2019/11/21 - Amazon Lex Runtime Service - 2 updated api methods
Changes Amazon Lex now supports Sentiment Analysis
{'sentimentResponse': 'string'}
Sends user input (text or speech) to Amazon Lex. Clients use this API to send text and audio requests to Amazon Lex at runtime. Amazon Lex interprets the user input using the machine learning model that it built for the bot.
The PostContent operation supports audio input at 8kHz and 16kHz. You can use 8kHz audio to achieve higher speech recognition accuracy in telephone audio applications.
In response, Amazon Lex returns the next message to convey to the user. Consider the following example messages:
For a user input "I would like a pizza," Amazon Lex might return a response with a message eliciting slot data (for example, PizzaSize ): "What size pizza would you like?".
After the user provides all of the pizza order information, Amazon Lex might return a response with a message to get user confirmation: "Order the pizza?".
After the user replies "Yes" to the confirmation prompt, Amazon Lex might return a conclusion statement: "Thank you, your cheese pizza has been ordered.".
Not all Amazon Lex messages require a response from the user. For example, conclusion statements do not require a response. Some messages require only a yes or no response. In addition to the message , Amazon Lex provides additional context about the message in the response that you can use to enhance client behavior, such as displaying the appropriate client user interface. Consider the following examples:
If the message is to elicit slot data, Amazon Lex returns the following context information:
x-amz-lex-dialog-state header set to ElicitSlot
x-amz-lex-intent-name header set to the intent name in the current context
x-amz-lex-slot-to-elicit header set to the slot name for which the message is eliciting information
x-amz-lex-slots header set to a map of slots configured for the intent with their current values
If the message is a confirmation prompt, the x-amz-lex-dialog-state header is set to Confirmation and the x-amz-lex-slot-to-elicit header is omitted.
If the message is a clarification prompt configured for the intent, indicating that the user intent is not understood, the x-amz-dialog-state header is set to ElicitIntent and the x-amz-slot-to-elicit header is omitted.
In addition, Amazon Lex also returns your application-specific sessionAttributes . For more information, see Managing Conversation Context .
See also: AWS API Documentation
Request Syntax
client.post_content(
botName='string',
botAlias='string',
userId='string',
sessionAttributes={...}|[...]|123|123.4|'string'|True|None,
requestAttributes={...}|[...]|123|123.4|'string'|True|None,
contentType='string',
accept='string',
inputStream=b'bytes'|file
)
string
[REQUIRED]
Name of the Amazon Lex bot.
string
[REQUIRED]
Alias of the Amazon Lex bot.
string
[REQUIRED]
The ID of the client application user. Amazon Lex uses this to identify a user's conversation with your bot. At runtime, each request must contain the userID field.
To decide the user ID to use for your application, consider the following factors.
The userID field must not contain any personally identifiable information of the user, for example, name, personal identification numbers, or other end user personal information.
If you want a user to start a conversation on one device and continue on another device, use a user-specific identifier.
If you want the same user to be able to have two independent conversations on two different devices, choose a device-specific identifier.
A user can't have two independent conversations with two different versions of the same bot. For example, a user can't have a conversation with the PROD and BETA versions of the same bot. If you anticipate that a user will need to have conversation with two different versions, for example, while testing, include the bot alias in the user ID to separate the two conversations.
JSON serializable
You pass this value as the x-amz-lex-session-attributes HTTP header.
Application-specific information passed between Amazon Lex and a client application. The value must be a JSON serialized and base64 encoded map with string keys and values. The total size of the sessionAttributes and requestAttributes headers is limited to 12 KB.
For more information, see Setting Session Attributes .
JSON serializable
You pass this value as the x-amz-lex-request-attributes HTTP header.
Request-specific information passed between Amazon Lex and a client application. The value must be a JSON serialized and base64 encoded map with string keys and values. The total size of the requestAttributes and sessionAttributes headers is limited to 12 KB.
The namespace x-amz-lex: is reserved for special attributes. Don't create any request attributes with the prefix x-amz-lex: .
For more information, see Setting Request Attributes .
string
[REQUIRED]
You pass this value as the Content-Type HTTP header.
Indicates the audio format or text. The header value must start with one of the following prefixes:
PCM format, audio data must be in little-endian byte order.
audio/l16; rate=16000; channels=1
audio/x-l16; sample-rate=16000; channel-count=1
audio/lpcm; sample-rate=8000; sample-size-bits=16; channel-count=1; is-big-endian=false
Opus format
audio/x-cbr-opus-with-preamble; preamble-size=0; bit-rate=256000; frame-size-milliseconds=4
Text format
text/plain; charset=utf-8
string
You pass this value as the Accept HTTP header.
The message Amazon Lex returns in the response can be either text or speech based on the Accept HTTP header value in the request.
If the value is text/plain; charset=utf-8 , Amazon Lex returns text in the response.
If the value begins with audio/ , Amazon Lex returns speech in the response. Amazon Lex uses Amazon Polly to generate the speech (using the configuration you specified in the Accept header). For example, if you specify audio/mpeg as the value, Amazon Lex returns speech in the MPEG format.
If the value is audio/pcm , the speech returned is audio/pcm in 16-bit, little endian format.
The following are the accepted values:
audio/mpeg
audio/ogg
audio/pcm
text/plain; charset=utf-8
audio/* (defaults to mpeg)
bytes or seekable file-like object
[REQUIRED]
User input in PCM or Opus audio format or text format as described in the Content-Type HTTP header.
You can stream audio data to Amazon Lex or you can create a local buffer that captures all of the audio data before sending. In general, you get better performance if you stream audio data rather than buffering the data locally.
dict
Response Syntax
{
'contentType': 'string',
'intentName': 'string',
'slots': {...}|[...]|123|123.4|'string'|True|None,
'sessionAttributes': {...}|[...]|123|123.4|'string'|True|None,
'sentimentResponse': 'string',
'message': 'string',
'messageFormat': 'PlainText'|'CustomPayload'|'SSML'|'Composite',
'dialogState': 'ElicitIntent'|'ConfirmIntent'|'ElicitSlot'|'Fulfilled'|'ReadyForFulfillment'|'Failed',
'slotToElicit': 'string',
'inputTranscript': 'string',
'audioStream': b'bytes'|file
}
Response Structure
(dict) --
contentType (string) --
Content type as specified in the Accept HTTP header in the request.
intentName (string) --
Current user intent that Amazon Lex is aware of.
slots (JSON serializable) --
Map of zero or more intent slots (name/value pairs) Amazon Lex detected from the user input during the conversation. The field is base-64 encoded.
Amazon Lex creates a resolution list containing likely values for a slot. The value that it returns is determined by the valueSelectionStrategy selected when the slot type was created or updated. If valueSelectionStrategy is set to ORIGINAL_VALUE , the value provided by the user is returned, if the user value is similar to the slot values. If valueSelectionStrategy is set to TOP_RESOLUTION Amazon Lex returns the first value in the resolution list or, if there is no resolution list, null. If you don't specify a valueSelectionStrategy , the default is ORIGINAL_VALUE .
sessionAttributes (JSON serializable) --
Map of key/value pairs representing the session-specific context information.
sentimentResponse (string) --
The sentiment expressed in and utterance.
When the bot is configured to send utterances to Amazon Comprehend for sentiment analysis, this field contains the result of the analysis.
message (string) --
The message to convey to the user. The message can come from the bot's configuration or from a Lambda function.
If the intent is not configured with a Lambda function, or if the Lambda function returned Delegate as the dialogAction.type in its response, Amazon Lex decides on the next course of action and selects an appropriate message from the bot's configuration based on the current interaction context. For example, if Amazon Lex isn't able to understand user input, it uses a clarification prompt message.
