Latest revision as of 18:37, 25 June 2008

Performs a morphological analysis over unrestricted text and easily integrates a Natural Language Processing System as it regroups words and their constituents according to rewriting rules; changes the tags according to correspondence rules; and splits phrases according to a list of separators. Works over real texts, such as newspapers and books.

Goals

PAsMo rewrites the text according to the morphological features of is words and a set of given rewriting rules. PAsMo is also used to translate tags, easing the eventual integration with the syntactic analyzer and the proceeding modules in a Natural Language chain. Separation in phrases is preformed according to a list of separator words. The possibility of adding initial (and final) junks to use in the rules is possible.

Features

Team / Authors

• Joana L. Paulo
• David Martins de Matos

Platforms

• Windows / Linux / Mac OS X

Developing status

• Stable
• Last stable version: 3.5

Integration in other systems

As PAsMo writes the tags and regroups the words it is mainly used to ease the integration of a morphological analyzer with the following module in the NLP chain.

Implementation issues

PAsMo is written in C++ and the algorithm was enhanced, reducing processing time. Also XML input and output is available in order to ease communication with other modules and allow data verification.

Client/Server platforms

A distributed version of PAsMo is available allowing the use of the system in a client/server platform through Sun Remote Procedure Call (RPC).