Advertising Keyword Recommendation based on Supervised Link Prediction in Multi-Relational Network
Xiao Yang, Tao Deng, Zhi Miao Guo, Zongyao Ding · 2017
In the sponsored search system, advertisers bid on keywords that are related to their products or services to participate in repeated auctions and display their ads on the search result pages. Since there is a huge inventory of possible terms, instructive keyword recommendation is an important component to help advertisers optimize their campaigns and improve ad monetization. In this paper, by constructing a heterogeneous network which contains four types of links between advertisers and keywords based on different data resources and mining complex representations of network structure and task-guided attributes of nodes, we propose an approach to keyword recommendation based on supervised link prediction in multi-relational network. This method can retrieve ample candidates and provide informative ranking scores for recommended list of keywords, experimental results with real sponsored search data validate the effectiveness of the proposed algorithm in producing valuable keywords for advertisers.