings
OWL had rapidly become a de facto standard for ontology development across diverse fields including biology, medicine, geography, geology, astronomy, agriculture, and defense by 2007. A key factor in this success was the availability of sophisticated tools with built-in reasoning support — tools like Protege, SWOOP, and TopBraid Composer, which use DL reasoners (FaCT++, Racer, Pellet) to detect inconsistencies, identify synonymous classes, compute class hierarchies, and answer complex queries. The paper highlights a concrete example of reasoning's value: when the Medical Entities Dictionary (MED), a large ontology with 100,210 classes and 261 properties used at Columbia Presbyterian Medical Center, was converted to OWL and checked by a reasoner, it revealed "systematic modelling errors" and missed subClass relationships that "could have cost the hospital many missing results in various decision support and infection control systems." Challenges for the future included balancing increased expressive power with scalability — the need to reason over ontologies with tens or hundreds of thousands of classes and data sets with millions of individuals. New reasoning systems like the OWL Instance Store (combining DL reasoning with relational databases), HermiT (using hypertableau techniques), and Kaon2 (reducing OWL to datalog programs) were being developed to meet these scalability challenges. The proposed OWL 1.1 extension addressed user-requested features including qualified cardinality constraints and transitive part-whole relations.
Relevance
While this keynote is not directly about accessibility, it was presented at W4A for good reason: the Semantic Web vision has profound implications for web accessibility. By making web content machine-understandable through semantic annotations, the same content can be automatically adapted, transformed, and presented in ways suited to different users' needs — whether that means generating text descriptions of visual content, restructuring navigation for screen reader users, or simplifying complex information for users with cognitive disabilities. The ontological approach to knowledge representation provides a foundation for accessibility-related ontologies that can formally describe user capabilities, device characteristics, content properties, and adaptation rules. The paper's observation that HTML markup is "primarily intended for presentation" and creates difficulties for automated processes directly parallels the accessibility argument for semantic HTML and ARIA. The subsequent development of schema.org, linked data, and knowledge graphs built on these foundations has increasingly intersected with accessibility — for example, enabling richer alt text generation, more intelligent content adaptation, and better assistive technology interoperability. For accessibility practitioners, understanding the Semantic Web provides context for why semantic markup matters: it's not just about following rules, but about making content meaning explicit so that technology can bridge the gap between how information is presented and how individual users need to receive it.
Tags: semantic web · ontology · OWL · knowledge representation · web standards · linked data
Standards referenced: OWL · RDF · DAML+OIL · OIL