Multi-language services provided by Text Moderation 2.0 - Content Moderation

Text Moderation 2.0 enhances multi-language model capabilities. It supports more languages and can automatically identify languages. In consideration of the characteristics of global business, Text Moderation 2.0 adopts separate moderation policies and a separate internationalization labeling system. This topic describes the details of the multi-language services provided by Text Moderation 2.0 and how to use them.

Feature introduction

Compared with the multi-language services of Text Moderation 1.0, Text Moderation 2.0 utilizes separate moderation policies and a separate internationalization labeling system to meet the requirements of global business. In addition, it provides more features to simplify service usage and assist in manual review.

Comparison item	Text Moderation 2.0	Text Moderation 1.0
Multi-language capability	Supports 38 languages.	Supports 18 languages.
Moderation capability	Utilizes multiple models and incorporates language and regional characteristics. The moderation policies are more accurate.	Utilizes a single model and incorporates language characteristics. The moderation policies take into account accuracy and recall.
Labeling system	Utilizes a separate internationalization labeling system for global business, adds internationalization labels such as profanities and regions, and supports multiple risk labels and subcategory labels.	Inherits the labeling system from earlier versions and supports only one single risk label for a text moderation task.
Operation	You do not need to specify an input language, and Text Moderation 2.0 can automatically identify it. After moderation is complete, the language type and translated English content are returned to assist in manual review.	You need to specify an input language. No translation is returned.

Supported languages

The multi-language services of Text Moderation 2.0 support 38 languages.

Language type	Name	Language code
English	English	en
Simplified Chinese	Chinese	zh
Traditional Chinese	Traditional Chinese	zh-tw
Indonesian	Indonesian	id
Malay	Malay	ms
Thai	Thai	th
Vietnamese	Vietnamese	vi
Tagalog	Tagalog	tl
Hindi	Hindi	hi
Arabic	Arabic	ar
Turkish	Turkish	tr
French	French	fr
German	German	de
Russian	Russian	ru
Portuguese	Portuguese	pt
Spanish	Spanish	es
Italian	Italian	it
Dutch	Dutch	nl
Polish	Polish	pl
Japanese	Japanese	ja
Korean	Korean	ko
Urdu	Urdu	ur
Uighur	Uighur	ug
Bengali	Bengali	bn
Persian	Persian	fa
Swedish	Swedish	sv
Danish	Danish	da
Norwegian	Norwegian	no
Icelandic	Icelandic	is
Finnish	Finnish	fi
Belarusian	Belarusian	be
Lithuanian	Lithuanian	lt
Czech	Czech	cs
Slovak	Slovak	sk
Hungarian	Hungarian	hu
Greek	Modern Greek	el
Romanian	Romanian	ro
Irish	Irish	ga

Internationalization labels

The multi-language services of Text Moderation 2.0 adopt an internationalization labeling system. If the content contains multiple types of risks, multiple labels can be returned at the same time. Label categories include, but are not limited to, those listed in the following table.

Label type	Category
Level-1 labels	violence: violence contraband: contraband sexuality: pornography profanity: profanity and abuse pullinTraffic: advertising content regional: regional hostility C_customized: custom library that is hit
Subcategory labels (riskTips)	Subcategory labels are returned in the `xxx_yyy` format. Example: `contraband_Drugs`.

Billing methods

Text Moderation 2.0 supports the pay-as-you-go billing method.

Pay-as-you-go

After you activate the Text Moderation 2.0 service, the default billing method is pay-as-you-go. Daily fees are calculated based on the actual usage. If the service is not called, no charge is made.

Moderation type	Supported business scenario (service)	Unit price
Advanced text moderation (text_advanced)	Multi-language moderation in international business scenarios: comment_multilingual_global	USD 0.3 per thousand calls

Use the Text Moderation 2.0 service

Step 1: Activate Text Moderation 2.0

Open the service activation page to activate the Text Moderation 2.0 service.

After you activate the Text Moderation 2.0 service, the default billing method is pay-as-you-go. Daily fees are settled based on the actual usage. If the service is not called, no charge is made. After you call API operations, the billing system automatically charges you based on your usage. For more information, see Billing rules.

Step 2: Grant permissions to a RAM user

Before you call API operations or use SDKs as a Resource Access Management (RAM) user, you must grant permissions to the RAM user. You can create an AccessKey pair for your Alibaba Cloud account and the RAM user. When you call API operations, you must use the AccessKey pair to complete identity verification. For information about how to obtain an AccessKey pair, see Obtain an AccessKey pair.

Log on to the RAM console as a RAM administrator.
Create a RAM user.

For more information, see Create a RAM user.
Grant the AliyunYundunGreenWebFullAccess system policy to the RAM user.

For more information, see Grant permissions to a RAM user.

After completing the preceding operations, you can call the Content Moderation API as the RAM user.

Step 3: Install and use SDKs

The following table lists the supported regions. For more information about the API operation, see Text Moderation 2.0 service in Text Moderation 2.0 SDKs and usage guide.

Region	Public endpoint	Internal endpoint
Singapore	green-cip.ap-southeast-1.aliyuncs.com	green-cip-vpc.ap-southeast-1.aliyuncs.com

Note

If you need SDK sample code in other programming languages, you can call this operation in OpenAPI Explorer. OpenAPI Explorer dynamically generates the sample code of the operation for different SDKs.

API

Usage notes

Service endpoint: https://green-cip.{region}.aliyuncs.com.

You can call the TextModeration operation to create a text moderation task. For more information about how to construct an HTTP request, see Request syntax. You can also select an existing HTTP request. For more information, see Use the Text Moderation 2.0 service.

Billing method
You are charged for calling this operation. You are charged by using the pay-as-you-go billing method only for requests whose HTTP status code is 200. For more information about the billing method, see Billing methods.

