A communiqué from David Steinberg reports on the version 0.1 release of a specification for the Exchangable Faceted Metadata Language (XFML). The draft specification defines an XML format which "allows for easy creation of advanced, automatically generated navigation for your website; you can even automatically generate links to related topics on other websites. It also allows for merging of metadata between different websites. XFML is an open, free format, like SOAP or RSS, designed to be easy to understand, and easy to code for, yet powerful and flexible... XFML borrows many of the ideas of [XML] topic maps: it works with faceted metadata, although you could easily use it with a classical hierarchical taxonomy by using only one facet. Conceptually, in XFML metadata is separated (as topics) from content: you create a map that exists on its own, regardless of whether you have any content that relates to the map. This provides a lot of power in that it allows you to work with your map as an entity of its own, you can import facets of other maps, merge topics and so on." XFML is based upon the notion of "imperfect and ever changing taxonomies: with its strict separation of metadata and content, the metadata in an XFML map can evolve more easily than in most current CMS systems where metadata is intertwined with content definition, and where adjusting the structure of the metadata typically involves a lot of work."

From the version 0.1 spec: "XFML is based on the idea of faceted metadata. Faceted metadata schemes are generally more powerful, flexible and useful for websites than more classic hierarchical taxonomies. Taxonomies published as XFML can be reused by other people (unless there are license restrictions). XFML maps are designed so they can be easily merged... Publishing XFML is conceptually similar to publishing RSS: you get many of the same advantages: it is easy, and you get a lot of incoming links and additional functionality possibilities that would otherwise be extremely difficult to obtain..."

From the website description: "XFML addresses the following issues:

• It is impractical to create a centralized metadata store for the web. However, we need shared metadata to make the web easier to get around in. XFML takes a distributed approach, yet allows individual authors to connect their metadata schemes by merging topics. Thus it allows for creation of a distributed, loosely connected metadata network that fulfills much of the same functions a centralized store would.
• Taxonomies are labour intensive to create. XFML allows taxonomies to be easily shared and published.
• Taxonomies are created by fallible humans and subject to change. XFML is based around imperfect and ever changing taxonomies: with its strict separation of metadata and content, the metadata in an XFML map can evolve more easily than in most current cms systems where metadata is intertwined with content definition, and where adjusting the structure of the metadata typically involves a lot of work.
• Faceted taxonomies are generally more powerful for websites than classic hierarchical taxonomies. XFML is based on facets.
• Most current content management systems implement a certain level of metadata already, yet there is no standard way of publishing this metadata, let alone make meaningful connections between metadata in different systems. Allowing meaningful metadata connections between separate systems is what makes XFML so powerful.
• Easy to implement standards get adopted faster. XFML is not perfect, neither does it try to be a comprehensive solution for all possible metadata needs. It is designed to be useful in the real world from day one, and easy to understand and implement..."

From the 'Information Architects' slide presentation:

"... it's about faceted metadata... XFML is based around imperfect and ever changing taxonomies: with its strict separation of metadata and content, the metadata in an XFML map can evolve more easily than in most current cms systems where metadata is intertwined with content definition, and where adjusting the structure of the metadata typically involves a lot of work. Most current content management systems implement a certain level of metadata already, yet there is no standard way of publishing this metadata, let alone make meaningful connections between metadata in different systems. Allowing meaningful metadata connections between separate systems is what makes XFML so powerful. XFML consists of topics and pages... A topic can be anything at all. 'Information architecture' could be a topic. Or 'faceted classification'. Or 'Feelings of love'. Anything you can possibly think of could be a topic. Which topics you choose for your XFML map (we call it a map) is up to you and depends on how you want to organise the information. Topics are organised in mutually exclusive facets... you can have hierarchy within a facet; topics within a facet don't have to be mutually exclusive... An XFML map can also contain pages. With each page, it can have a list of topics that OCCUR on that page; these are called occurrences... So you can add a webpage to an XFML document (by adding its URL) And then indicate which topics occur on that webpage And also what type of occurrence it is. Examples of occurrence types are Webpage, Picture, Weblog entry, Discussion, Mailing list archive, etc, and you can make up your own occurrence types. You can indicate that your topic is the same as someone elses topic on another website (in another published XFML document) by merging them. Merging is simply done by indicating the URL of the XFML map, plus adding #nameopftopic You can merge a topic with lots of other topics in lots of maps Merging is what lets your website automatically generate links to related content on other sites... Topics organised in facets; topics can be merged between maps; pages with occurrences (and occurrencetypes)..."