Cross-lingual Visual Verb Sense Disambiguation
This work addresses the problem of verb sense ambiguity in machine translation for multilingual applications, representing an incremental extension from noun to verb disambiguation.
The paper tackled cross-lingual verb sense disambiguation by introducing the MultiSense dataset with 9,504 images annotated in English, German, and Spanish, and showed that visual context improves disambiguation models and enhances text-only machine translation results.
Recent work has shown that visual context improves cross-lingual sense disambiguation for nouns. We extend this line of work to the more challenging task of cross-lingual verb sense disambiguation, introducing the MultiSense dataset of 9,504 images annotated with English, German, and Spanish verbs. Each image in MultiSense is annotated with an English verb and its translation in German or Spanish. We show that cross-lingual verb sense disambiguation models benefit from visual context, compared to unimodal baselines. We also show that the verb sense predicted by our best disambiguation model can improve the results of a text-only machine translation system when used for a multimodal translation task.