When you create an intent you can assign messages to groups. When messages are assigned to groups Amazon Lex returns one message from each group in the response. The message field is an escaped JSON string containing the messages. For more information about the structure of the JSON string returned, see msg-prompts-formats .
If the Lambda function returns a message, Amazon Lex passes it to the client in its response.
messageFormat (string) --
The format of the response message. One of the following values:
PlainText - The message contains plain UTF-8 text.
CustomPayload - The message is a custom format for the client.
SSML - The message contains text formatted for voice output.
Composite - The message contains an escaped JSON object containing one or more messages from the groups that messages were assigned to when the intent was created.
dialogState (string) --
Identifies the current state of the user interaction. Amazon Lex returns one of the following values as dialogState . The client can optionally use this information to customize the user interface.
ElicitIntent - Amazon Lex wants to elicit the user's intent. Consider the following examples: For example, a user might utter an intent ("I want to order a pizza"). If Amazon Lex cannot infer the user intent from this utterance, it will return this dialog state.
ConfirmIntent - Amazon Lex is expecting a "yes" or "no" response. For example, Amazon Lex wants user confirmation before fulfilling an intent. Instead of a simple "yes" or "no" response, a user might respond with additional information. For example, "yes, but make it a thick crust pizza" or "no, I want to order a drink." Amazon Lex can process such additional information (in these examples, update the crust type slot or change the intent from OrderPizza to OrderDrink).
ElicitSlot - Amazon Lex is expecting the value of a slot for the current intent. For example, suppose that in the response Amazon Lex sends this message: "What size pizza would you like?". A user might reply with the slot value (e.g., "medium"). The user might also provide additional information in the response (e.g., "medium thick crust pizza"). Amazon Lex can process such additional information appropriately.
Fulfilled - Conveys that the Lambda function has successfully fulfilled the intent.
ReadyForFulfillment - Conveys that the client has to fulfill the request.
Failed - Conveys that the conversation with the user failed. This can happen for various reasons, including that the user does not provide an appropriate response to prompts from the service (you can configure how many times Amazon Lex can prompt a user for specific information), or if the Lambda function fails to fulfill the intent.
slotToElicit (string) --
If the dialogState value is ElicitSlot , returns the name of the slot for which Amazon Lex is eliciting a value.
inputTranscript (string) --
The text used to process the request.
If the input was an audio stream, the inputTranscript field contains the text extracted from the audio stream. This is the text that is actually processed to recognize intents and slot values. You can use this information to determine if Amazon Lex is correctly processing the audio that you send.
audioStream (bytes or seekable file-like object) --
The prompt (or statement) to convey to the user. This is based on the bot configuration and context. For example, if Amazon Lex did not understand the user intent, it sends the clarificationPrompt configured for the bot. If the intent requires confirmation before taking the fulfillment action, it sends the confirmationPrompt . Another example: Suppose that the Lambda function successfully fulfilled the intent, and sent a message to convey to the user. Then Amazon Lex sends that message in the response.
{'sentimentResponse': {'sentimentLabel': 'string', 'sentimentScore': 'string'}}
Sends user input to Amazon Lex. Client applications can use this API to send requests to Amazon Lex at runtime. Amazon Lex then interprets the user input using the machine learning model it built for the bot.
In response, Amazon Lex returns the next message to convey to the user an optional responseCard to display. Consider the following example messages:
For a user input "I would like a pizza", Amazon Lex might return a response with a message eliciting slot data (for example, PizzaSize): "What size pizza would you like?"
After the user provides all of the pizza order information, Amazon Lex might return a response with a message to obtain user confirmation "Proceed with the pizza order?".
After the user replies to a confirmation prompt with a "yes", Amazon Lex might return a conclusion statement: "Thank you, your cheese pizza has been ordered.".
