Creating, documenting, publishing and maintaining an RDF Schema vocabulary is a complex, time-consuming task. This makes vocabulary maintainers reluctant to evolve their creations quickly in response to user feedback; it prevents use of RDF for casual, ad-hoc data publication about niche topics; it leads to poorly documented vocabularies, and contributes to poor compliance of vocabularies with bestpractice recommendations. Neologism is a web-based vocabulary editor and publishing system that dramatically reduces the time required to create, publish and modify vocabularies. By removing a lot of pain from this process, Neologism will contribute to a generally more interesting, relevant and standards-compliant Semantic Web.
Anyone who wants to publish information as RDF on the Semantic Web rst faces the choice which RDF Schema vocabulary or OWL ontology to use. Some areas, such as social networks (FOAF), online communities (SIOC) or general document metadata (DC) are covered by established vocabularies. Outside of these domains, registries like SchemaWeb3 and search services like Falcons Concept Search4 assist in the task of nding vocabularies for niche topics, but what they nd might be of insu cient quality, or might not cover all required terms, and at present many areas of interest are not covered by any vocabulary at all.
In summary, most e orts to publish information on the Semantic Web rst require an e ort to create, extend or modify an RDF Schema vocabulary or OWL ontology. But this is a complex and time-consuming task in itself. It involves: { Creating the formal speci cation of the vocabulary in RDFS or OWL, { writing documentation that is clear and helpful for users of the ontology, { keeping both documents in sync as the vocabulary evolves,
{ archiving older versions of the documents,
{ de ning and maintaining mappings to related vocabularies,
{ con guring the web server in accordance with W3C best practices [
In this paper we present an online vocabulary editor and publishing system based on Drupal5, implemented in PHP and ActionScript, which will support vocabulary authors in the tasks above and thereby dramatically reduce the time required to create, publish and modify vocabularies. The work presented in this paper is in progress. 2
We de ne vocabularies as simple, \lightweight" ontologies, such as FOAF, DC,
SIOC and SKOS. They usually comprise less than 50 terms. Expressivity is
limited to RDF Schema plus selected OWL features, e.g. inverse functional
properties and class disjointness. Their value is in providing common terminology
for exchanging information between programs. The actual information is in the
RDF instance data that is expressed with the vocabulary's terms, while in more
complex ontologies, the actual information lies in the de nitions of the classes
and properties. A vocabulary is created by publishing a description of its terms
in natural using HTML or formal using RDFS/OWL language. Since classes
and properties are identi ed by URIs, it is considered a good practice to make
these URIs resolvable [
Vocabulary maintenance with text editors and custom scripts. Many popular
vocabularies such as FOAF and SIOC are maintained by a process involving
hand-authoring of RDF and HTML les and custom scripts, e.g. SpecGen7.
Often, complex custom Apache con gurations are employed to follow best practices
regarding content negotiation, MIME types and resolvable URIs [
O ine ontology editors. OWL ontology editors such as Protege [
Neologism11 is a web-based vocabulary editor and publishing platform designed to address these issues. It is currently being implemented and will soon be released as an open-source project. This section presents Neologism's current state. Public interface. To non-authenticated users on the Web, Neologism presents a very simple interface: a homepage that lists one or more vocabularies, and for each of them a vocabulary page containing some general information about the vocabulary (Figure 1), followed by the descriptions of all its classes and properties.
Editor. After a vocabulary maintainer logs in, additional links become visible on the vocabulary page and allow adding new terms, as well as editing of existing terms. Terms are created and edited through a web form (Figure 2). The form allows entry of an ID (to become part of the term's URI), label, comment, subclasses, subproperties, domain, range, disjoint classes, inverse properties, and marking a property as inverse functional. Authenticated users can also create new vocabularies and modify the vocabulary metadata.
Overview diagram. The vocabulary page provides access to a diagram that shows the vocabulary's classes and their relationships (Figure 3). The vocabulary maintainer can arrange the diagram into a sensible layout and then save its current state which will henceforth be shown to all users.
RDFS output, URIs and content negotiation. The URIs identifying classes and properties are always generated by appending the hash character and the term's ID to the URI of the vocabulary page. This makes sure that the vocabulary page is returned when these URIs are resolved. HTTP requests to the vocabulary page are subject to content negotiation. Web browsers will see the HTML variant shown in Figure 1. RDF-aware clients will receive the RDFS/OWL speci cation, either in RDF/XML or N3 syntax. In a nutshell, Neologism publishes standardscompliant vocabularies on the Web without requiring any additional e ort on the part of vocabulary maintainers.
Implementation. Neologism is implemented in PHP as a Drupal module. Drupal reduces development time by providing many features for free, such as account management. It also makes integration with a larger Drupal-based site very easy, for example to provide a news blog and discussion forum for each vocabulary. All data is stored in a MySQL database. RAP12 is used to serialize RDF/XML and N3. The PHP Content Negotiation library13 is used instead of the usual Apache rules to implement content negotiation, and Vapour14 was used to validate its correctness. The overview diagram is implemented using Adobe Flex and coded in ActionScript; the ObjectHandles and Tweener libraries are used for animation and object handling. 5
Hosted Neologism service. Currently, vocabulary maintainers must install Neologism on their own webspace. A central hosted service, which could be easily built on the Drupal platform, would remove this barrier.
Branching and revision tracking. Neologism does not yet o er revision control.
Some desirable features for vocabulary revision control are: archival of all prior
versions; grouping of several small edits into a single version to avoid putting
the vocabulary into an inconsistent intermediate state; publishing changes as a
draft before accepting them as a new version.
12 http://www4.wiwiss.fu-berlin.de/bizer/rdfapi/
13 http://ptlis.net/source/php-content-negotiation/
14 http://vapour.sourceforge.net/
Plugin system. We intentionally kept the set of supported class and property
annotations small to simplify the user experience, and don't support many
possible further annotations, such as OWL cardinality constraints, plural and
inverse labels15, multilingual labels or associating Fresnel lenses [
Consistency checking. Neologism doesn't check the created vocabulary for consistency. This can become an issue when a vocabulary is integrated with several external vocabularies. A solution could be the integration of an external reasoning service that performs consistency checks and is invoked through an API over the Web. 6
We have shown a web-based vocabulary publishing system that simpli es the process of creating, publishing and maintaining RDF vocabularies by (i) instant web-based publishing, (ii) focus on a limited subset of RDFS and OWL, (iii) avoiding instance editing or browsing, and (iv) handling URI management and HTTP content negotiation. We hope that the presented system will encourage the creation of new vocabularies and thereby contribute to a generally more interesting, relevant and standards-compliant Semantic Web.