How Page Analysis Works
Verity page analysis works by extracting metadata and significant content from a page, then applying computer vision (CV) and natural language processing (NLP) to the extracted data. For detailed information about Verity technology and processing, refer to the Verity Description of Methodology.
Note: Verity does not download or analyze the CSS, JavaScript, navigation, footer, sidebars, third-party ads, user-generated content, or other areas extraneous to the main textual content of the page.
Page Analysis Response
The Verity page analysis response is aggregated into the following sections:
IAB Categories
Identifies and scores contextual targeting categories. Includes standard IAB v1.0 and IAB v2.0 categories.Keywords
Extracts and scores page keywords according to prominence.Brand Safety
Identifies and scores brand safety threats according to GumGum's Threat classification taxonomy, in compliance with The 4A’s Advertising Assurance Brand Safety Floor.Sentiments
Analyzes the overall opinions expressed within the content of a page, and classifies the percentage of positive, negative, and neutral sentiments.
The format is JavaScript Object Notation (JSON) Content-Type: application/json; charset=UTF-8.
Check out an example page analysis response here.
Page Analysis Categories and Keywords
The following table summarizes the classifications and keywords that are detected during page analysis.
Brand Safety (Threat) Categories | Contextual | Keywords |
---|---|---|
Violence | IAB v1.0 IAB v2.0 | Contextual targeting keywords |
Page Analysis Language Support
Verity classification categories and keywords are supported in the languages specified in the following table:
Language | Keywords | Categories | |||
---|---|---|---|---|---|
Threat | IAB V2 | IAB V1 | Sentiment | ||
English | ✔ | ✔ | ✔ | ✔ | ✔ |
Japanese | ✔ | ✔ | ✔ |