This page introduces and describes the capabilities of Vertex AI Search for media. The page also provides links to more information, tutorials and checklists, to get you started with Vertex AI Search for media.
Vertex AI Search includes two capabilities specific for the media industry:
-
Media recommendations. Get recommendations for media content such as videos, news, and music. With media recommendations, audiences can discover more personalized content, like what to watch or read next, with Google-quality results customized using optimization objectives.
Media search. Get Google-quality search results with advanced query and document understanding designed for media content.
How media apps and generic apps differ
There are many similarities between media apps (and data stores) and generic apps (and data stores) in Vertex AI Search. However, there are some key feature differences:
Media apps use user events. You upload user events to personalize recommendations and rank search results for your audience.
Media apps require media metadata to conform to a predefined schema or to use a custom schema that contains a minimum set of key properties.
Predefined schema (GA). This lets recommendations and search ranking use Google-defined, media-specific fields such as content ratings, aggregated ratings, persons, and production year to help generate results based on media engagement.
Custom schema (Public preview). The custom schema gives you more flexibility than the predefined schema. However, your schema fields must map to the following required key properties:
title,category,uri,media_available_time, andmedia_duration. Thecategoryproperty must be an array of strings, and the other four properties are strings.In addition to the required key properties, Google recommends that you map as many other schema fields as possible to the suggested key properties. The suggested key properties represent similar media metadata to that in the predefined schema—for example, content ratings, aggregated ratings, persons, and production year.
Media recommendations apps offer you a choice of models. Media recommendations apps let you choose what kind of recommendation to generate, such as recommending other content that users might like, similar items, or the most popular items.
Media recommendations apps offer you a choice of optimization objectives. For example, you can decide whether to optimize recommendations for click-through-rate to increase the number of interactions with content or conversion rate to increase the consumption of content.
The following table outlines some functional differences between media and generic data stores.
| Media apps and data stores | Generic apps and data stores |
|---|---|
| Data stores are always structured. | Data stores can be of any type (website, unstructured, structured). |
| Require structured data with a predefined schema or a custom schema where you map your data fields to some required key properties. | No key properties are required for structured data. |
| For recommendations, user events are required. | User events are not required. |
| Imported historical user events are joined synchronously. | Imported historical user events are joined asynchronously. |
For more information, see About media data stores and documents and About apps and data stores.
Getting started tutorials
If you are new to Vertex AI Search, try out the getting started tutorials. These tutorials guide you step-by-step through the creation of an app. Data (documents and user events) are provided for the tutorials so all you need is a Google Cloud project and a billing account to create your first app:
Checklists
There is a lot of commonality between working with media apps and working with generic apps, but some features apply only to generic apps and other features only to media apps.
Use the following checklists to guide you through typical workflows specific to media: