09.04.2013 Views

SNOMED CT® Release Format 1 (RF1) Guide - ihtsdo

SNOMED CT® Release Format 1 (RF1) Guide - ihtsdo

SNOMED CT® Release Format 1 (RF1) Guide - ihtsdo

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Example: Excluding the Fully Specified Name field from the <strong>Release</strong> <strong>Format</strong> 1 Concepts table<br />

increases normalization of the data, since the Fully Specified Name is also available in the Description<br />

table.<br />

• Less normalized representations of the model.<br />

Example: Replicating the | is a | Relationships with the source and target ConceptId fields reversed<br />

decreases normalization of the data but might have some performance benefit.<br />

• Omission of some of the tables.<br />

Example: If a terminology server is not required to support cross mapping, the components required<br />

for this functionality may be omitted.<br />

• Replacement of some of the supporting tables with proprietary alternatives.<br />

Example: The word search support tables could be completely replaced by other tables or indices<br />

generated by the terminology server when loading the distribution files.<br />

4.1.3. Non-relational structures<br />

Although the primary distribution format is relational, this does not require terminology servers to utilize a<br />

relational database as the primary or only storage format. The requirements for terminology services may<br />

also be met by representing some or all of the distributed data in other forms including object-oriented<br />

databases, Extensible Mark-up Language (XML) and/or proprietary data structures. These structures may<br />

be used separately or, in some cases, in combination with a relational database.<br />

A possible conclusion from analysis is that most terminology services are concept-centered. One structure<br />

that reflects this is a hierarchical design in which all the information about each concept is bound together as<br />

a discrete unit.<br />

Table 235: Representing <strong>SNOMED</strong> CT in a hierarchical structure<br />

<strong>SNOMED</strong> CT<br />

[1..n] Concepts<br />

[1..n] Descriptions<br />

[1..n] Subsets related to the Description)<br />

Including support for language and term usage subsets.<br />

[1..n] Relationships<br />

[1..n] Subsets related to the concept)<br />

Including support for subsets and crossmaps.<br />

Terminology services <strong>Guide</strong> (<strong>RF1</strong>) | 301<br />

Such a structure needs support from indices and/or other data access tools to allow searches. However, it<br />

has the merit that whenever a concept is accessed, all the relevant information is immediately at hand without<br />

following relational joins to multiple rows in other tables.<br />

© 2002-2012 International Health Terminology Standards Development Organisation CVR #: 30363434

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!