The first workshop on Multi-Language Processing in a Globalising World was held in Dublin City University on September 4th and 5th, 2017.
Globalisation has, on the one hand, brought significant growth in free international trade and cross-cultural communication, as well as access to newly-developed technology, media, education, healthcare, consumer goods, and more. On the other hand, it may have negative impacts on local societies, such as cultural homogenisation. The goal of this workshop was to create a forum in which natural language processing researchers and linguists could come together and discuss these issues.
Supported by the ADAPT Centre and DCU, the workshop’s overall aim was to provide a research forum dedicated to state-of-the-art methods and techniques on multi-language and cross-language processing and exploring the use of such technologies in specific tasks.
ADAPT researchers Professor Qun Liu and Dr. Jinhua Du were the main organisers of the event, in which shared tasks on Cross-lingual Word Segmentation and Morpheme Segmentation were major components coordinated by Dr. Chao-Hong Liu and Dr. Teresa Lynn. The corpora which were prepared for the shared tasks included ten languages, all of them curated voluntarily by researchers from all over the world. These included Amharic, Basque, Mandarin Chinese, Farsi, Filipino, Finnish, Japanese, Kazakh, Marathi, Uyghur and Vietnamese.
Three invited speakers gave talks in the workshop. Prof Željko Agić from the IT University of Copenhagen introduced recent developments in cross-lingual NLP, Prof Trond Trosterud of the Arctic University of Norway gave a great review on NLP for morphologically rich languages, and Dr Teresa Lynn of DCU talked about Irish language technologies. Prof Tim Baldwin also gave a guest speech on language identification, which also summarised recent developments in multi-language NLP.
Attended researchers came from Ireland, China, the U.S., Australia, Spain, Finland and Norway. NLP technologies for the Mongolian and Tibetan languages, among others, were also presented in the workshop.
The workshop was highly successful, with proceedings archiving eight workshop papers and four shared task papers.
Share this article: