Targoman: A robust Statistical Machine Translation framework

Targoman is a robust Statistical Machine Translation framework which can use moses or jane trained models to provide a robust decoding process. Further information can be found at Targoman blog.

See INSTALL for build instructions.

Targoman Project Tree

├── Apps │ ├── E4SMT
│ ├── TargomanLoadBalancer
│ ├── TargomanSMTConsole
│ ├── TargomanSMTServer
│ ├── TargomanWebAPI
│ └── TargomanWebUI
├── ExternalToolsAndLibs
│ └── KenLM
├── NLPLibs │ ├── TargomanLM
│ ├── TargomanStem (Not yet published) │ ├── TargomanTextProcessor │ └── TargomanTransliterate (Not yet published) ├── Scripts ├── Samples ├── TargomanCommon ├── TargomanSMT
├── Templates
└── Tests
#####Sub-directories Specification:

• E4SMT: Application based on TargomanTextProcessor which converts text to IXML and vice versa it also validates XML files and some more stuff.
• TargomanLoadBalancer:Application that monitors translation servers and load-balances requests between them.
• TargomanSMTConsole: Application based on TargomanSMT as a multithreaded console application.
• TargomanSMTServer: Application based on TargomanSMT as a multithreaded server application. This app comunicates with TargomanLoadBalancer and can be set-up on clouds.
• TargomanWebAPI: WebAPI to contact with TargomanLoadBalancer and do other DB based stuff.
• TargomanWebUI: WebInterface of Targoman
• KenLM: KenLM Langugae model library
• TargomanLM: Targoman Language Model Library
• TargomanStem: Targoman Stemmer Library (Not yet published)
• TargomanTextProcessor: Targoman TextProcessor Library. Used to normalize characters, tokenize text, tag special tokens and vice versa
• TargomanTransliterate: Targoman Transliteration Library (Not yet published)
• Scripts: Train, Tune, and other usefull scripts (Not yet published)
• Samples: Sample config file and corpora to test Targoman.
• TargomanCommon: Usefull and common tools as a compact library
• TargomanSMT: Core Targoman SMT Library
• Templates: Templates used to generate subprojects
• Tests: BlackBox Tests