Discovering implicit concepts in popular TV programs from social media

With large audience base for popular TV programs, vendors usually place program-related product information implicitly by product placement, or embedded marketing schemes. They target at specific groups of audience who might be interested in the products for promotion purpose. Thus, contextual advertising technique is usually applied in matching product information with relevant Web pages. However, it cannot be directly applied to TV programs without suitable metadata. In this paper, we propose a Web mining approach to implicit concept discovery for popular TV programs. First, major topics in programs are extracted by keyphrase extraction and topic filtering methods. Then, implicit relevance analysis is performed between topics and related categories in various social Web forums to identify implicit concepts. In the experiments, we evaluated the performance of our proposed approach for popular Korean, Japanese, Chinese, and Taiwanese dramas. As the experimental results show, an average precision of 65.11% can be achieved for popular dramas. This validates the effectiveness of our proposed approach in discovering implicitly relevant concepts from social media.