Today, regular dictionaries include 120 to 150 classifiers; the 8822-word Syllabus of Graded Words and Characters for Chinese Proficiency lists 81; and a 2009 list compiled by Gao Ming and Barbara Malt includes 126.
As an alternative to purely dictionary-based approaches, rule-based methods [ 3] involve developing regular expressions which can combine dictionary-based information with orthographic and lexical characteristics of targeted entities and their surrounding contexts.
The structure of the template includes the following main fields: Regular Expression, Dictionaries, Relations, Object Attributes and Relation Attributes.
These approaches range from linear Conditional Random Fields (CRFs) to a combination of CRFs with regular expression and dictionary matching, followed by a post-processing step to tag those chemical names in a corpus of Medline abstracts.
We note that regular expressions or dictionary words used to identify any type of chemical entity by the Regular Expression tagger were also used as a feature to identify the same type of entities by the CRFs tagger in the relevant runs.
Rule-based systems usually tackle the de-identification task with pattern matching, regular expressions and dictionary lookups [ 6- 9].
The ADESSA system, a real-time decision support service for detecting or monitoring adverse drug events, uses a regular expression and dictionary-based method to extract ADRs [ 15, 16].
Rule-based systems (HMS Scrubber, MeDS, and MIT deid system) used the original regular expression sets, dictionaries and de-identification algorithms implemented by their developers, and were directly tested with our 225 documents reference standard.
But that regular side of the dictionary is spiced up and enlivened by Roald Dahl's own richly invented vocabulary.
The integration of the output from the various recognition approaches used in a run (CRF, dictionary, and regular expression matching) is done through a post-processing step.
