Analyzes the provided text and returns scores for requested attributes.
suggestscore(body, x__xgafv=None)
Suggest comment scores as training data.
analyze(body, x__xgafv=None)
Analyzes the provided text and returns scores for requested attributes.
Args:
body: object, The request body. (required)
The object takes the form of:
{ # The comment analysis request message.
"comment": { # Represents a body of text. # The comment to analyze.
"text": "A String", # UTF-8 encoded text.
"type": "A String", # Type of the text field.
},
"doNotStore": True or False, # Do not store the comment or context sent in this request. By default, the
# service may store comments/context for debugging purposes.
"languages": [ # The language(s) of the comment and context (if none are specified, the
# language is automatically detected). If multiple languages are specified,
# the text is checked in all of them that are supported. Both ISO and BCP-47
# language codes are accepted.
# Current Language Restrictions:
# * Only English text ("en") is supported.
# If none of the languages specified by the caller are supported, an
# `UNIMPLEMENTED` error is returned.
"A String",
],
"sessionId": "A 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.
"context": { # Context is typically something that a Comment is referencing or replying to # The context of the comment.
# (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": { # A type of context specific to a comment left on a single-threaded comment # Information about the source for which the original comment was made, and
# any parent comment info.
# message board, where comments are either a top level comment or the child of
# a top level comment.
"article": { # Represents a body of text. # The source content about which the comment was made (article text, article
# summary, video transcript, etc).
"text": "A String", # UTF-8 encoded text.
"type": "A String", # Type of the text field.
},
"parentComment": { # Represents a body of text. # Refers to text that is a direct parent of the source comment, such as in a
# one-deep threaded message board. This field will only be present for
# comments that are replies to other comments and will not be populated for
# direct comments on the article_text.
"text": "A String", # UTF-8 encoded text.
"type": "A String", # Type of the text field.
},
},
"entries": [ # A list of messages. For example, a linear comments section or forum thread.
{ # Represents a body of text.
"text": "A String", # UTF-8 encoded text.
"type": "A String", # Type of the text field.
},
],
},
"clientToken": "A String", # Opaque token that is echoed from the request to the response.
"requestedAttributes": { # Specification of requested attributes. The AttributeParameters serve as
# configuration for each associated attribute. The map keys are attribute
# names. The following attributes are available:
# "ATTACK_ON_AUTHOR" - Attack on author of original article or post.
# "ATTACK_ON_COMMENTER" - Attack on fellow commenter.
# "ATTACK_ON_PUBLISHER" - Attack on publisher of article/post.
# "INCOHERENT" - Difficult to understand, nonsensical.
# "INFLAMMATORY" - Intending to provoke or inflame.
# "OBSCENE" - Obscene, such as cursing.
# "OFF_TOPIC" - Not related to the original topic.
# "SPAM" - Commercial/advertising spam content.
# "UNSUBSTANTIAL" - Trivial.
"a_key": { # Configurable parameters for attribute scoring.
"scoreThreshold": 3.14, # 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": "A String", # What type of scores to return. If unset, defaults to probability scores.
},
},
"spanAnnotations": True or False, # 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.
"communityId": "A 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.
}
x__xgafv: string, V1 error format.
Allowed values
1 - v1 error format
2 - v2 error format
Returns:
An object of the form:
{ # The comment analysis response message.
"languages": [ # 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 = detected_languages
# else
# effective_languages = Request.languages
"A String",
],
"detectedLanguages": [ # Contains the languages detected from the text content, sorted in order of
# likelihood.
"A String",
],
"attributeScores": { # 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).
"a_key": { # This holds score values for a single attribute. It contains both per-span
# scores as well as an overall summary score..
"summaryScore": { # Analysis scores are described by a value and a ScoreType. # Overall score for comment as a whole.
"type": "A String", # The type of the above value.
"value": 3.14, # Score value. Semantics described by type below.
},
"spanScores": [ # Per-span scores.
{ # This is a single score for a given span of text.
"begin": 42, # "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.
"score": { # Analysis scores are described by a value and a ScoreType. # The score value.
"type": "A String", # The type of the above value.
"value": 3.14, # Score value. Semantics described by type below.
},
"end": 42,
},
],
},
},
"clientToken": "A String", # Same token from the original AnalyzeCommentRequest.
}
suggestscore(body, x__xgafv=None)
Suggest comment scores as training data.
Args:
body: object, The request body. (required)
The object takes the form of:
{ # The comment score suggestion request message.
"comment": { # Represents a body of text. # The comment being scored.
"text": "A String", # UTF-8 encoded text.
"type": "A String", # Type of the text field.
},
"languages": [ # The language(s) of the comment and context (if none are specified, the
# language is automatically detected). If multiple languages are specified,
# the text is checked in all of them that are supported. Both ISO and BCP-47
# language codes are accepted.
# Current Language Restrictions:
# * Only English text ("en") is supported.
# If none of the languages specified by the caller are supported, an
# `UNIMPLEMENTED` error is returned.
"A String",
],
"sessionId": "A 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.
"context": { # Context is typically something that a Comment is referencing or replying to # The context of the comment.
# (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": { # A type of context specific to a comment left on a single-threaded comment # Information about the source for which the original comment was made, and
# any parent comment info.
# message board, where comments are either a top level comment or the child of
# a top level comment.
"article": { # Represents a body of text. # The source content about which the comment was made (article text, article
# summary, video transcript, etc).
"text": "A String", # UTF-8 encoded text.
"type": "A String", # Type of the text field.
},
"parentComment": { # Represents a body of text. # Refers to text that is a direct parent of the source comment, such as in a
# one-deep threaded message board. This field will only be present for
# comments that are replies to other comments and will not be populated for
# direct comments on the article_text.
"text": "A String", # UTF-8 encoded text.
"type": "A String", # Type of the text field.
},
},
"entries": [ # A list of messages. For example, a linear comments section or forum thread.
{ # Represents a body of text.
"text": "A String", # UTF-8 encoded text.
"type": "A String", # Type of the text field.
},
],
},
"clientToken": "A String", # Opaque token that is echoed from the request to the response.
"communityId": "A 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.
"attributeScores": { # 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.
"a_key": { # This holds score values for a single attribute. It contains both per-span
# scores as well as an overall summary score..
"summaryScore": { # Analysis scores are described by a value and a ScoreType. # Overall score for comment as a whole.
"type": "A String", # The type of the above value.
"value": 3.14, # Score value. Semantics described by type below.
},
"spanScores": [ # Per-span scores.
{ # This is a single score for a given span of text.
"begin": 42, # "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.
"score": { # Analysis scores are described by a value and a ScoreType. # The score value.
"type": "A String", # The type of the above value.
"value": 3.14, # Score value. Semantics described by type below.
},
"end": 42,
},
],
},
},
}
x__xgafv: string, V1 error format.
Allowed values
1 - v1 error format
2 - v2 error format
Returns:
An object of the form:
{ # The comment score suggestion response message.
"requestedLanguages": [ # The list of languages provided in the request.
"A String",
],
"detectedLanguages": [ # The list of languages detected from the comment text.
"A String",
],
"clientToken": "A String", # Same token from the original SuggestCommentScoreRequest.
}