Peer39 uses its semantic technology to provide page-level intelligence for web pages. With language, a single expression or term can encompass different concepts, and page-level semantic classification is the most intelligent way to connect with the intended content. The Peer39 advanced semantic technology knows the difference between shark pajamas, loan sharks, and shark attack news and provides page-level intelligence that gives advertisers more control in defining, sourcing, and targeting their ideal ad environment - all before the ad is served.
Our proprietary classification technology is based on Natural Language Processing and Machine Learning.
For more documentation about Peer39, see the Peer39 Category.
Peer39 provides page-level classification data as attributes, which can be used as standard targeting variables across all web pages from any supply source. Each of the Peer39 attributes is based on a proprietary algorithm that is carefully built by our research team and is tuned to achieve optimal precision and recall.
After Peer39 fully classifies a page, it is common to see anywhere from 20-30 attributes returned for a given page; depending on the level of customization for your taxonomy, Peer39 may return 100 attributes or more for a given page.
For Verification, there are three main classification channels that contain the classification attributes:
Peer39 classifies pages semantically in advertiser-focused categories by detecting the main topics covered by the content of the page. These categories are carefully selected and optimized to be most relevant to the verticals that are commonly targeted. Examples include Automotive, Parenting and Kids, and Sports.
The standard, Peer39 safety attributes identify the negative content that is most often considered objectionable or misaligned with advertising messages, such as Mature, Alcohol, Hate Speech and Profanity, and Torrents.
Through offering a variety of standard negative categories, Peer39 provides advertisers and buyers the flexibility to decide and identify what content is ultimately safe for their brand.
Peer39 evaluates negative categories dynamically based on the specific page-level content and not generalized solely by domain. For example, Peer39 can flag negative content seen within a domain's pages and not necessarily flag the entire domain as negative. While some domains may be negative as whole (for example, pornographic sites or torrents sites), most domains include a mixture of content, which may include some negative content in small portions, and Peer39 can identify only the pages that contain negative content. News sites, for example, may show a percentage of impressions classified as Negative News, but the entire site's impressions are not classified as negative.
Additionally, the negative categories' algorithm is binary - either a page will contain objectionable content or not.
Video Player Attributes
Peer39 offers a channel of classification, which describes the characteristics of a video player on a page. For more information, see the Video Player Metrics section in REFERENCE: Media Relevance Metrics and Categories Grid.
Please note that while Peer39 does not support content classification of the in-player video content itself, we classify the surrounding textual content on the page, which is often aligned closely to the actual video content within the player.