Not all Amazon Lex messages require a user response. For example, a conclusion statement does not require a response. Some messages require only a "yes" or "no" user response. In addition to the message , Amazon Lex provides additional context about the message in the response that you might use to enhance client behavior, for example, to display the appropriate client user interface. These are the slotToElicit , dialogState , intentName , and slots fields in the response. Consider the following examples:
If the message is to elicit slot data, Amazon Lex returns the following context information:
dialogState set to ElicitSlot
intentName set to the intent name in the current context
slotToElicit set to the slot name for which the message is eliciting information
slots set to a map of slots, configured for the intent, with currently known values
If the message is a confirmation prompt, the dialogState is set to ConfirmIntent and SlotToElicit is set to null.
If the message is a clarification prompt (configured for the intent) that indicates that user intent is not understood, the dialogState is set to ElicitIntent and slotToElicit is set to null.
In addition, Amazon Lex also returns your application-specific sessionAttributes . For more information, see Managing Conversation Context .
See also: AWS API Documentation
Request Syntax
client.post_text(
botName='string',
botAlias='string',
userId='string',
sessionAttributes={
'string': 'string'
},
requestAttributes={
'string': 'string'
},
inputText='string'
)
string
[REQUIRED]
The name of the Amazon Lex bot.
string
[REQUIRED]
The alias of the Amazon Lex bot.
string
[REQUIRED]
The ID of the client application user. Amazon Lex uses this to identify a user's conversation with your bot. At runtime, each request must contain the userID field.
To decide the user ID to use for your application, consider the following factors.
The userID field must not contain any personally identifiable information of the user, for example, name, personal identification numbers, or other end user personal information.
If you want a user to start a conversation on one device and continue on another device, use a user-specific identifier.
If you want the same user to be able to have two independent conversations on two different devices, choose a device-specific identifier.
A user can't have two independent conversations with two different versions of the same bot. For example, a user can't have a conversation with the PROD and BETA versions of the same bot. If you anticipate that a user will need to have conversation with two different versions, for example, while testing, include the bot alias in the user ID to separate the two conversations.
dict
Application-specific information passed between Amazon Lex and a client application.
For more information, see Setting Session Attributes .
(string) --
(string) --
dict
Request-specific information passed between Amazon Lex and a client application.
The namespace x-amz-lex: is reserved for special attributes. Don't create any request attributes with the prefix x-amz-lex: .
For more information, see Setting Request Attributes .
(string) --
(string) --
string
[REQUIRED]
The text that the user entered (Amazon Lex interprets this text).
dict
Response Syntax
{
'intentName': 'string',
'slots': {
'string': 'string'
},
'sessionAttributes': {
'string': 'string'
},
'message': 'string',
'sentimentResponse': {
'sentimentLabel': 'string',
'sentimentScore': 'string'
},
'messageFormat': 'PlainText'|'CustomPayload'|'SSML'|'Composite',
'dialogState': 'ElicitIntent'|'ConfirmIntent'|'ElicitSlot'|'Fulfilled'|'ReadyForFulfillment'|'Failed',
'slotToElicit': 'string',
'responseCard': {
'version': 'string',
'contentType': 'application/vnd.amazonaws.card.generic',
'genericAttachments': [
{
'title': 'string',
'subTitle': 'string',
'attachmentLinkUrl': 'string',
'imageUrl': 'string',
'buttons': [
{
'text': 'string',
'value': 'string'
},
]
},
]
}
}
Response Structure
(dict) --
intentName (string) --
The current user intent that Amazon Lex is aware of.
slots (dict) --
The intent slots that Amazon Lex detected from the user input in the conversation.
Amazon Lex creates a resolution list containing likely values for a slot. The value that it returns is determined by the valueSelectionStrategy selected when the slot type was created or updated. If valueSelectionStrategy is set to ORIGINAL_VALUE , the value provided by the user is returned, if the user value is similar to the slot values. If valueSelectionStrategy is set to TOP_RESOLUTION Amazon Lex returns the first value in the resolution list or, if there is no resolution list, null. If you don't specify a valueSelectionStrategy , the default is ORIGINAL_VALUE .
