W3C Vocabulary Services

In order to promote the widespread interoperability of data, the W3C is beginning to offer a set of services to help people select, create, and maintain IRI vocabularies useful for creating reusable data.

Introduction

Effective data sharing requires people or systems to know how elements in a dataset are supposed to be understood. For example, a csv file can only be used by people who know what the column headings mean. When software is written to collect and analyze data, its developers need to understand these structural elements. When data is coming from many different sources, the data producers have to agree on structural elements (column names) or else the data consumers will have to learn and write code for each different style.

One technique for addressing this problem is to use web addresses (URLs, URIs, or more recently IRIs) as identifiers for elements of the structure (eg column headings). This establishes a single authoritative source of information about the identifier's meaning — the web page — while still allowing everyone who can create a website to create as many new identifiers as desired. It also allows people to easily and unambiguously refer to the identifiers, for example when recommending them to a colleague or asking a question about the meaning.

While this use of IRIs has been adopted in some technical communities, several barriers remain to wider adoption:

• Lack of vocabulary hosting. In practice, creating and maintaining a website for your identifiers can be difficult. Many organizations structure their website around marketing requirements, not this kind of technical work, and the people who need to create new vocabularies of IRIs have little immediate motivation for creating a public-facing site for their work.

• Lack of collaboration. Creating a vocabulary of identifiers that will be suitable for a range of applications usually requires input from a range of people. Finding and organizing the people with expertise in the domain of the vocabulary is one challenge; finding people with expertise in creating vocabularies is another. The result is often vocabularies developed by one or two people and suited only for narrow audiences.

• Lack of market information. There is no definitive listing of vocabularies, so it can be hard to determine whether a vocabulary for a particular purpose might exist. Even when one or more candidate vocabularies are found, it can be hard to find out how suitable they are: who else is using them, what are the relative strengths and weaknesses of the design, what software exists to support them, etc.

With the growing world-wide demand for data interoperability, these barriers are becoming increasingly problematic. Fortunately, W3C is well-positioned to address these problems. The vocabulary services outlined below build on existing W3C services and strengths. These services will make it much easier for people to obtain high quality vocabularies and help create new ones; this in turn will promote data sharing, reuse, and interoperability.

In general, these services are inexpensive to provide and can be offered for free to the public. W3C may, however, decide to charge for certain services and/or limit their use to W3C members.

ServicesVocabulary Hosting

Summary:We will promote and clarify the existing policy of giving www.w3.org/ns space to any W3C group, including community groups. The goal is to allow vocabularies useful for open data interchange to be hosted by W3C and maintained by the people who care about them.

BackgroundThe longstanding W3C policy has been to give out namespaces upon request by any W3C group. When the policy was created, groups could only be created with the approval of the W3C Advisory Committee (representing the W3C Membership), and every group included a member of the W3C staff. Since then, W3C has begun to support Community Groups and Business Groups, which can be quickly created, without membership approval, and do not have ongoing staff participation. The namespace policy was interpreted to include these new groups, as long as the "shortname" was based on the group name.

In practice, few groups have taken advantage of this policy. It is not widely known, and the process for updating the namespace document is not specified.

Plan@@@ with this in mind...

The chairs of W3C groups, including Community Groups, will have access to a web form which allows them to reserve and update namespace documents. The form will ask for some metadata, like what decision process was used by the group, and request that the group seek review from public-vocabs@w3.org before making significant updates.

At some point this interface may be expanded to provide software tools which support group development of vocabularies. It may also be extended to cover namespace documents on domains other than w3.org, in order to allow vocabularies to potentially become independent of W3C.

Vocabulary Groups

Summary: We will re-purpose the public-vocabs@w3.org group into a general group of vocabulary-development experts, with a mission to help other groups produce high-quality vocabularies. We will promote the use of Community Groups for coordination among the people interested in developing and maintaining individual vocabularies.

Background@@@ public-vocabs created....; who are the experts...?

Plan@@@ outreach?

@@@ review sessions?

Vocabulary Market InformationDirectory

Summary: We will collect, maintain, and distribute information about vocabularies, with the aim of helping people identify and decide among alternatives. This will be an open data application, freely interoperating with other suchrelated information services thatservices.

Even though at present vocabularies are generally available free of charge, we may exist. Background @@@ various ontology directories Plan @@@consider vocabulary adoption as a market, with "consumers" trying to identify "products" and choose among them. From this perspective, consumers in the current vocabulary market have very little information about available products and their features. This is hardly surprising: there is little or no "advertising", there no simple UIbusiness case for list, search, add, thumbs-up @@@ experiments"retailers" to attract and guide consumers, and there is little available "product information" as might be printed on a package.

The current market has had some "retailers" who have since gone away (schema.net), some promising newcomers (LOV), and some successful efforts in subdomains with available funding (BioPortal). We plan to improve the flow of information in this market in two complementary ways:

1. Vocabulary Directory Website. We will provide a "retail" website where people can maintain and search listings of vocabularies, along with useful metadata ("product information"). Metadata may include simple endorsements ("like", "+1", star ratings) and more detailed information like reviews, the list of open/closed issues, and the list of public users/implementations.

2. Vocabulary Market Database. The directory will be an open data synndication, usingapplication, making its internal data available for others and consuming data feeds from others. People who have existing standardsmetadata will be able to easily provide it to the directory, and prototypingpeople will be able to create new ones where necessary.interfaces for exploring and exploiting the data. The system will be architected to give the directory website no special status; people will be able to create alternative vocabulary directories (other "retail" sites) backed by the same data. (This is a "dogfood" project, using open data technologies to support the open data ecosystem.)

By making the directory an open data application, and by branding it as a W3C service, we are likely to be able to make the service essentially complete, listing all in-use vocabularies. By using an open data architecture, we avoid stiffling the market; we provide an extremely low barrier to entry for either new vocabulary producers or innovative retailers.

Copyright© 2013 W3C ^® (MIT , ERCIM , Keio), All Rights Reserved.

Date: 2013-04-25 00:41:5217:57:43 $