QPS limit

You can call this operation up to 100 times per second per account. If the number of calls per second exceeds the limit, throttling is triggered. As a result, your business may be affected. We recommend that you take note of the limit when you call this operation.

Request parameters

Parameter	Type	Required	Example	Description
Service	String	Yes	comment_multilingual_global	The type of the moderation service. Valid value: comment_multilingual_global: Multi-language moderation in international business scenarios.
ServiceParameters	JSONString	Yes		The parameters required by the moderation service. The value is a JSON string. For more information about the description of each string, see ServiceParameters.

Table 1 ServiceParameters

Parameter	Type	Required	Example	Description
content	String	Yes	Content to be moderated	The text to be moderated, which must be within 600 characters in length.
accountId	String	No	10123****	The account ID, which uniquely identifies an account.
deviceId	String	No	20240307****	The device ID, which uniquely identifies a device.
deviceToken	String	No	MzQvo1d7scyZ3tl_RcJ******	The device token that is obtained by using the risk control SDK.

Response parameters

Parameter	Type	Example	Description
Code	Integer	200	The returned HTTP status code. For more information, see Error codes.
Data	JSONObject		The moderation results. For more information, see Data.
Message	String	OK	The message that is returned in response to the request.
RequestId	String	AAAAAA-BBBB-CCCCC-DDDD-EEEEEEEE****	The request ID.

Table 2 Data

Parameter	Type	Example	Description
Labels	String	sexual_content	Labels. Multiple labels are separated by commas (,). Valid values: violence: violence contraband: contraband sexuality: pornography profanity: profanity and abuse pullinTraffic: advertising content regional: regional hostility C_customized: custom library that is hit
Reason	String	{"customizedWords": "HXXXXX","detectedLanguage": "en","riskTips": "sexuality_Suggestive","riskWords": "pxxxxy","translatedContent": "pxxxxy sxxxx,HXXXXX","customizedLibs": "ZXXXXX"}	The JSON string used to locate the cause. Valid values: riskLevel: the risk level, which is returned based on the risk level recommended by the system. Valid values: high medium low Note The following handling suggestions are recommended: Customers handle high-risk content directly. Manual review should be performed on medium-risk content. Low-risk content can be handled when more risky content is detected. Customers can handle the content on which no risk is detected based on their business requirements. riskTips: subcategory label. riskWords: risky words. customizedWords: customized words. customizedLibs: customized libraries. translatedContent: translated content.
AccountId	String	10123****	The ID of the Alibaba Cloud account. Note If you specify the accountId parameter in the request, the value of the accountId parameter is returned in the response.
DeviceId	String	20240307****	The device ID. Note If you specify the deviceId parameter in the request, the value of the deviceId parameter is returned in the response.

Examples

Sample requests

{
    "Service": "comment_multilingual_global",
    "ServiceParameters": {
        "content": "testing content",
        "accountId": "10123****"
    }
}

Sample responses:

If both the terms in a custom term library and risky words are detected, the following sample code is returned:

{
    "Code": 200,
    "Data": {
        "Labels": "C_customized,profanity",
        "Reason": "{\"riskLevel\":\"high\",\"customizedWords\":\"HXXXXX\",\"detectedLanguage\":\"en\",\"riskTips\":\"sexuality_Suggestive\",\"riskWords\":\"pxxxxy\",\"translatedContent\":\"pxxxxy sxxxx,HXXXXX\",\"customizedLibs\":\"ZXXXXX\"}"
    },
    "Message": "OK",
    "RequestId": "AAAAAA-BBBB-CCCCC-DDDD-EEEEEEEE****"
}

If only the terms in a custom term library are detected, the following sample code is returned:

{
    "Code": 200,
    "Data": {
        "Labels": "C_customized",
        "Reason": "{\"riskLevel\":\"high\",\"customizedWords\":\"HXXXXX\",\"detectedLanguage\":\"en\",\"translatedContent\":\"HXXXXX\",\"customizedLibs\":\"ZXXXXX\"}"
    },
    "Message": "OK",
    "RequestId": "AAAAAA-BBBB-CCCCC-DDDD-EEEEEEEE****"
}

If only risky words are detected, the following sample code is returned:

{
    "Code": 200,
    "Data": {
        "Labels": "sexuality",
        "Reason": "{\"riskLevel\":\"high\",\"detectedLanguage\":\"ar\",\"riskTips\":\"sexuality_Suggestive\",\"riskWords\":\"pxxxxy\",\"translatedContent\":\"pxxxxy sxxxx\"}"
    },
    "Message": "OK",
    "RequestId": "AAAAAA-BBBB-CCCCC-DDDD-EEEEEEEE****"
}

If neither terms in a custom term library nor risky words are detected, the following sample code is returned:

{
    "Code": 200,
    "Data": {
        "Labels": "",
        "Reason": "{\"detectedLanguage\":\"en\",\"translatedContent\":\"AXXXXXX\"}"
    },
    "Message": "OK",
    "RequestId": "AAAAAA-BBBB-CCCCC-DDDD-EEEEEEEE****"
}

Error codes

HTTP status code	Error code	Description
200	OK	The request is successful.
400	BAD_REQUEST	The request is incorrect. It may be caused by invalid request parameters. Check the request parameters carefully.
407	NOT_SUPPORT	The language is not identified or not supported.
408	PERMISSION_DENY	It may be that your account is not authorized, your account is overdue, your account is not activated, or your account is banned.
500	GENERAL_ERROR	Error. A temporary error may occur on the server. We recommend that you try again. If the error code persists, submit a ticket on Support and Services.
581	TIMEOUT	The operation timed out. We recommend that you try again. If the error code persists, submit a ticket on Support and Services.
588	EXCEED_QUOTA	The QPS exceeds the quota.