Advertising slogan generation system reflecting user preference on the web

Abstract

Increased demand for Web advertising has resulted in a corresponding increase in the need to develop online personalized advertisements. This paper proposes an advertising-slogan generation system reflecting Web-user preferences. Using a social networking service (SNS) site as the knowledge base for word preferences, and by employing an advertising slogan corpus, the proposed system aims to generate slogans that reflect advertising posts on an SNS. Using model slogans selected from a corpus containing 24,472 slogans, the proposed system generates slogan candidates using the knowledge obtained from a post on an SNS. These slogan candidates are selected based on the following three indexes: the natural level given by a large-scale balanced corpus, a semantic-relations score using advertising slogans, and the preference level obtained from SNS sites. In particular, the proposed system extracts preference data from these SNS fan pages and estimates the preference level on each word based on a bag-of-words model. This enables the proposed system to select slogans in a timely fashion. The authors conducted a subjective experiment to examine the quality of the generated slogans. The results show that (1) the natural and semantic-relation levels are effective for selecting slogans that reflect a post, and (2) the preference-level index contributes to the selection of preferred slogans that are interesting to users.