In-Image Advertising CONCLUSION We have presented an innovative platform for in-image advertising. The platform is able to au- tomatically decompose the Web page into several coherent blocks, select the suitable images from these blocks for advertising, detect the nonintru- sive advertisement insertion positions within the images, and associate the relevant advertisements with these positions. In contrast to conventional ad-networks, in-image advertising is fully dedi- cated to image and aims to monetize the visual content rather than surrounding text. We formalize the association of advertisements with images as an optimization problem and show a practical solution over a large-scale of real-world data. Chatterjee, P., Hoffman, D. L., & Novak, T. P. (2003). Modeling the clickstream: Implica- tions for web-based advertising efforts. Mar- keting Science, 22(4), 520­541. doi:10.1287/ mksc.22.4.520.24906 Chen, W.-Y., Lee, B. N., & Chang, E. Y. (2006). Fotowiki: distributed map enhancement service. Proceedings of the 14th annual ACM international conference on Multimedia (pp. 803-804). New York: ACM. Dai, H., Zhao, L., Nie, Z., Wen, J.-R., Wang, L., & Li, Y. (2006). Detecting online commercial intention. In Proceedings of International World Wide Web Conference. ACM. He, X., Cai, D., Wen, J.-R., Ma, W.-Y., & Zhang, H.-J. (2007). Clustering and searching WWW images using link and page layout. ACM Transac- tions on Multimedia Computing, Communications, and, 3(2). REFERENCES