DebajyotyBanik/Framework-Faster-SMT-Decoder

Our proposed framework can be applied to utilize the beauty of SMT system for building end-to-end translation architecture. The widely used statistical machine translation toolkit (moses) can also be improved with the proposed framework. Statistical and neural based classifiers are introduced as a bridge between the SMT decoder and random input text. Here, the required machine translation parameters for decoding are dynamically modified as per requirements for better resource utilization.

Code: The source code for the proposed work Dataset and text_data: Prepared datasets

The detailed readme is available inside the respective folders.

Cite:

The Decoder Framework for Optimized Pruning Strategy in Faster Statistical Machine Translation Debajyoty Banik