It has been a while since my last post because I’ve been researching Information Architecture. I knew that information architecture is, or should be, the lynch pin for application and network architecture. However I’ve been running up against the bias that information architecture is solely a component of application design, primarily web design, and since everything was moving to the web anyways, why would an Enterprise Information Architecture be needed. So I took a course in Information Architecture in December and came away with the following conclusions:
1. Information Architecture identifies the information that is meaningful to an organization. Information is valuable if it is relevant. Defining the information that is relevant to the organization, project, etc. is essential to making an organization effective. Defining the relationships between the information elements adds meaning, where none may have been before and creates the basis for knowledge management.
2. Key Word Search is not the complete answer. In speaking with application developers I often encounter skepticism about the value of developing an ontology to support search. Search tools are powerful enough now to to find everything.. Granted, but I don’t want to find everything – just what is relevant. It comes back to point 1. Here is a list of reasons why developing an ontology to support search is important.
According to Kent Bimson: Ontologies help to find:
- Specific names versus class names (Mary versus Person)
- Pronoun references – (Her versus Mary)
- Paraphrasing eg: Prime Minister versus Mr. Harper
- Modal verbs – (could be, possibly, etc.)
- Tenses – (was versus is)
3. Data Architects and Library trained resources should be key resources on any IM team. These two resources have very complimentary skill sets. Library trained resources understand how to classify and create ontologies, primarily in the world of unstructured information. Data architects also know how to classify and create relationships between data elements for structured information using data mapping techniques programmers can use to develop applications or solutions. The combination of these two resources covers both the structured and unstructured arena of information to provide a complete view of an enterprise’s information.
4. Terminology Matters – Each area of expertise has their own specific language to describe information and information relationships. Defining what these terms are and the semantic relationship between them will create a common language that makes it easier for teams to work together. (As an aside, the terminology and definitions that Microsoft introduced with SharePoint has muddied the terminology waters between application developers and business analysts and records managers.)
5. Use graphics and representational languages to describe relationships between elements. As we move toward a more integrated model of information exchange and sharing, being able to represent the relationships from a conceptual to a programmable level is essential (see point 4). Newer standards are being created, in particular OWL a web ontology language that is based on RDF XML AND DAML+OIL.
6. Standford University is doing a lot of work on semantics and ontologies. They have a tutorial on building ontologies and a free open source ontological application for building and maintaining ontologies called Protogé. Here’s a list from Wikipedia of other Ontology tools.
7. Service Oriented Architecture appears to be the way to go. Still working on learning more about this, but the general idea is that a call for information is routed through a service center that interprets the call, identifies all the relationships and their associated information stores, and then routes the call accordingly. While this approach reduces duplications, supports more intuitive responses, and centralizes all the relationships, it also can increase the complexity of the solution architecture and can add a layer of administration to the solution and its governance.
Feel free to add comments and additional references. It’s a big topic that requires input from all fields involved in managing information.