Product Changes

The objective of this release of Protege 1.4.1 is to provide bugs fixes found in previous Protege releases as well as completing feature request for formatting of the complex properties and improving performance for batch edits. 

New Feature Highlights

Although much of the work in 1.4 involved refactoring existing code to simplify maintenance and enhance performance in various spots, a number of new features have been introduced in various areas. 

• Searching:  The ability to use negation (NOT operator) in queries in the LQT

• Searching:  Lucene Match, Property Present, and Property Absent, added to the list of available search types in the LQT (which already included Exact, Contains, Starts With, Ends With, and Sounds Like).  The Lucene Match allows queries to be constructed with multiple query strings separated by boolean operators (NOT, AND), thereby extending the range and flexibility of lexical queries that can be conducted in the LQT.

• Property Restrictions:  a simple GUI dialog for creating restrictions has been added to replace the Role expression dialog for simple restrictions; this is essentially the dialog used in the OWLClasses tab of Protege.  The Role expression dialog now verifies that expressions have a valid Role Group format.

• Export of resultsets from the LQT for reports in a spreadsheet format now includes Superclasses.  Reports can now also be generated for concepts that are input as a list of codes.  Hence, the mechanisms for generating a resultset for outputting reports includes concepts specified a) from lexicographic searches (in the LQT tab), b) by their tree placement (Report Writer sub-tab), and c) by a list of codes (also from the Report Writer sub-tab).

• Batch Edits / Batch Loads:  validation of complex properties and enumerations in property values.

Behavior To Be Aware Of

A number of GForge items were created that need additional comment. With the exception of GForge items 23284, 23276, and 23057 (see below), which will be addressed in 1.4.1, the items listed here have been closed.

• 19949 Return of 60,082 results for a search

Searching complex properties or the rdf:ID (:NAME) with specific Contains queries will result in the entire vocabulary being returned, as if a single wildcard character was the query string. This is because complex properties contain embedded XML and the rdf:ID contains a full URI. For instance, the following searches will return the entire NCIt vocabulary

entity searched    query string
FULL_SYN            name
FULL_SYN            group
FULL_SYN            source
FULL_SYN            >
:NAME                 /
:NAME                 :
:NAME                 http
:NAME                 Thesaurus
:NAME                 owl

Similarly, a Contains search for "reviewer" or "source" in DEFINITION will return all the concepts with a DEFINITION property.

By the same token, doing Exact searches on the above properties and the :NAME will fail unless the complex property value or URI is entered in its entirety.

• 23284 Batch edit to delete associations needs full association rdfs:ID

This is deemed an inconsistency in behavior and is slated for a fix in 1.4.1. For now, editors need to be aware that deleting associations in batch requires that the association's rdf:ID be entered in its entirety.

• 19970 Batch edit records with sequential escaped characters fail

Exporting a txt file containing escapable characters from Excel results in Excel's escape character, the double-quote, to be exported too. Files derived from these exports used for a batch edit will fail. The files need to be cleaned up by searching and fixing the misplaced double-quotes.

• 21973 Browser does not refresh after bulk load

To increase the performance of the bulk load, the event listeners are turned off during the batch load (and edit). This is by design and results in the behavior listed in the gforge.

• 21569 Bulk Load prevents others from moving around in Protege

This applies to GForge 21632 as well (Batch load of new concepts freezes GUI for other editors)

Although issues that affected the performance of batch loads and edits were examined and their effects minimized in the 1.4.0 release, changes in the edit filters to satisfy QA requirements canceled out the performance increase. As discussed in other venues, batch edits and batch loads need to be dealt with through guidelines. As of this time, batch loads are being restricted to specific time periods of the day after editors have been notified. Batch edits can be done without this restriction but notification must be sent out so that a) competing batch edits are not done at the same time, b) editors are aware that there might be an increased lag time in Protege's response due to the ongoing batch edit.

• 22004 Error message appears during a bulk edit when there is an error in the input file. The bulk edit freezes until the OK button on the error message is pressed.

This issue has to do with concurrent edits being done by other editors on the same classes that are being batch-edited. The warning messages are the correct behavior, and can only be prevented if the classes are not edited by more than one person (or batch edit) at the same time.

• 17711 Missing warning message to the user : causes protege to freeze.

On selecting a "hasValue" modifier when owl:Thing is selected in the class widget, Protege appears to freeze. It is not frozen, the client is loading all the classes into the widget. Editors need to be aware that if "hasValue" is to be used, in the class widget the "class hierarchy" view must be selected, instead of the "instance" view.

• 23276 In Copy pane, the Preferred_Name and PT are automatically sync'ed with an improper AQ

A warning message is displayed on changing a PT or PN, e.g. like when an AQ is created. If the editor does not wish to change the PT/PN, canceling out of the action will restore the concept to its initial state and other edits will also be lost and they'll need to be recreated. This gforge is slated for modification in protege 1.4.1. For the time being, editors need to be aware of this behavior.

• 23057 Lucene indexing failure with database data

Searches in the LQT with "lucene match" might yield incorrect results if multiple concepts exist where the rdf:ID (the class name) only differs in case, e.g. "tocopherol" and "Tocopherol". The observed incorrect result is that only one class is returned instead of both. This is due to a fetch action on the database being done in a case-insensitive manner. This gforge is slated for modification in protege 1.4.1. Editors need to be aware that under these conditions "lucene match" searches (and "sounds like") won't be accurate.

List of New Features and Bug Fixes

Other Items Closed

• The defects in the following list were not reproducible, some of these were deemed fixed as a by-product of refactoring.

9298    Prompt - accept and reject icons in 'changed by' panel of 'table' view don't work

The defects in the following list were either considered invalid or did not require a fix

8608 Add Cut, Copy, and Paste options to main tree that modifies Superclass
8918 Allows paste of Preferred Name but can not delete
11600 Automatically switches Restriction when select a Restriction Property
15182 Removal of XML tags from query export
17144 Definition under the Edit tab not visible when you choose CDE/Motif and Window under Look and Feel
17571 Search under the Class browser is not cleared after the search is done.
17926 Ontology imports cause data problems when loading the explanation server from a Protege database
18583 Hierarchy View too cluttered
19383 Problems configuring the LQP
19595 Welcome to Protege User's Guide link goes to the Geek
20300 PATO concepts not found when criteria has 'Exact match' but found with criteria 'contains'.
20475 Exception while adding property (right-click - Add Property) in the copy tab.
20480 Expanding OWL:Thing using right click menu
21084 Changes tab displays edits even after cleaning the changes ontology.
21541 Prompt log output is acting differently than expected
21624 Associations pointing to PATO concepts display ID not name

CORE Product Dependencies

No caCORE dependencies. This software is used for editing and maintaining NCI Thesaurus and BiomedGT

Release History