@datafire/google_commentanalyzer

Client library for Perspective Comment Analyzer API

Installation and Usage

npm install --save @datafire/google_commentanalyzer

let google_commentanalyzer = require('@datafire/google_commentanalyzer').create({
  access_token: "",
  refresh_token: "",
  client_id: "",
  client_secret: "",
  redirect_uri: ""
});

.then(data => {
  console.log(data);
});

Description

The Perspective Comment Analyzer API provides information about the potential impact of a comment on a conversation (e.g. it can provide a score for the "toxicity" of a comment). Users can leverage the "SuggestCommentScore" method to submit corrections to improve Perspective over time. Users can set the "doNotStore" flag to ensure that all submitted comments are automatically deleted after scores are returned.

Actions

oauthCallback

Exchange the code passed to your redirect URI for an access_token

google_commentanalyzer.oauthCallback({
  "code": ""
}, context)

Input

• input object
  • code required string

Output

• output object
  • access_token string
  • refresh_token string
  • token_type string
  • scope string
  • expiration string

oauthRefresh

Exchange a refresh_token for an access_token

google_commentanalyzer.oauthRefresh(null, context)

Input

This action has no parameters

Output

• output object
  • access_token string
  • refresh_token string
  • token_type string
  • scope string
  • expiration string

commentanalyzer.comments.analyze

Analyzes the provided text and returns scores for requested attributes.

google_commentanalyzer.commentanalyzer.comments.analyze({}, context)

Input

• input object
  • body AnalyzeCommentRequest
  • $.xgafv string (values: 1, 2): V1 error format.
  • access_token string: OAuth access token.
  • alt string (values: json, media, proto): Data format for response.
  • callback string: JSONP
  • fields string: Selector specifying which fields to include in a partial response.
  • key string: API key. Your API key identifies your project and provides you with API access, quota, and reports. Required unless you provide an OAuth 2.0 token.
  • oauth_token string: OAuth 2.0 token for the current user.
  • prettyPrint boolean: Returns response with indentations and line breaks.
  • quotaUser string: Available to use for quota purposes for server-side applications. Can be any arbitrary string assigned to a user, but should not exceed 40 characters.
  • upload_protocol string: Upload protocol for media (e.g. "raw", "multipart").
  • uploadType string: Legacy upload protocol for media (e.g. "media", "multipart").

Output

• output AnalyzeCommentResponse

commentanalyzer.comments.suggestscore

Suggest comment scores as training data.

google_commentanalyzer.commentanalyzer.comments.suggestscore({}, context)

Input

• input object
  • body SuggestCommentScoreRequest
  • $.xgafv string (values: 1, 2): V1 error format.
  • access_token string: OAuth access token.
  • alt string (values: json, media, proto): Data format for response.
  • callback string: JSONP
  • fields string: Selector specifying which fields to include in a partial response.
  • key string: API key. Your API key identifies your project and provides you with API access, quota, and reports. Required unless you provide an OAuth 2.0 token.
  • oauth_token string: OAuth 2.0 token for the current user.
  • prettyPrint boolean: Returns response with indentations and line breaks.
  • quotaUser string: Available to use for quota purposes for server-side applications. Can be any arbitrary string assigned to a user, but should not exceed 40 characters.
  • upload_protocol string: Upload protocol for media (e.g. "raw", "multipart").
  • uploadType string: Legacy upload protocol for media (e.g. "media", "multipart").

Output

• output SuggestCommentScoreResponse

Definitions

AnalyzeCommentRequest

• AnalyzeCommentRequest object: The comment analysis request message. LINT.IfChange
  • clientToken string: Opaque token that is echoed from the request to the response.
  • comment TextEntry
  • communityId string: Optional identifier associating this AnalyzeCommentRequest with a particular client's community. Different communities may have different norms and rules. Specifying this value enables us to explore building community-specific models for clients.
  • context Context
  • doNotStore boolean: Do not store the comment or context sent in this request. By default, the service may store comments/context for debugging purposes.
  • languages array: The language(s) of the comment and context. If none are specified, we attempt to automatically detect the language. Specifying multiple languages means the text contains multiple lanugages. Both ISO and BCP-47 language codes are accepted. The server returns an error if no language was specified and language detection fails. The server also returns an error if the languages (either specified by the caller, or auto-detected) are not all supported by the service.
    • items string
  • requestedAttributes object: Specification of requested attributes. The AttributeParameters serve as configuration for each associated attribute. The map keys are attribute names. The available attributes may be different on each RFE installation, and can be seen by calling ListAttributes (see above). For the prod installation, known as Perspective API, at blade:commentanalyzer-esf and commentanalyzer.googleapis.com, see go/checker-models (internal) and https://github.com/conversationai/perspectiveapi/blob/master/2-api/models.md#all-attribute-types.
  • sessionId string: Session ID. Used to join related RPCs into a single session. For example, an interactive tool that calls both the AnalyzeComment and SuggestCommentScore RPCs should set all invocations of both RPCs to the same Session ID, typically a random 64-bit integer.
  • spanAnnotations boolean: An advisory parameter that will return span annotations if the model is capable of providing scores with sub-comment resolution. This will likely increase the size of the returned message.

