tencent cloud

Feedback

Synchronized Moderation of Text Content

Last updated: 2024-06-12 16:01:53

    Feature Overview

    This API is designed for the direct moderation of text content, operating on a synchronous request basis.
    The following lists the characteristics of this API:
    Note:
    Plaintext needs to be Base64-encoded (UTF8 encoding only) before direct moderation is implemented.
    Recognition of multiple non-compliant scenes is supported, including pornographic, illegal, and advertising scenes.
    Setting Moderation Policy supports diverse business scenarios.

    Authorization Guidelines

    In the authorization policy, set action to ci:CreateAuditingTextJob. View all actions.

    Service Activation

    Cloud Infinite (CI) needs to be enabled with buckets bound before you use this function. For details, see Bind Bucket.

    Prerequisites

    Before using this API, please check all relevant limitations. For details, see Specifications and Limits.

    Pricing

    Each moderation scene is billed separately. For example, if you select moderation for both pornographic and advertising scenes, moderating one piece of text will result in charges for two moderation actions.
    Length limit: Plaintext needs to be Base64-encoded before direct moderation is implemented. The original text length before encoding must not exceed 10,000 UTF8-encoded characters.
    Language support in text moderation: Currently, the system supports detection in Chinese, mixed Chinese and English, and Arabic numerals.
    The COS SDK offers complete demo, automated integration, signature calculations, and other capabilities. The SDK allows for a more convenient and swift way to call APIs. Click here to view SDK documentation.

    Request

    Request Example

    POST /text/auditing HTTP/1.1
    Host: <BucketName-APPID>.ci.<Region>.myqcloud.com
    Date: <GMT Date>
    Authorization: <Auth String>
    Content-Length: <length>
    Content-Type: application/xml
    
    <body>
    Note:
    Authorization: Auth String (For more information, see Request Signature.)
    When it is used through a sub-account, related permissions need to be assigned. For details, see Authorization Granularity Details.

    Request Header

    This API only uses common request headers. For details, see Common Request Headers.

    Request Body

    The implementation of this request action requires the following request body:
    <Request>
    <Input>
    <Content></Content>
    <DataId></DataId>
    </Input>
    <Conf>
    <Callback></Callback>
    <CallbackVersion></CallbackVersion>
    <CallbackType></CallbackType>
    <BizType></BizType>
    <Freeze>
    <PornScore></PornScore>
    </Freeze>
    </Conf>
    </Request>
    The following details the data description:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    Required or Not
    Request
    No
    The specific configuration item for text moderation.
    Container
    Yes
    The data description for Request of the Container type is as follows:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    Required or Not
    Input
    Request
    Content requiring moderation.
    Container
    Yes
    Conf
    Request
    Configuration of moderation rules.
    Container
    Yes
    The data description for Input of the Container type is as follows:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    Required or Not
    Content
    Request.Input
    When the input is plaintext, it should first be Base64-encoded. The text length before encoding cannot exceed 10,000 UTF-8 encoded characters. If this length limit is exceeded, the API will return an error.Note: Currently, detection and moderation are supported in Chinese, mixed Chinese and English, and Arabic numerals. If you need to implement moderation in other languages, please contact us.
    String
    No
    DataId
    Request.Input
    This field in the moderation result will return the original content, with a length limit of 512 bytes. You can use this field to uniquely identify the data to be moderated.
    String
    No
    UserInfo
    Request.Input
    User business field.
    Container
    No
    UserInfo content of the Container node:
    Node Name (Keyword)
    Description
    Type
    Required or Not
    TokenId
    It usually indicates account information, limited to 128 bytes in length.
    String
    No
    Nickname
    It usually indicates nickname information, limited to 128 bytes in length.
    String
    No
    DeviceId
    It usually indicates device information, limited to 128 bytes in length.
    String
    No
    AppId
    It usually indicates a unique identifier for the App, limited to 128 bytes in length.
    String
    No
    Room
    It usually indicates room number information, limited to 128 bytes in length.
    String
    No
    IP
    It usually indicates IP address information, limited to 128 bytes in length.
    String
    No
    Type
    It usually indicates the business type, limited to 128 bytes in length.
    String
    No
    ReceiveTokenId
    It usually indicates the user account for receiving messages, limited to 128 bytes in length.
    String
    No
    Gender
    It usually indicates gender data, limited to 128 bytes in length.
    String
    No
    Level
    It usually indicates level information, limited to 128 bytes in length.
    String
    No
    Role
    It usually indicates role details, limited to 128 bytes in length.
    String
    No
    The data description for Conf of the Container type is as follows:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    Required or Not
    BizType
    Request.Conf
    It indicates a unique identifier for the moderation policy. You can set the scenes you want to moderate, such as pornography, advertising, or illegal content, through the moderation policy page on the console. Follow these guidelines to configure: Setting Moderation Policy. BizType can be obtained from the console. When you specify BizType, this moderation request will enable moderation according to the scenes configured in the policy. If BizType is left blank, the default moderation policy will be applied automatically.
    String
    No
    Callback
    Request.Conf
    The moderation results can be sent to your callback address in a callback form. It supports addresses starting with http:// or https://, such as: http://www.callback.com. When Input is set to Content, this parameter is not effective, and results will be returned directly.
    String
    No
    CallbackVersion
    Request.Conf
    The structure of the callback content. Valid values: Simple (the callback content includes basic information), Detail (the callback content includes detailed information). The default value is Simple.
    String
    No
    CallbackType
    Request.Conf
    The type of callback segment. Valid values: 1 (all text segments are recalled), 2 (non-compliant text segments are recalled). The default value is 1.
    Integer
    No
    Freeze
    Request.Conf
    This field allows you to automatically block text files based on the different scores given according to the moderation results. It is effective only when the text moderated in input is object.
    Container
    No
    The data description for Freeze of the Container type is as follows:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    Required or Not
    PornScore
    Request.Conf.Freeze
    The value range is [0,100]. A block operation will be carried out automatically when the pornographic content moderation score equals or exceeds the given score. If it is left blank, an automatic block operation will not occur. The default value is null.
    Integer
    No
    AdsScore
    Request.Conf.Freeze
    The value range is [0,100]. The file will be automatically blocked when the advertising moderation score equals or exceeds the specified score. If it is left blank, an automatic block operation will not occur. The default value is null.
    Integer
    No
    IllegalScore
    Request.Conf.Freeze
    The value range is [0,100]. The file will be automatically blocked if the illegal content moderation result equals or exceeds this score. If it is left blank, an automatic block operation will not occur. The default value is null.
    Integer
    No
    AbuseScore
    Request.Conf.Freeze
    The value range is [0,100]. The file will be automatically blocked if the verbal abuse moderation result equals or exceeds this score. If it is left blank, an automatic block operation will not occur. The default value is null.
    Integer
    No
    For block parameters in other moderation scenarios, please contact our service team.

