Papavassiliou Vassilis

Areas of interest

Language Resource Acquisition and Processing
Document Image Analysis

Vassilis Papavassiliou received the Diploma degree in the Electrical and Computer Engineering from the National Technical University of Athens (NTUA), Greece, in 1998. The title of his thesis was “Development of a Speech Recognition System for Educational Software”. The results of this thesis were incorporated in the “Filoglossia” educational multimedia product. He also holds a M.Sc. degree in Language Technology graduating from NTUA & National & Kapodistrian University of Athens in 2000. The subject of his M.Sc. thesis focused on a statistical approach of the literary style with applications to author identification. In 2010, he obtained his Ph.D. degree at NTUA on handwritten document image analysis under the supervision of Professor Carayannis. During pursing his Ph.D., he participated in teaching of the “Digital Signal Processing” course at the School of Electrical and Computer Engineering – N.T.U.A.

He has been associated with the Institute for Language and Speech Processing (ILSP / R.C. “Athena”) since 1998. He has worked on several educational multimedia projects as a software engineer (“A knight in the castle of letters“, “Trilingual Term Dictionary” and “Nea Logomatheia”) and has also participated, as a research assistant, in the R&D projects (e.g. “Panoptis”, “SAMS”, “PENED 2003 – Applications of Pattern Recognition on Language Technology”). Since 2011, he focuses on applied research in computational linguistics and has participated in several Research and Development projects in the areas of language resource acquisition and processing, including the European Language Resource Coordination (ELRC), PANACEA, and Abu-MaTran. He has been involved in the development of systems for acquiring monolingual and bilingual data from the web and in building domain-specific monolingual and bilingual resources.

Awards

Winners of the ICDAR2009 Handwriting Segmentation Contest, 26-29 July 2009, Barcelona, Spain. Team members: V. Papavassiliou, T. Stafylakis, V. Katsouros, G. Carayannis.
Winners of the ICDAR2007 Handwriting Segmentation Contest, 23-26 September 2007, Curitiba, Brazil. Team members: V. Papavassiliou, T. Stafylakis, V. Katsouros, G. Carayannis.

@article{Toral:2016,
title = {Crawl and crowd to bring machine translation to under-resourced languages},
author = {Antonio Toral and Miquel Espl á and Filip Klubi vc and Nikola Ljube vs and Vassilis Papavassiliou and Prokopidis Prokopis and Raphael Rubino and Andy Way},
url = {http://link.springer.com/content/pdf/10.1007%2Fs10579-016-9363-6.pdf},
doi = {10.1007/s10579-016-9363-6},
issn = {1574-0218},
year = {2017},
date = {2017-01-01},
journal = {Language Resources and Evaluation},
pages = {1–33},
abstract = {We present a widely applicable methodology to bring machine translation (MT) to under-resourced languages in a cost-effective and rapid manner. Our proposal relies on web crawling to automatically acquire parallel data to train statistical MT systems if any such data can be found for the language pair and domain of interest.
If that is not the case, we resort to (1) crowdsourcing to translate small amounts of text (hundreds of sentences), which are then used to tune statistical MT models, and (2) web crawling of vast amounts of monolingual data (millions of sentences), which are then used to build language models for MT. We apply these to two respective use-cases for Croatian, an under-resourced language that has gained relevance since it recently attained official status in the European Union. The first use-case regards tourism, given the importance of this sector to Croatia's economy, while the second has to do with tweets, due to the growing importance of social media. For tourism, we crawl parallel data from 20 web domains using two state-of-the-art crawlers and explore how to combine the crawled data with bigger amounts of general-domain data.
Our domain-adapted system is evaluated on a set of three additional tourism web domains and it outperforms the baseline in terms of automatic metrics and/or vocabulary coverage. In the social media use-case, we deal with tweets from the 2014 edition of the soccer World Cup. We build domain-adapted systems by (1) translating small amounts of tweets to be used for tuning by means of crowdsourcing and (2) crawling vast amounts of monolingual tweets. These systems outperform the baseline (Microsoft Bing) by 7.94 BLEU points (5.11 TER) for Croatian-to-English and by 2.17 points (1.94 TER) for English-to-Croatian on a test set translated by means of crowdsourcing.
A complementary manual analysis sheds further light on these results.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}