woensdag 6 juni 2012

Parsing English Sentences

My company Sirius Computing has improved the Sirius English Text Analyser with grammar and phraseology support. The 'Grammar' tab makes it possible to manage grammatical patterns (constituents and grammatical categories). Each grammatical pattern may be linked to several examples. The following screenshot shows the list of grammatical patterns for the constituent "direct object" that begins with the grammatical category "article".
Via the tab "Text analysis" it is possible to manually annotate constituents. The following screenshot shows, for example, how a direct object can be added as a grammatical pattern.
The 'Phraseology' tab makes it possible to manage verb phrases and term + preposition pairs. Verb phrases are managed separately because the verb and adverb or preposition can be separated by other words. Term + preposition pairs are used to extend phrases with prepositional phrases. For example, "The title of the book" is detected as a noun phrase because the term + preposition pair [title; of] exists.
Phrases and multi word expressions can also be manually annotated via the tab "Text analysis". Multi word expressions (example: as soon as) are managed just like other terms in the Terminology tab.
The English Text Analyser uses the information of grammatical patterns and phrases to automatically annotate phrases and parse English sentences into clauses and constituents.