AMIRA 2.0 is a suite of tools for the processing of Modern Standard Arabic text. It takes in raw text and produces segmented output labeled with part of speech tag information as well as shallow syntactic structures (chunk level information). AMIRA is designed to be a very flexible platform. The user can choose different tokenization schemes. The output of tokenization handles morphotactics, however the current version of AMIRA does not separate out inflectional affixes even if they are reflected in the POS tag set. For part of speech tagging, the user can specify different levels of granularity for the POS tag set. The richest of which explicitly marks inflectional variants such as number, gender and person. The current version of AMIRA does not produce CASE and MOOD information. For shallow syntactic structure, the current version does reflect some level of embedded structure. It accepts Arabic script input as well as the Buckwalter transliteration input encoding formats. The output can be produced in the user's choice of encoding, by default, it will produce the output in the same encoding as the input data.
|
Software 