Extracting Advertisements from Content Marketing Articles based on TopicCNN

With the rapid development of Internet technology, online social networks have been an essential part of people's life. With the increase in the number of users on various social platforms, more and more companies begin to value the importance of social media marketing. One of the most common ways is Content marketing (CM), which inserts advertisements into regular articles in a roundabout and covert way. However, CM articles often contain false and exaggerated information, misleading consumers, and harming their economic interests. The advertisement also affects the reading experience of users and affects the accuracy of data analysis. Extracting advertisements from CM articles is beneficial to both users and social platforms. Generally speaking, the topics of the inserted advertisement are very different from that of the regular content. Moreover, advertisement usually contains more positive words and phrases to attract more attention. According to these characteristics, we propose a method for identifying the content of advertisements based on the fusion of topic and semantic features, which are used to train a supervised classifier based on a manually labeled dataset. We also conduct extensive experiments to validate the effectiveness of the proposed method.