    Response

    Response Header

    This API only returns common response headers. For more information, see Common Response Headers.

    Response Body

    The response body returns application/xml data. The following displays complete node data:
    <Response>
    <JobsDetail>
    <DataId></DataId>
    <JobId></JobId>
    <State></State>
    <CreationTime></CreationTime>
    <Code>Success</Code>
    <Message>Success</Message>
    <SectionCount></SectionCount>
    <Result>1</Result>
    <PornInfo>
    <HitFlag></HitFlag>
    <Count></Count>
    </PornInfo>
    <Section>
    <StartByte></StartByte>
    <PornInfo>
    <HitFlag></HitFlag>
    <Score></Score>
    <Keywords></Keywords>
    </PornInfo>
    </Section>
    </JobsDetail>
    <RequestId></RequestId>
    </Response>
    The specific data content is as follows:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    Response
    No
    The specific response content returned by text moderation.
    Container
    `Response` content of the Container node:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    JobsDetail
    Response
    Detailed information of text moderation.
    Container
    RequestId
    Response
    When a request is send, the server automatically creates an ID specific to the request, assisting in locating encountered problems.
    String
    JobsDetail content of the Container node:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    Code
    Response.JobsDetail
    Error code. It is returned only if State is Failed. For more information, see Error Codes.
    String
    DataId
    Response.JobsDetail
    Unique business identifier added in the request.
    String
    Message
    Response.JobsDetail
    Error description. It is returned only when State is Failed.
    String
    JobId
    Response.JobsDetail
    Task ID of this text moderation operation.
    String
    CreationTime
    Response.JobsDetail
    Creation time of the text moderation.
    String
    State
    Response.JobsDetail
    Status of the text moderation. Valid values include Success (moderation succeeded) and Failed (moderation failed).
    String
    Content
    Response.JobsDetail
    Base64-encoded text content.
    String
    ContextText
    Response.JobsDetail
    When the text context correlation moderation capability is activated, this field, along with the current text under moderation and its associated text, will be returned in their original forms.Note: When you use the context correlation capabilities, it is necessary to incorporate the UserInfo field during the initiation of a text moderation operation. It indicates that the correlated context is specific to a particular user ID, and the text content uploaded by different user IDs will not be correlated.To enable context correlation capabilities, please contact our service team for activation.
    String
    Label
    Response.JobsDetail
    This field returns the moderation results which correspond to the malicious tag with the highest priority and are recommended by the model. It is recommended that you handle different types of violations and suggested values based on the business requirements. Returned values include: Normal: normal, Porn: pornography, Ads: advertising, along with other types of unsafe or inappropriate content.
    String
    Result
    Response.JobsDetail
    This field indicates the moderation result of the current assessment. You can perform subsequent operations based on the results.
    Valid values: 0 (normal), 1 (sensitive and non-compliant files), and 2 (possibly sensitive, with manual moderation recommended).
    Integer
    SectionCount
    Response.JobsDetail
    The number of content segments for text moderation. The value is fixed at 1.
    Integer
    PornInfo
    Response.JobsDetail
    The moderation result of the pornographic content moderation scene.
    Container
    AdsInfo
    Response.JobsDetail
    The moderation result of the advertising content moderation scene.
    Container
    IllegalInfo
    Response.JobsDetail
    The moderation result of the illegal content moderation scene.
    Container
    AbuseInfo
    Response.JobsDetail
    The moderation result of the abusive content moderation scene.
    Container
    Section
    Response.JobsDetail
    The specific result information of text moderation.
    Container Array
    UserInfo
    Response.JobsDetail
    User business field.
    Container
    ListInfo
    Response.JobsDetail
    Account whitelist/blacklist results.
    Container
    PornInfo, AdsInfo, IllegalInfo, AbuseInfo content of the Container node:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    HitFlag
    Response.JobsDetail.*Info
    It is used to return moderation results of the corresponding scene. Returned values: 0: normal, 1: confirmed as violation content of the current scene, 2: suspected as violation content of the current scene.
    Integer
    Count
    Response.JobsDetail.*Info
    The number of segments that match the moderation classification.
    Integer
    Score
    Response.JobsDetail.*Info
    This field indicates the confidence of the matched moderation information, with a range of 0 (lowest confidence) to 100 (highest confidence). A higher confidence indicates a higher possibility that the content belongs to the returned moderation information. For example, if the confidence for pornography is 99, the content is highly likely to be pornographic content.
    Integer