AnalyzeCommentResponse

• AnalyzeCommentResponse object: The comment analysis response message.
  • attributeScores object: Scores for the requested attributes. The map keys are attribute names (same as the requested_attribute field in AnalyzeCommentRequest, for example "ATTACK_ON_AUTHOR", "INFLAMMATORY", etc).
  • clientToken string: Same token from the original AnalyzeCommentRequest.
  • detectedLanguages array: Contains the languages detected from the text content, sorted in order of likelihood.
    • items string
  • languages array: The language(s) used by CommentAnalyzer service to choose which Model to use when analyzing the comment. Might better be called "effective_languages". The logic used to make the choice is as follows: if !Request.languages.empty() effective_languages = Request.languages else effective_languages = detected_languages[0]
    • items string

ArticleAndParentComment

• ArticleAndParentComment object: A type of context specific to a comment left on a single-threaded comment message board, where comments are either a top level comment or the child of a top level comment.
  • article TextEntry
  • parentComment TextEntry

AttributeParameters

• AttributeParameters object: Configurable parameters for attribute scoring.
  • scoreThreshold number: Don't return scores for this attribute that are below this threshold. If unset, a default threshold will be applied. A FloatValue wrapper is used to distinguish between 0 vs. default/unset.
  • scoreType string (values: SCORE_TYPE_UNSPECIFIED, PROBABILITY, STD_DEV_SCORE, PERCENTILE, RAW): What type of scores to return. If unset, defaults to probability scores.

AttributeScores

• AttributeScores object: This holds score values for a single attribute. It contains both per-span scores as well as an overall summary score..
  • spanScores array: Per-span scores.
    • items SpanScore
  • summaryScore Score

Context

• Context object: Context is typically something that a Comment is referencing or replying to (such as an article, or previous comment). Note: Populate only ONE OF the following fields. The oneof syntax cannot be used because that would require nesting entries inside another message and breaking backwards compatibility. The server will return an error if more than one of the following fields is present.
  • articleAndParentComment ArticleAndParentComment
  • entries array: A list of messages. For example, a linear comments section or forum thread.
    • items TextEntry

Score

• Score object: Analysis scores are described by a value and a ScoreType.
  • type string (values: SCORE_TYPE_UNSPECIFIED, PROBABILITY, STD_DEV_SCORE, PERCENTILE, RAW): The type of the above value.
  • value number: Score value. Semantics described by type below.

SpanScore

• SpanScore object: This is a single score for a given span of text.
  • begin integer: "begin" and "end" describe the span of the original text that the attribute score applies to. The values are the UTF-16 codepoint range. "end" is exclusive. For example, with the text "Hi there", the begin/end pair (0,2) describes the text "Hi". If "begin" and "end" are unset, the score applies to the full text.
  • end integer
  • score Score

SuggestCommentScoreRequest

• SuggestCommentScoreRequest object: The comment score suggestion request message.
  • attributeScores object: Attribute scores for the comment. The map keys are attribute names, same as the requested_attribute field in AnalyzeCommentRequest (for example "ATTACK_ON_AUTHOR", "INFLAMMATORY", etc.). This field has the same type as the attribute_scores field in AnalyzeCommentResponse. To specify an overall attribute score for the entire comment as a whole, use the summary_score field of the mapped AttributeScores object. To specify scores on specific subparts of the comment, use the span_scores field. All SpanScore objects must have begin and end fields set. All Score objects must be explicitly set (for binary classification, use the score values 0 and 1). If Score objects don't include a ScoreType, PROBABILITY is assumed. attribute_scores must not be empty. The mapped AttributeScores objects also must not be empty. An INVALID_ARGUMENT error is returned for all malformed requests.
  • clientToken string: Opaque token that is echoed from the request to the response.
  • comment TextEntry
  • communityId string: Optional identifier associating this comment score suggestion with a particular sub-community. Different communities may have different norms and rules. Specifying this value enables training community-specific models.
  • context Context
  • languages array: The language(s) of the comment and context. If none are specified, we attempt to automatically detect the language. Both ISO and BCP-47 language codes are accepted.
    • items string
  • sessionId string: Session ID. Used to join related RPCs into a single session. For example, an interactive tool that calls both the AnalyzeComment and SuggestCommentScore RPCs should set all invocations of both RPCs to the same Session ID, typically a random 64-bit integer.

SuggestCommentScoreResponse

• SuggestCommentScoreResponse object: The comment score suggestion response message.
  • clientToken string: Same token from the original SuggestCommentScoreRequest.
  • detectedLanguages array: The list of languages detected from the comment text.
    • items string
  • requestedLanguages array: The list of languages provided in the request.
    • items string

TextEntry

• TextEntry object: Represents a body of text.
  • text string: UTF-8 encoded text.
  • type string (values: TEXT_TYPE_UNSPECIFIED, PLAIN_TEXT, HTML): Type of the text field.