Natural Language Processing researcher and Perl programmer
I am a Natural Language Processing (NLP) specialist. I hold a PhD degree in this field of Artificial Intelligence (thesis subject: "Word Sense Disambiguation").
I am a good Perl programmer and I also can write C/C++/C# or Java code. I put a lot of emphasis on algorithms and their space and time requirements and I prefer to code stand-alone applications. I am a theory oriented kind of guy and not a specific technology adept (in principle I can write code in any language as long as the documentation is available).
Radu is an excellent programmer. His coding is lean and clean. During the project he came up with his own ideas to make the tool even better than I imagined in the first place. Excellent communication, reliable and straightforward.
Research Institute for Artificial Intelligence, Romanian Academy
thg 8 2001 - Hiện tại
I am a Natural Language Processing researcher. I currently work on/with:
- machine translation (e.g. Moses)
- part-of-speech tagging (e.g. HMMs, neural nets)
- word sense disambiguation
- parsing (syntactic analysis, dependency parsing)
- question answering
- word alignment
- automatic bilingual dictionary extraction from parallel or comparable corpora
Full resume can be found here:
Junior C++ Programmer
thg 1 2001 - thg 3 2001 (2 tháng, 1 ngày)
C++ programming in the company's COSWIN application.
PhD in Computer Science/Natural Language Processing
Romania 2003 - 2007
Master in Theoretical Lingustics
Universitatea din Bucuresti, Romania 2002 - 2003
Graduate Studies in Computer Science
Universitatea 'Politehnica' din Bucuresti, Romania 2001 - 2002
Nội dung đã đăng
Large tagset labeling using Feed Forward Neural Networks. Case study on Romanian Language
Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (ACL 2013)
High accuracy POS tagging with neural nets, using a large tagset named Morpho-Syntactic Descriptors.
The Romanian wordnet in a nutshell
Language Resources and Evaluation Journal
This is an account of the current development activities of the Romanian semantic network, aligned to Princeton WordNet.
Wikipedia as an SMT Training Corpus
Proceedings of Recent Advances in Natural Language Processing (RANLP 2013)
We have mined for parallel sentences in the Wikipedia seen as a comparable corpora. We have tested the quality of the corpus by doing SMT with Moses trained on the extracted data.