New features | ||
---|---|---|
Previous | Next | |
Snow Owl 1.6 New and Noteworthy | Sneak peeks |
Snow Owl now supports a formal concept model--a set of rules that govern how concepts can be modeled. As an initial set of concept model rules, we have incorporated the first and only release from the IHTSDO Machine Readable Concept Model (MRCM) Project Group from October 2009.
We have additionally expanded the scope of the IHTSDO concept model to include support for reference sets, descriptions, and concrete datatypes.
In the following discussion, we will use the terms from the MRCM PG.
Each constraint contains a description and additional metadata indicating the constraint author and effective date.
Constraints can be specified at the strengths identified by the MRCM PG:
Constraints can be applied to the following forms:
Concept set definitions are used for defining the constraint domain and the range of acceptable values and types of a relationship predicate. The MRCM PG release includes concept set definitions defined by specifying a concept and whether to include its descendants. We have added two more concept set definition types:
Concept model predicates constrain the relationships, descriptions, and concrete datatypes for a particular concept model domain. The predicates allow defining:
We have imported the MRCM PG Phase I (October 2009) release as an initial concept model.
We have included a constraint editor, which is intended to be a developer-level tool rather than suitable for end users. This tool is described at the end of these release notes.
Snow Owl's development framework can now build appropriate user interfaces based on the concept model. For example, it is now simple to create a generic editor that can display appropriate choices of description types, relationships, and concrete datatypes based on the concept being edited. We have used this feature to create a constraint-driven reference set editor.
A new reference set editor allows editing concepts based on their MRCM constraints. When editing a concept, the rules for description types, attributes, and concrete datatypes for the concept are determined based on the MRCM and a suitable user interface is provided.
The descriptions section is built based on the sanctioned MRCM constraints. Required description types will always appear, even if their descriptions are not set. A combo box at the bottom of the section allows adding unsanctioned description types.
The attributes section displays the sanctioned relationship types for the current concept. The possible values for each relationship type are also constrained based on MRCM rules. A blank combo box at the bottom of this section allows adding unsanctioned relationship types.
The concrete domain section displays the sanctioned concrete domains for the current concept. An appropriate widget will be built based on the datatype: a checkbox for Boolean; a calendar widget for DateTime; and fields to enter String, and numerical data types.
This section provides basic information about the concept, including the concept ID and module, its effective time, whether it is fully-defined, and its status.
MRCM rules are now used to validate both the ontology and post-coordinated expressions.
MRCM violations are detected and displayed when validating a concept. Based on the MRCM rules, the concept validator can verify the following properties of a concept:
The MRCM violations are displayed as errors in the Problems view with a short summary of the violated constraint.
Because post-coordinated expressions only define relationships, the MRCM based validation verifies only the cardinality of their relationships with the expected type and destination.
Support for the following datatypes has been added in this release:
Snow Owl will provide integrated reasoners sufficient information to classify the above datatypes. We continue to recommend the ELK reasoner and include it as the default reasoner.
To improve classification performance, we normalize datatypes before reasoning.
We used the Australian Medicines Terminology (AMT) v3.0 previous release to test the classification performance on an ontology with large numbers of numerical datatypes. We found that the full classification and semantic differencing (i.e. identifying the ontological changes) process takes about a minute on a fast desktop computer (3.4 GHz i7).
We have changed the stated and inferred relationship icons to more clearly distinguish them.
We support importing concrete domains as specified in the 'RF2 Specification Change Request - Addition of Concrete Domains' change request.
We improved our current post-coordination processes, to enhance the creation of semantically correct post-coordinated expressions and enable pre-coordination.
Snow Owl 1.6 now supports decomposing a terminology concept or a post-coordinated expression into its normal form based on the rules provided by IHTSDO. The decomposition can be invoked from the SNOMED CT Compositional Grammar editor by selecting the icon. An expression can be overwritten with its normal form, or the normal form can be saved separately to the local workspace.
Two selected post-coordinated expression can be tested to check whether one subsumes the other or they are equivalent with each other. Equivalence can occur if the expressions are in different state of decomposition. The subsumption tester dialog can be invoked from the Project Explorer's context menu by right-clicking on one or two selected expressions.
A post-coordinated expression can be saved as a pre-coordinated concept into the hierarchy by selecting the icon from the toolbar. The user needs to specify a description only, all the other properties for the concept and its descriptions and relationships are set to default values.
After saving the concept the concept editor opens to review the newly created concept definition or further continue editing.
We have added support for SNOMED CT RF2 description format reference sets, which provide metadata about description types. These reference sets specify the maximum length and format for each description type.
Snow Owl's editors and wizards now allow entering any description types present in the imported metadata.
We now import description format reference sets per the SNOMED CT RF2 specification.
We have provided an editor to view and modify description format type reference sets.
The filtered navigators and quick searches for SNOMED CT, ATC, and ICD-10 now present more relevant results when searching for short, commonly-used terms.
We are very interested in improving the relevancy of our search results. If you have comments or suggestions regarding a particular search string and the results that are returned, please send them to snowowl@b2international.com.
The Preferred and Inclusion terms of the ICD-10 codes have now been added to the indexes, which improves the search functionality and helps finding more clinically relevant results. For example, searching for "Granulomatous enteritis" now returns "Chrone disease".
The automap functionality in Snow Owl 1.6 has been extended to support newer versions of Excel for automap file generation. The supported file formats are: csv, xls and xlsx.
For each mapping the automap editor now indicates whether automatic ("Auto") or manual ("Revised") mappings have been executed. To support quality assurance the user can now set the mapping state from "Not accepted" to "Accepted" for a valid mapping. When generating the Simple map type reference set only the "Accepted" maps are included.
To support the validation of a mapping, opening a SNOMED CT concept and reviewing its definition in a SNOMED CT editor is now accessible by double-clicking the concept in the automap editor.
Before executing the automapping the user can now decide whether the entire terminology or just a certain top-level hierarchy should be queried to find the SNOMED CT equivalent concepts.
Previous | Top | Next |
Snow Owl 1.6 New and Noteworthy | Sneak peeks |