Sensitivity-Proof Content Advertising Based on Two-Stage Text Classification

Sensitive information, e.g. sex-related, violate news, hinders a lot to content advertising, especially for mobile environment. This paper presents a sensitivity-proof content advertising method by utilizing two-stage text classification. The two stages are advertisement (Ad) classification and sensitivity detection in sequence. A policy is established to combine the results of the two stages for advertisement selection and recommendation. The paper also presents some creative feature extraction and selection methods to promote classification performance. From the experiment results, we believe that these methods will be of great value to the Ad application in Samsung Electronics´ next generation mobile products.