You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
All readers/writers and components should have a standard switch which disables any type mapping (e.g. POS taggers and components which read POS annotations would only every generate the POS type, but not any sub-types like POS_NOUN).
When preparing data then then importing it into an annotation editor like WebAnno or INCEpTION, having the mapped types is counter productive because UIMA does not allow to change the type of an annotation. Simply deleting and recreating an annotation is also not sufficient because the annotation might be referenced by other annotations (e.g. the POS annotation is referenced from the "pos" feature of the Token annotation) and all these references would have to be updated as well to point to the new annotation. Thus, it is better in certain cases to entirely disable the type mapping and stick only to the main types (POS, NamedEntity, Lemma, etc.).
This entails that the requirement that the root annotation of constituency and dependency parse trees to be ROOT is relaxed and that they are represented as Constituent and Dependency with root being the tag value.
* master:
#1322 - Upgrade to OpenNLP 1.9.1
#1308 - integrate mystem
#1327 - Update LIF support
#1327 - Update LIF support
#1329 - Span annotations with slot features may disappear from WebAnno TSV
#1329 - Span annotations with slot features may disappear from WebAnno TSV
#1329 - Span annotations with slot features may disappear from WebAnno TSV
#1327 - Update LIF support
#1323 - File extension generated by BinaryCasWriter does not contain dot
#858 - Out-of-tagset tags should map to the generic type
#1239 - Rename NYTCollectionReader to NitfReader
#858 - Out-of-tagset tags should map to the generic type
#1317 - Standard parameter to disable type mapping
No issue. If a DKProTextContext is available, then TestRunner generates an XMI file from the processed data and stores it in the test output folder.
No issue - Log names of files with license issues to the console.
#1160 - Better support for CoNLL-U v2 (1.11.0)
% Conflicts:
% dkpro-core-asl/pom.xml
All readers/writers and components should have a standard switch which disables any type mapping (e.g. POS taggers and components which read POS annotations would only every generate the POS type, but not any sub-types like POS_NOUN).
When preparing data then then importing it into an annotation editor like WebAnno or INCEpTION, having the mapped types is counter productive because UIMA does not allow to change the type of an annotation. Simply deleting and recreating an annotation is also not sufficient because the annotation might be referenced by other annotations (e.g. the POS annotation is referenced from the "pos" feature of the Token annotation) and all these references would have to be updated as well to point to the new annotation. Thus, it is better in certain cases to entirely disable the type mapping and stick only to the main types (POS, NamedEntity, Lemma, etc.).
This entails that the requirement that the root annotation of constituency and dependency parse trees to be
ROOT
is relaxed and that they are represented asConstituent
andDependency
withroot
being the tag value.Related issues
The text was updated successfully, but these errors were encountered: