Creating sentiment dictionaries via triangulation

Results are presented that verify the triangulation hypothesis, by evaluating triangulated lists and comparing them to non-triangulated machine-translated word lists.

Abstract

The paper presents a semi-automatic approach to creating sentiment dictionaries in\nmany languages. We first produced high-level gold-standard sentiment dictionaries\nfor two languages and then translated them automatically into third languages.\nThose words that can be found in both target language word lists are likely to be\nuseful because their word senses are likely to be similar to that of the two source\nlanguages. These dictionaries can be further corrected, extended and improved. In\nthis paper, we present results that verify our triangulation hypothesis, by evaluating\ntriangulated lists and comparing them to non-triangulated machine-translated word\nlists.