Personalized Ads Based on Grid Modeling

In addition to provide major funding for many Internet companies, online advertising creates a disutility to consumers, subsequently reducing market share. However, previous works focus only on the topical relevance of ads and, in doing so, neglect consumer attitudes. From the view of text processing, they focus only on the topic dimension of texts, while paying no attention to the sentiment dimension. This work proposes a feature extraction process to match advertisement and targeted users by extracting features from the user’s profile and advertisement specification. First, the proposed platform relies on mine characteristics supplied by a user to his avatar including preferred color, style and feeling. Second, the system selects the best matching advertisement based on the user’s variable interests (as expressed on his blog). These features are scored and finally these advertisements are conveyed to the target users by product. Experimental results in several topics demonstrate that the proposed framework works well in detecting a user’s potential preferences, and in recommending suitable advertisements