Alibaba Cloud Content Moderation is a leading service in the field of content moderation. It is built on Alibaba's extensive security expertise and combines content administration experience from platforms in diverse sectors, such as e-commerce, community, and payment, with advanced AI algorithms and cloud computing infrastructure. The service provides businesses and developers with mature, stable, and out-of-the-box Content Moderation solutions. These solutions help you quickly detect various risks in text, images, videos, voice, and documents in the complex and ever-changing internet environment to help you improve your platform's content quality and user experience.
Product features
The content detection algorithm models have been trained on extensive samples and have undergone diverse evaluations. They are also continuously iterated and upgraded daily. The out-of-the-box model allows enterprises and developers to quickly obtain automated content moderation capabilities. In addition, we provide systematic methods and processes to support custom requirements such as algorithm model adjustments, optimization for specific detection targets, and optimization of detection thresholds.
Currently, the core features of Content Moderation include three components: the Content Moderation API, and the console.
Content Moderation API
The Content Moderation API detects content by calling Alibaba Cloud interfaces. It detects and identifies whether text, images, videos, voices, and documents contain pornography, politically sensitive content, violence, horror, inappropriate advertising, spam, or undesirable content. It is applicable to business scenarios with public-facing content, such as video websites, live streaming platforms, social platforms, media platforms, vertical communities, forums, e-commerce websites, storage services, CDN platforms, and AIGC products.
Console
For the two detection methods mentioned above, Content Moderation provides the Alibaba Cloud Management Console to meet analysis and control needs, such as data statistics, content moderation analysis, personalized control configuration, and risk library management. This feature is suitable for users who need analysis, have different detection requirements for multiple scenarios, or require personalized control and management.
Relationship between Content Moderation 2.0 and 1.0
Content Moderation 2.0 is a new content moderation product. Compared to Content Moderation 1.0, it has preset common business scenarios and is optimized for product effectiveness, label richness, configuration flexibility, and price.
Capability comparison between Content Moderation Enhanced Edition and 1.0
Item | Content Moderation 2.0 | Content Moderation 1.0 |
Billing method and pricing |
Note For more information about the pricing of Content Moderation 2.0, see Content Moderation Pricing. |
Note For more information about the pricing of Content Moderation 1.0, see Content Moderation Pricing. |
Detection configuration | Detection scope (10+ major categories) Custom libraries | Detection scope (5+ major categories) Custom libraries |
Default capacity |
|
|
Scenario modality | Modality: Image, text, voice, video Preset business scenarios: common baseline, social entertainment live streaming, audio and video media, and more | Modality: Image, text, video Preset business scenario: General |
Detection results | Interpretable labels (50+, multiple violation labels can be returned simultaneously) Confidence score | Interpretable labels (40+, only one violation label can be returned) Suggested action |
Activation and billing
For more information about the activation and billing of Content Moderation 2.0, see Activation and billing.
For more information about the activation and billing of Content Moderation 1.0, see Activation and billing.
Item | Content Moderation 2.0 | Content Moderation 1.0 |
Pay-as-you-go | The pay-as-you-go billing method for Content Moderation 2.0 meters and bills based on different content types (such as images, text, and voices) and detection volume. For the same content, when multiple threat scenarios are detected simultaneously, the cost is 50% to 70% lower than that of version 1.0 under the same usage conditions. | The pay-as-you-go billing method for Content Moderation 1.0 meters and bills based on the specific type of detected content (such as images, text, videos, ), detection scenarios (including but not limited to image pornography detection, image QR code detection, and text anti-spam), daily scan volume tiers, the suggested action after algorithm detection (for example, review, block, or pass), and the region used for content detection (for example, China (Shanghai)). |