A Flexible Distributed Architecture for Natural Language Analyzers

Abstract

Many modern NLP applications require basic language processors such as PoS taggers, parsers, etc. All these tools are usually pre-existing, and must be adapted to fit in the requirements of the application to be developed. This adaptation procedure is usually time consuming and increases the application development cost. Our proposal to minimize this effort is to use standard engineering solutions for software reusability. In that sense, we converted all our language processors to classes which may be instantiated and accessed from any application via a CORBA broker. Reusability is not the only advantatge, since the distributed CORBA approach also makes it possible to access the analyzers from any remote application, developed in any language, and running on any operating system.

Resources

Details