Overview of LVG
This annotator wraps the National Library of Medicine (NLM) SPECIALIST lexical tools. It generates a canonical form for words, and this normalization step is helpful for dictionary lookup.* It also generates a list of lemma entries with Penn Treebank tags. These tags could be useful for a part of speech (POS) tagger. However, for the OpenNLP POS tagger, we use a tag dictionary rather than lemma information.
Refer to the documentation for the POS tagger annotator.
*Note: LVG adds variants that the dictionary lookup will use in an attempt to discover terms whose form in the text is not present in the dictionary database. e.g., the singular variants of plural forms, capitalization variants, etc. While LVG often increases the number of dictionary terms found in the text, there is, of course, a risk of introducing false positive returns from the dictionary.
Analysis engines (annotator) - LvgAnnotator.xml
controls whether only certain sections will be annotated by this annotator
list of sections not to be processed by this annotator
controls whether to look up information in a cache before using norm
location of norm cache file
words for which canonicalForm is never set and Lemma entries are never posted
mapping of part of speech tags used by LVG to POS tags from lexical tools to Penn Treebank tags
controls whether any lemma entries are posted to the CAS
controls whether to look up lemma information in a cache before using lvg
the location of the cache file
- lvg.properties - The LVG config file resources/lvg/data/config/lvg.properties defines the location and attributes of the LVG database and the jdbc driver used.
- LVG database - The database engine used is hsqldb. The database file included is a sample. Refer to the LVG section of the install instructions for details on how to replace the sample.