Page History

...

...

Overview of Dictionary Lookup

...

This can be done using class eduorg.mayoapache.bmictakes.uimadictionary.lookup.ae.FirstTokenPermLookupInitializerImpl as the lookupInitializer, and using class OrangeBookFilterConsumerImpl as the lookupConsumer, provided you have the RxNorm dictionary, and you configure the LookupDescriptorFile resource to use your RxNorm dictionary.

...

Dictionary entries need to have been tokenized the way the pipeline tokenizes the document text. For example, the lookup algorithm will not find a lookup hit if a dictionary entry is "ear, skin" but the document text contains the same text ("ear, skin") and the pipeline has tokenized that text as the three tokens "ear" "," "skin". To find a lookup hit for the three tokens, the dictionary entry should be tokenized, with a space before the comma: "ear , skin".

...

Editing dictionary lookup AE descriptors in Eclipse
The analysis engine descriptors for this annotator use elements of type configurableDataResourceSpecifier. These cannot be modified from the Parameters or Resources tabs of the Component Descriptor Editor (at least not in UIMA 2.2). To view these values or edit them, use the Sources tab or open the descriptor with a text editor.

...

A LookupDescriptorFile such as lookup/LookupDesc.xml, found in resources/, defines the dictionary(s) used, and the classes that interact with the dictionary(s). The implementation tag identifies the type of dictionary: Lucene index (luceneImpl), database (jdbcImpl), or delimited flat file (csvImpl). See class eduorg.mayoapache.bmictakes.uimadictionary.lookup.ae.LookupParseUtilities.java for implementation details.

...

...

To better understand the dictionary lookup annotator code you could start by reading the Javadoc API for the classes DictionaryLookupAnnotator.java and FirstTokenPermutationImpl.java.'

...

DictionaryLookupAnnotatorUMLS.xml

...

This is a skeleton of how you could use a dictionary contained in a database that can be accessed via a JDBC driver instead of using a Lucene index or flat file.

Refer also to the page on dictionaries in the cTAKES documentation on SourceForge.

Sample dictionaries

This project includes two sample dictionaries that are used by default:

...

The programs used to create these Lucene indexes are scripts/java/org/apache/eductakes/mayodictionary/bmilookup/dictionarytoolstools/CreateLuceneIndexForExampleDrugs.java and scripts/java/edu/mayo/bmi/dictionarytoolsorg/apache/ctakes/dictionary/lookup/tools/CreateLuceneIndexForSnomedLikeSample.java

...

To view the contents of a Lucene index, you could use a tool such as Luke.

...

To create a dictionary yourself, you could download a copy of the UMLS Metathesaurus and build upon the program mentioned above to create a Lucene index of the desired vocabulary.

Alternatively, you could use a different program in that same package that reads create a Lucene index from a pipe-delimited file :\\
by using a different program (scripts/java/edu/mayo/bmi/dictionarytools to create a Lucene indexorg/apache/ctakes/dictionary/lookup/tools/CreateLuceneIndexFromDelimitedFile.java) in that same package.