抄録
This paper proposes a kansei-aware landscape picture creating system from natural language sentence. The proposed system creates a landscape picture by focusing on nouns. In the system, several concepts such as “sea” and “sky” and their many graphical parts are stored. Semantic estimation among input words plays an important role in the creating of a meaningful image representation of the input sentence. We use Google N-gram to cope with this issue, a lexical resource. As a result, restrictions on the input sentences can be greatly weakened and relevant constituents can be selected. In addition, the proposed system can express sense of season by selecting appropriate graphic parts with suitable color to the input sentence. Furthermore, the relation between colors and adjectives is also considered in the proposed system to reflect kansei aspects of the input sentence. Evaluation experiments have been carried out to show the effectiveness of the proposed system.
