Image-Based Detection Criteria for Cultural Differences in Translation
- 202 Downloads
These days, the improved accuracy of machine translation system enable us conduct intercultural collaboration. Even though machine translation could translate words correctly, we sometimes face trouble in our communications because we think about different images for the same word due to our different backgrounds and culture. To make the machine translation users notice the difference, a previous study proposed a cultural difference detection method based on image feature similarity. However, the proposal have a deficiency about any similarity criteria for judging cultural differences. This paper proposes a method for calculating the criteria for detecting cultural differences. Specifically, a threshold value is used to determine the presence or absence of a cultural difference from the image similarity. An experiment compares the detection results with the results of human determination as to the cultural difference contained in the pair. The experiment changes the threshold value to determine the optimum threshold value. We prepared 1000 concepts and judged the cultural difference by the proposed method. We divided the concepts into 200 and verified them by 5-fold cross validation. As a result, the average threshold value closest to human judgement calculated from validation data was 0.4 and the accuracy for test data was 80.4%.
KeywordsIntercultural collaboration Multilingual communication Machine translation Image feature
This research was partially supported by a Grant-in-Aid for Scientific Research (B) (18H03341, 2018–2020) and a Grant-in-Aid Young Scientists (A) (17H04706, 2017–2020) from Japan Society for the Promotion of Sciences (JSPS).
- 1.Deutscher, G.: Through the Language Glass: Why the World Looks Different in Other Languages. Metropolitan Books, New York (2010)Google Scholar
- 2.Yoshino, T., Miyabe, M., Suwa, T.: A proposed cultural difference detection method using data from Japanese and Chinese Wikipedia. In: 2015 International Conference on Culture and Computing (Culture Computing), pp. 159–166. IEEE (2015)Google Scholar
- 4.Hofstede, G.H., Hofstede, G.J., Minkov, M.: Cultures and Organizations: Software of the Mind, vol. 2. Mcgraw-Hill, New York (2005)Google Scholar
- 6.Koda, T.: Cross-cultural comparison of interpretation of avatars’ facial expressions. In: IEEE/IPSJ Symposium on Applications and the Internet (SAINT-06) (2006)Google Scholar
- 7.Pituxcoosuvarn, M., Lin, D., Ishida, T.: A method for automated detection of cultural difference based on image similarity. In: Nakanishi, H., Egi, H., Chounta, I.-A., Takada, H., Ichimura, S., Hoppe, U. (eds.) CRIWG+CollabTech 2019. LNCS, vol. 11677, pp. 129–143. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-28011-6_9CrossRefGoogle Scholar
- 8.Fellbaum, C.: WordNet. In: The Encyclopedia of Applied Linguistics (2012)Google Scholar
- 9.Bond, F., Isahara, H., Fujita, S., Uchimoto, K., Kuribayashi, T., Kanzaki, K.: Enhancing the Japanese WordNet in the 7th Workshop on Asian Language Resources, in conjunction with ACL-IJCNLP 2009 (2009)Google Scholar