TransPic:Using Images as Interlingua for Machine Translations Systems

JIITA, vol. 2, no. 3, pp.100-109, Sep, 2018

DOI: DOI: 10.22664/ISITA.2018.2.3.100

TransPic:Using Images as Interlingua for Machine Translations Systems

Mark Vincent B. Melgo 1), Robert R. Roxas 2,*)
1,2) Dept. of Computer Science, College of Science, University of the Philippines Cebu, Philippines

Abstract: This paper proposes a new approach to machine translation systems that use images as interlingua. The system is just like the usual Text-to-Text systems but with a pictorial output that conveys the image representation of the translated text. The system differs from traditional approaches of machine translation because it generates images together with the translated texts, which would allow the users to better understand the translation, thus avoiding misunderstandings and miscommunication. The system initially contains 945 words together with their corresponding images and are categorized into 29 different categories. Two sets of survey were done on the understandability of images that have been generated, particularly the images for pronouns. The results showed that such images are highly understandable.

Keyword: machine translation; interlingua; TransPic system

Full Paper: