Image the Imageless: Harvesting Connotation Knowledge for Visual Expression

By Ping Xiao and Josep Blat.

Published by The International Journal of Visual Design

Format Price
Article: Print $US10.00
Published Online: May 5, 2016 $US5.00

To convey abstract meanings with visual elements is a frequent challenge in design. A popular vehicle of abstract meanings is the connotation of physical entities. Making designs demands a good deal of this knowledge, for the sake of novelty and appropriateness. We present two automatic knowledge extraction systems, namely VRAC and CDVS, for collecting the knowledge of physical entities and their abstract connotation. VRAC discovers connotation knowledge in general purpose semantic databases. CDVS searches in a large collection of stock photos and their annotations. In the evaluation against thirty eight distinct visual representations of six abstract concepts used in past successful advertisements, both VRAC and CDVS provide a good coverage of visual representations consisting of single objects. Besides, CDVS has the potential to excel in harvesting the abstract connotation of scenes. VRAC enjoys the agility and flexibility of being entirely textual. It can communicate fluently with other automatic tools manipulating concepts, which are very likely to appear in future creative systems. Finally, we point out that the knowledge of emergent culture symbols is a necessary complement for VRAC and CDVS.

Keywords: Visual Design, Design Knowledge, Design Automation, Connotation, Knowledge Extraction

The International Journal of Visual Design, Volume 10, Issue 2, June, 2016, pp.25-36. Article: Print (Spiral Bound). Published Online: May 5, 2016 (Article: Electronic (PDF File; 637.305KB)).

Ping Xiao

PhD Candidate, Department of Information and Communication Technologies, Universitat Pompeu Fabra, Barcelona, Catalunya, Spain

Josep Blat

Professor of Computer Science, Department of Information and Communication Technologies, Universitat Pompeu Fabra, Barcelona, Catalunya, Spain