Changes between Version 1 and Version 2 of RdsWipWorldView/PathsToInteroperability
- Timestamp:
- 08/27/08 15:46:59 (16 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
RdsWipWorldView/PathsToInteroperability
v1 v2 14 14 = RDS/WIP World View: Paths to Interoperability = 15 15 16 There are also at least three common approaches to interoperability :16 There are also at least three common approaches to interoperability - these are outlined here, making analogy to human languages: 17 17 18 1. Force everyone to speak your language natively. That19 means push ISO 15926 definitions out into every software20 application that counts in your domain and force software18 1. Point to Point: translate directly from any language to every other 19 language as needed. This approach is costly and error prone for 20 everyone. 21 21 22 developers to adapt to its structures and way of thinking. 22 2. Lingua Franca: collectively identify a language as usefully common 23 and translate only to and from it. The burden on a lingua franca 24 in terms of solving engineering problems is that it must have the 25 expressive power to represent information from many different sources. 23 26 24 2. Translate from your language to every other language that 25 you need to. Costly for everyone and a problem that I26 think ISO 15926 was conceived to redress by the former27 or the latter options.27 3. Mandated Language: force everyone to speak one language natively. 28 That means push definitions out into every software application 29 in the domain and force software vendors to adapt to its structures 30 and way of thinking. 28 31 29 3. Collectively identify a language as usefully common and 30 translate only to and from it. And again, ISO 15926 has 31 a role to play here as that common language. 32 The RDS/WIP provides support for all of these approaches. In all 33 cases, it provides a place to publish reference data for the language and mappings to and from other languages. 32 34 33 What I think is important though, is that for the latter 34 the *cost* of translating peer to peer or peer to common, 35 by investing in definitional machinery that can then be 36 used to create a cheap, strong translation engine. 37 38 If we take the top-down approach and run it to its 39 conclusion right down at the FOL level on existing 40 templates, and 41 42 In one case, you force everyone to speak your language. 43 44 45 46 47 48 35 Note however that the mappings themselves must be written in some sort of language, and in many cases, it is the expressive power of at least one of the addressed languages that determines how precise and fidelity-retaining a mapping can be. 49 36 50 37 ----