Technology
Technology
Our technology is based on cutting edge machine learning algorithms for content categorization, semantic keyword and entity analysis, and sentiment analysis, which are all essential for ensuring comprehensive brand safety. Categorization helps identify topics and broad categories such as baseball or hate speech but for brand suitability, these are too broad. An article about baseball injuries may not be appropriate for a sports equipment manufacturer. Keyword and entity analysis helps in this situation. On the other hand, keyword analysis by itself can lead to inaccuracies. The keyword 'home run' in an article talking about how the election debate was a home run for the presidential candidate may not be suitable for a sports ad. Content categorization helps resolve the ambiguity. Finally, categorization and keyword analysis may help us identify that a webpage is talking about car tires but if it is discussing the negative environmental impact of used tires, it may not be appropriate for car tire manufacturers. Sentiment analysis helps us make the right choice in this case. With a combination of these technologies, our platform provides robust and unparalleled brand safety for brands and content optimization for publishers.
Content categorization
Our categorization engine categorizes unstructured content into meaningful categories and identifies topics. As an advertiser you want to reach relevant audiences. A chicken burger ad is not appropriate for the audience consuming articles about vegan food. As a publisher you want to understand how different content categories are performing and which type of content you should produce more of. Our categorization solution uses the IAB taxonomy but can also be customized for other taxonomies.
Categorization
Semantic keyword and entity analysis
We can identify keywords as well as entities such as people, places, times, and events. This is much more powerful compared to simple keyword matching. For example, if we want to target all entertainment content but avoid a musical band 'Kings', simple keyword filtering will also discard any content about the Kings county in New York as well as content about royal families around the world. Our technology can identify the semantics associated with keywords and hence can avoid these situations.
Keywords
Sentiment analysis
With the sentiment analysis technology, our platform can identify whether the content has a positive, negative, or neutral sentiment. We extract this information at the whole document level as well as with respect to different entities in the document. For example, a tourist's blog about a trip may have an overall positive sentiment but may have a negative review for a hotel, and so may not be appropriate for the hotel company's ad. This fine-grained information helps make better-informed decisions.
Sentiment
Offensive image detection
Our technology uses the latest advances in deep learning to identify offensive images such as pornography, hate content, accidents, emergencies, crime and violence, and disasters. A combination of the most sophisticated AI approaches in text as well as image analysis enables our platform to offer a comprehensive solution that is hard to find in the market.
Technology
API
Our solutions are available through our API. With a simple HTTP call, our cloud-based platform processes the content you provide and returns the information as a JSON object. Please contact us for a free trial.
Technology