    Section content of the Container node:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    StartByte
    Response.JobsDetail.Section
    The location within the text where the segment begins (that is, 10 represents the 11th UTF-8 character). It starts from 0.
    Integer
    Label
    Response.JobsDetail.Section
    This field returns the moderation results which correspond to the malicious tag with the highest priority and are recommended by the model. It is recommended that you handle different types of violations and suggested values based on the business requirements. Returned values include: Normal: normal, Porn: pornography, Ads: advertising, along with other types of unsafe or inappropriate content.
    String
    Result
    Response.JobsDetail.Section
    This field indicates the moderation result of the current assessment. You can perform subsequent operations based on the results. Valid values: 0 (normal), 1 (sensitive and non-compliant files), and 2 (possibly sensitive, with manual moderation recommended).
    Integer
    PornInfo
    Response.JobsDetail.Section
    The moderation result of the pornographic content moderation scene.
    Container
    AdsInfo
    Response.JobsDetail.Section
    The moderation result of the advertising content moderation scene.
    Container
    IllegalInfo
    Response.JobsDetail.Section
    The moderation result of the illegal content moderation scene.
    Container
    AbuseInfo
    Response.JobsDetail.Section
    The moderation result of the abusive content moderation scene.
    Container
    PornInfo, AdsInfo, IllegalInfo, AbuseInfo content of the Container node:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    HitFlag
    Response.JobsDetail.Section.*Info
    It is used to return moderation results of the corresponding scene. Returned values: 0: normal, 1: confirmed as violation content of the current scene, 2: suspected as violation content of the current scene.
    Integer
    Score
    Response.JobsDetail.Section.*Info
    The moderation result score within this segment. Higher scores indicate more sensitive content.
    Integer
    Keywords
    Response.JobsDetail.Section.*Info
    Keywords matched under the current moderation scene. Multiple keywords are separated by ",".
    String
    LibResults
    Response.JobsDetail.Section.*Info
    This field is used to return results identified through the text risk library. Note: This field is not returned when no samples within the risk library have been matched.
    Container Array
    SubLabel
    Response.JobsDetail.Section.*Info
    This field indicates the specific sub-tags matched in the moderation. For example: the SexBehavior sub-tag under Porn. Note: This field may return null, indicating that no specific sub-tags are matched.
    String
    `LibResults` content of the `Container` node:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    LibType
    Response.JobsDetail.Section.*Info.LibResults
    Type of the matched text risk library. Valid values include 1 (preset text library) and 2 (custom text risk library).
    Integer
    LibName
    Response.JobsDetail.Section.*Info.LibResults
    Name of the risk library hit.
    String
    Keywords
    Response.JobsDetail.Section.*Info.LibResults
    Keywords matched in the library. This parameter may return multiple values, indicating multiple keywords matched.
    String Array
    UserInfo content of the Container node:
    Node Name (Keyword)
    Description
    Type
    Required or Not
    TokenId
    It usually indicates account information, limited to 128 bytes in length.
    String
    No
    Nickname
    It usually indicates nickname information, limited to 128 bytes in length.
    String
    No
    DeviceId
    It usually indicates device information, limited to 128 bytes in length.
    String
    No
    AppId
    It usually indicates a unique identifier for the App, limited to 128 bytes in length.
    String
    No
    Room
    It usually indicates room number information, limited to 128 bytes in length.
    String
    No
    IP
    It usually indicates IP address information, limited to 128 bytes in length.
    String
    No
    Type
    It usually indicates the business type, limited to 128 bytes in length.
    String
    No
    ReceiveTokenId
    It usually indicates the user account for receiving messages, limited to 128 bytes in length.
    String
    No
    Gender
    It usually indicates gender data, limited to 128 bytes in length.
    String
    No
    Level
    It usually indicates level information, limited to 128 bytes in length.
    String
    No
    Role
    It usually indicates role details, limited to 128 bytes in length.
    String
    No
    ListInfo content of the Container node:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    ListResults
    Response.JobsDetail.ListInfo
    All matched business field risk library results.
    Container Array
    ListResults content of the Container node:
    Node Name (Keyword)
    Parent Node
    Description
    Type
    ListType
    Response.JobsDetail.ListInfo.ListResults
    Type of matched business field risk library. Valid values include 0 (white library) and 1(black library).
    Integer
    ListName
    Response.JobsDetail.ListInfo.ListResults
    Name of the matched business field risk library.
    String
    Entity
    Response.JobsDetail.ListInfo.ListResults
    Matched entry in the business field risk library.
    String

