Rather than imagining a Semantic Web where the meaning of things is determined as in a formal, authoritative taxonomy, my concept of a Semantic Content Management System proposes that by proposing some unifying standards for expressing semantic content, multiple "directories of meaning" could be constructed to work cooperatively.
For purposes of example, we won't begin by discussing news. Imagine instead that a microbiologist working in a lab studying the DNA of a particular pathogen wishes to publish the results of her most recent study. We must also suppose that she has been given a Semantic Content Management System, which allows her to construct her own directory of meaning for everything she has learned about her subject.
If our microbiologist simply imagines an information structure for her directory, it will be of little value. So the SCMS she uses will come with a standard information template in place, and will use familiar wizard-style interfaces to walk her through the creation of her directory. As she writes and edits her article, she will be prompted to define terms -- a lot of work at the beginning, but with each valid expression of meaning, the job gets easier. This is because the SCMS remembers phrases and associates them with her definitions within her directory of meaning, automatically assigning semantic tags to repeat appearances unless she over-rides the tag.
In some instances, our microbiologist will add lots of meanings at once -- not by writing natural language text, but by ingesting data files. Because of the architecture of spreadsheets, every piece of data they contain may be expressed as an RDF triple.
At the conclusion of her writing, the microbiologist will have produced both a directory of meaning -- that is, a formal, machine-readable expression of every concept or datapoint she developed in her study -- as well as a meaning model that consists of the metadata embedded in and around her natural language text article. Her directory will be hosted on a server, providing a unique address for each URI it contains. The meaning model contained in the natural language text document that communicates her study will reference that directory no matter where it is published, no matter where it appears.
Another microbiologist working in this field will also be free to create a custom directory of meaning, but because most of the tasks required to create a directory of meaning are redundant, it would be in the next scholar's best interest to make use of the definitions provided by the first microbiologist. In this manner will directories of meaning tend to become self-propagating.
And because terms will be standardized and published, when it becomes necessary over time to edit or rebuild sections of a directory of meaning, many of the changes will be easily replaced across all instances -- a tricky task in natural language text, but less so in a directory of explict meanings and terms. Poetic license will not be a concern.
Over time, scientists would likely find that their directories would be easier to create and manage if common terms and definitions were maintained in curated directories of meaning. Perhaps the journal Cell will publish its directory of meaning. Perhaps an association of scientists working in the field will take on the task. What matters is that the individual directories are capable of working together.
In this sense, directories of meaning will be extensions of basic semantic concepts and techniques: controlled vocabularies, taxonomies, ontologies, schema. However, there is nothing implicit in the concept of a directory of meaning that requires it to be interoperable with any other. The incentive to make directories interoperable comes from mutual benefit. And while this incentive exists today, a user-friendly interface for putting it in action does not.
To that end, and because creating and maintaining semantic structures can be an expensive, time-consuming task, many organizations will choose to subscribe to general directories of meaning for their common terms, maintaining and hosting a directory only to those definitions that are unique to its area of research, work and publishing. These directories will be "read into" the user's SCMS.
In a journalistic application of directories of meaning, individual news organizations would create new entries in their directories of meaning with each news item they produce. In the short term, this would be certain to create competing directories and numerous examples of differently structured statements of the same fact. Over time, however, it will be likely to create shared standards that will allow for greater interoperability.
Recent Comments