Generating Ideas for Pictorial Advertisements: Starting from Pictorial Metaphors

Title of talk: Generating Ideas for Pictorial Advertisements: Starting from Pictorial Metaphors

Abstract:

Behind every good advertisement there is a creative concept, a Big Idea. In contrast to the countless number of ads, a small number of patterns of effective communication (idea templates) were uncovered, which are invariant across content and context. Moreover, researchers have been working on systematizing idea templates, that is, providing a coherent framework of differentiating templates.

The taxonomies of idea templates are the foundation of our computational approach of generating ideas for pictorial ads. Our computing endeavor started from the conceptual dimension of a single idea template called pictorial metaphor. Metaphor involves two concepts, tenor and vehicle. The best acknowledged effect of metaphors is highlighting certain aspect of the tenor or introducing some new information about it, resulted from the interaction between the conceptual domains of the tenor and vehicle. Not every metaphor is comprehensible or aesthetically pleasing. Only apt ones are.

We propose a two-stage approach of generating apt metaphor ideas for pictorial ads: Stage 1 searches for concepts that are salient in a given property to be highlighted; Stage 2 evaluates the aptness of the found concepts as metaphor vehicles. For Stage 1, we developed knowledge extraction methods of harvesting a large quantity of concepts and their stereotypical properties, especially connotations. At Stage 2, concepts found in Stage 1 are filtered by five metaphor aptness criteria, including concreteness and imageability, affect polarity, salience imbalance, secondary properties and similarity with tenor. In my talk, I will give details of the implementation of the above approach, as well as the results and analysis of experiments conducted to evaluate the approach and implementation.

Bio:

Ping Xiao had been conducting her PhD research in the Interactive Technologies Group of the Department of Information and Communication Technologies of Universitat Pompeu Fabra, Spain. She is at the final stage of her PhD and just joined the Discovery Group at the Department of Computer Science and HIIT. Here, she will work on computational creativity research in the Concept Creation Technology (ConCreTe) EU project. Mrs. Xiao is interested in all aspects of persuasive communication and particularly working on building creative systems that generate it.