    Error Code

    There is no specific error information for this request operation. For common error information, see Error Codes.

    Actual Case

    Request

    POST /text/auditing HTTP/1.1
    Authorization: q-sign-algorithm=sha1&q-ak=AKIDZfbOAo7cllgPvF9cXFrJD0a1ICvR****&q-sign-time=1497530202;1497610202&q-key-time=1497530202;1497610202&q-header-list=&q-url-param-list=&q-signature=28e9a4986df11bed0255e97ff90500557e0e****
    Host: examplebucket-1250000000.ci.ap-beijing.myqcloud.com
    Content-Length: 166
    Content-Type: application/xml
    <Request>
    <Input>
    <Content>54uZ5Ye75omL</Content>
    </Input>
    <Conf>
    <BizType>b81d45f94b91a683255e9a9506f45a11</BizType>
    </Conf>
    </Request>

    Response

    HTTP/1.1 200 OK
    Content-Type: application/xml
    Content-Length: 230
    Connection: keep-alive
    Date: Thu, 15 Jun 2017 12:37:29 GMT
    Server: tencent-ci
    x-ci-request-id: NTk0MjdmODlfMjQ4OGY3XzYzYzhf****
    <Response>
    <JobsDetail>
    <JobId>vab1ca9fc8a3ed11ea834c525400863904</JobId>
    <Content>54uZ5Ye75omL</Content>
    <State>Success</State>
    <CreationTime>2019-07-07T12:12:12+0800</CreationTime>
    <SectionCount>1</SectionCount>
    <Label>Illegal</Label>
    <Result>2</Result>
    <PornInfo>
    <HitFlag>0</HitFlag>
    <Count>0</Count>
    </PornInfo>
    <Section>
    <StartByte>0</StartByte>
    <Label>Illegal</Label>
    <Result>2</Result>
    <PornInfo>
    <HitFlag>0</HitFlag>
    <Score>0</Score>
    <Keywords/>
    </PornInfo>
    </Section>
    </JobsDetail>
    <RequestId>xxxxxxxxxxxxxx</RequestId>
    </Response>
    
    Contact Us

    Contact our sales team or business advisors to help your business.

    Technical Support

    Open a ticket if you're looking for further assistance. Our Ticket is 7x24 avaliable.

    7x24 Phone Support