WCSE

Ranking via Hypergraph Learning Integration of Textual Content and Visual Content

Abstract— Ranking has been widely researched in information retrieval and machine learning. Yet it is still a challenging problem, especially in visual product search. In this paper, we propose a novel hypergraph learning based ranking model by mining the correlations among products’ textual and visual features. We formally define a unified hypergraph based ranking framework for product search. Each product image is regarded as a vertex in a hypergraph. The hypergraph captures various high-order relations among different products’ information, including visual content, product categorization labels, and product descriptions. We conducted experiments on the proposed ranking algorithm on a data set collected from various e-commerce websites. The results of our comparison demonstrate the effectiveness of our proposed algorithm.

Index Terms— Ranking; Hypergraph Learning; Product Search; Visual Search.

Kaiman Zeng, Nansong Wu, Arman Sargolzaei, Kang K. Yen
Department of Electrical and Computer Engineering Florida International University, US

[Download]

Cite: Kaiman Zeng, Nansong Wu, Arman Sargolzaei, Kang K. Yen, "Ranking via Hypergraph Learning Integration of Textual Content and Visual Content," 2015 The 5th International Workshop on Computer Science and Engineering-Information Processing and Control Engineering (WCSE 2015-IPCE), pp. 367-372, Moscow, Russia, April 15-17, 2015.