Check out the new USENIX Web site. next up previous
Next: Head/Modifier frames Up: The AGFL Grammar Work Previous: Example: transduction

The EP4IR grammar

The ``English Phrases for IR" (EP4IR) grammar is a reasonably complete grammar of English, concentrating on the description of the noun phrase and the verb phrase. The grammar is provided with a large lexicon, providing detailed Part-Of-Speech information. The grammar is quite robust against incorrect input and unknown words. The EP4IR grammar and lexicon were developed in the [PEKING project] for Information Retrieval applications, and they are released along with the AGFL system.

From the grammar and lexicon, an English parser can be generated automatically using the AGFL system, which produces as its output not parse trees but Head/Modifier frames, more or less as described above. The following picture illustrates the generation of a parser and its use.


 
Figure: Generating and using a parser
\begin{figure*}
\begin{center}
\epsfxsize=12cm\epsfbox{sys.eps}\end{center}\end{figure*}

The HM frame representation is a very good starting point for diverse applications, and the transduction can be adapted with relative ease to yet other applications.

The grammar still many details of the language, which will hopefully seduce linguists to propose additions and improvements.



 

Kees Koster
2002-05-01