Interpretable Identification of Cybersecurity Vulnerabilities from News Articles

Proceedings of Recent Advances in Natural Language Processing, pages 26–34 Sep 1–3, 2021. https://doi.org/10.26615/978-954-452-072-4_004 26 Domain Adaptation for Hindi-Telugu Machine Translation using Domain Specific Back Translation Hema Ala Vandan Mujadia Dipti Misra Sharma LTRC LTRC LTRC IIIT Hyderabad IIIT Hyderabad IIIT Hyderabad hema.ala@research.iiit.ac.in vandan.mu@research.iiit.ac.in dipti@iiit.ac.in Abstract In this paper, we present a novel approach for domain adaptation in Neural Machine Translation which aims to improve the translation quality over a new domain. Adapting new domains is a highly challeng- ing task for Neural Machine Translation on limited data, it becomes even more diffi- cult for technical domains such as Chem- istry and Artificial Intelligence due to spe- cific terminology, etc. We propose Domain Specific Back Translation method which uses available monolingual data and gen- erates synthetic data in a different way. This approach uses Out Of Domain words. The approach is very generic and can be applied to any language pair for any do- main. We conduct our experiments on Chemistry and Artificial Intelligence do- mains for Hindi and Telugu in both direc- tions. It has been observed that the usage of synthetic data created by the proposed algorithm improves the BLEU scores sig- nificantly. 1 Introduction Neural Machine Translation (NMT) systems achieved a breakthrough in translation qual- ity recently, by learning an end-to-end system (Bahdanau et al., 2014)(Sutskever et al., 2014). These systems perform well on the general do- main on which they trained, but they fails to produce good translations for a new domain the model is unaware of. Adapting to a new domain is highly chal- lenging task for NMT systems, it becomes even more challenging when it comes to technical domains like Chemistry, Artificial Intelligence etc, as they contain many domain specific words. In a typical domain adaptation sce- nario like ours, we have a tremendous amount of general data on which we train an NMT model, we can assume this as a baseline model, now provided a new domain data, the chal- lenge is to improve the translation quality of that domain using available little amount of parallel domain data. We adopted two techni- cal domains namely, Chemistry and Artificial Intelligence for Hindi -> Telugu and Telugu -> Hindi experiments. The parallel data for the mentioned techni- cal domains is very less, hence we used back translation to create synthetic data. Instead of using synthetic data directly which may con- tain lots of noise we used domain monolingual data to create synthetic data in a different way (see section 3.4) and used such that translation of domain terms and context around them is accurate. 2 Background & Motivation As noted by Chu and Wang (2018) there are two important distinctions to make in do- main adaptation methods for Machine Trans- lation(MT). The first is based on data re- quirements, supervised adaptation relies on in- domain parallel data, and unsupervised adap- tation has no such requirement. There is also a difference between model-based and data-based methods. Model-based methods make explicit modifications to the model ar- chitecture …

Content truncated to 500 words