|English | Japanese|
People from different cultures and language collaborate to meet the needs of a more and more globalized society. The Intercultural Collaboration group of Ishida & Mastubara Laboratory aims to provide tools, mainly based on Services Computing, to transcend cultural barriers over distributed environments.
We have been identifying issues in intercultural collaboration through experiments and analyses, and have proposed a series of technologies and methodologies to support intercultural collaboration. For instance, from 2002 to 2005, we jointly conducted Intercultural Collaboration Experiment (ICE) with research institutes in Asia to analyze issues on using machine translation in intercultural, multilingual collaboration setting.
From 2006, we have been investigating technologies and methodologies to support Intercultural Collaboration through service-based infrastructures. Hence, we are working with the National Institute of Information and Communication Technologies (NICT), a Japanese governmental research institute, to develop the Language Grid, an infrastructure allowing its users to easily deploy and combine various language resources.
Our group is thus working on the use of language resources for intercultural collaboration and on how to improve service infrastructures to let people develop their own tools to break culture barriers. Please refer to the Research section below for summary of each research. For a list of publications by the Intercultural Collaboration Group, please refer to the Selected Publications section.
Intercultural Collaboration Using Machine Translation
The online language population is to said to comprise of 1/3 English, 1/3 European languages other than English, and 1/3 Asian languages. The language barrier can be viewed to exist in the online world same as the real world. One technology that can overcome the language barrier to support multilingual communication is machine translation. However, the quality of the machine translation currently has much room for improvement. Although the accuracy of machine translation is far from perfect, machine translation is helpful in such intercultural collaboration setting where no bilingual person is present. We have been conducting experiments to identify problems that arise during multilingual communication using machine translation, and to propose ways to maximize the effective usage of machine translation. Moreover, we are extending our research scope from machine translation to other language resources such as dictionaries and multilingual corpus, to investigate effective usage of these language resources in intercultural collaboration.
Figure: Japanese and Chinese subjects using machine translation embedded chat system to discuss tangram figure arrangement
Language Grid Playground Operation & Promotion
Language Grid Playground can be accessed at http://langrid.org/playground/ using any commercial Internet browser (Internet Explorer, Firefox, etc.). The Language Grid Playground showcases various language services provided by the Language Grid as well as the latest research and development efforts of this group. The source codes of the Language Grid Playground will be improved and published in the near future.
Language Grid is a multilingual service platform which enables easy registration and sharing of language services such as online dictionaries, bilingual corpora, and machine translations. Using the Language Grid, one can easily combine language resources (bilingual dictionary etc.) with language processing functions (machine translation etc.), or create community-specific language services by adding community-created language resources. The resources on the Language Grid can be used for nonprofit use, including the use by public agencies or nonprofit organizations for their main businesses or for research, and the use by profit organizations for social contributions.
Collaborative Translation by Monolinguals with Machine Translators
Machine translation has proven to be a very useful tool improving the communication between people of different language and cultural backgrounds. However, there are still cases where machine translation errors cause misconceptions in multilingual collaboration.
Collaborative translation is a concept where two non-bilingual people perform a translation task via machine translation using their native languages. The goal is to improve the machine translation quality by developing a collaborative translation protocol and improving the fluency of the target language translation by using the source language side to check the adequacy of the back translation
Social Capital in Multilingual Communication
The growth of the Internet along with increasing globalization has brought people of different cultural and language backgrounds closer together. The amount of people working in a computer-mediated environment with no common language or culture is increasing rapidly. With the increase of information available on the Internet, a problem with understanding and processing of information in multiple languages becomes apparent. Development of supporting systems for international collaboration is increasingly important.
The biggest obstacle for people working in different cultures and countries has always been the language barrier. Language ability greatly affects the formation social capital and breaking down cultural barriers. Social networks are essential in establishing close business and personal relationships. Machine translation has emerged as one solution to overcome the language barrier both in online and real world multicultural collaboration.
Context-Aware Coordination of Cascaded Machine Translations
Cascading multiple machine translators realizes translation for language pairs which single machine translators does not support. However, as the result of translation by cascaded translators, the meaning of the translated sentence can be different from the given one due to ambiguity of words. To solve this problem, we coordinate machine translators by propagating context using tuples of multilingual equivalent terms.
User Involvement in Quality of Service (QoS) Control for Language Service Composition
The success of the emerging services computing will rely on the Quality of Service (QoS). The current research and industry on QoS are not enough to catch up the speed of services computing. A new approach is inevitable to enhance QoS technology by increasing the involvement from user in QoS control. User involvement approach is very important to increase the participation from user as well as service quality improvement.
In this approach, an improved technique of constraint satisfaction problem will be used to optimize QoS constraints. Two categories of classes of service are also proposed to accomodate constraint satisfaction, i.e.: user defined class of service and provider defined class of service.
Provider-centered Trust Distribution for Composite Services
In current service composition mechanisms, the sole client decides of the participant partakers from a range of providers estimated suitable for the workflow. However, since providers cannot choose their partners, several problems such as commercially concurrent providers, disparity of quality for the composite service and data privacy can arise.
In this research, we propose to tackle this problem by building on Multi-Agent research on coalition formation to let the providers choose trustworthy partners, thus letting them decide on the team formation and lead to a smoother and more efficient composition. Challenges in this research include coalition formation negotiation protocol based on trust, trust metrics for service providers and adaptable workflows.
Please add the abbreviated domain to each account name.
Toru Ishida (Professor)
Ari Hautasaari (D3) arihau
Shi Chunqi (D3) shi
Amit Pariyar (D1) amit
Kemas Muslim Lhaksmana (D1) kemas.muslim
Trang Mai Xuan (D1) trangmx
Nan Jin (M2) jin
Shinsuke Goto (M2) s-goto
Hiromichi Cho (M2) h-cho
Tatsuya Nozoe (M1) nozoe
Kaori Kita (M1) kita
Daisuke Kitagawa (M1) kitagawa
Takuya Nishimura (M1) nishimura
Ann Lee (M1) lee
Yuri Horita (B4) horita
Yuka Fujiwara (B4) fujiwara