(string) --
(string) --
sessionAttributes (dict) --
A map of key-value pairs representing the session-specific context information.
(string) --
(string) --
message (string) --
The message to convey to the user. The message can come from the bot's configuration or from a Lambda function.
If the intent is not configured with a Lambda function, or if the Lambda function returned Delegate as the dialogAction.type its response, Amazon Lex decides on the next course of action and selects an appropriate message from the bot's configuration based on the current interaction context. For example, if Amazon Lex isn't able to understand user input, it uses a clarification prompt message.
When you create an intent you can assign messages to groups. When messages are assigned to groups Amazon Lex returns one message from each group in the response. The message field is an escaped JSON string containing the messages. For more information about the structure of the JSON string returned, see msg-prompts-formats .
If the Lambda function returns a message, Amazon Lex passes it to the client in its response.
sentimentResponse (dict) --
The sentiment expressed in and utterance.
When the bot is configured to send utterances to Amazon Comprehend for sentiment analysis, this field contains the result of the analysis.
sentimentLabel (string) --
The inferred sentiment that Amazon Comprehend has the highest confidence in.
sentimentScore (string) --
The likelihood that the sentiment was correctly inferred.
messageFormat (string) --
The format of the response message. One of the following values:
PlainText - The message contains plain UTF-8 text.
CustomPayload - The message is a custom format defined by the Lambda function.
SSML - The message contains text formatted for voice output.
Composite - The message contains an escaped JSON object containing one or more messages from the groups that messages were assigned to when the intent was created.
dialogState (string) --
Identifies the current state of the user interaction. Amazon Lex returns one of the following values as dialogState . The client can optionally use this information to customize the user interface.
ElicitIntent - Amazon Lex wants to elicit user intent. For example, a user might utter an intent ("I want to order a pizza"). If Amazon Lex cannot infer the user intent from this utterance, it will return this dialogState.
ConfirmIntent - Amazon Lex is expecting a "yes" or "no" response. For example, Amazon Lex wants user confirmation before fulfilling an intent. Instead of a simple "yes" or "no," a user might respond with additional information. For example, "yes, but make it thick crust pizza" or "no, I want to order a drink". Amazon Lex can process such additional information (in these examples, update the crust type slot value, or change intent from OrderPizza to OrderDrink).
ElicitSlot - Amazon Lex is expecting a slot value for the current intent. For example, suppose that in the response Amazon Lex sends this message: "What size pizza would you like?". A user might reply with the slot value (e.g., "medium"). The user might also provide additional information in the response (e.g., "medium thick crust pizza"). Amazon Lex can process such additional information appropriately.
Fulfilled - Conveys that the Lambda function configured for the intent has successfully fulfilled the intent.
ReadyForFulfillment - Conveys that the client has to fulfill the intent.
Failed - Conveys that the conversation with the user failed. This can happen for various reasons including that the user did not provide an appropriate response to prompts from the service (you can configure how many times Amazon Lex can prompt a user for specific information), or the Lambda function failed to fulfill the intent.
slotToElicit (string) --
If the dialogState value is ElicitSlot , returns the name of the slot for which Amazon Lex is eliciting a value.
responseCard (dict) --
Represents the options that the user has to respond to the current prompt. Response Card can come from the bot configuration (in the Amazon Lex console, choose the settings button next to a slot) or from a code hook (Lambda function).
version (string) --
The version of the response card format.
contentType (string) --
The content type of the response.
genericAttachments (list) --
An array of attachment objects representing options.
(dict) --
Represents an option rendered to the user when a prompt is shown. It could be an image, a button, a link, or text.
title (string) --
The title of the option.
subTitle (string) --
The subtitle shown below the title.
attachmentLinkUrl (string) --
The URL of an attachment to the response card.
imageUrl (string) --
The URL of an image that is displayed to the user.
buttons (list) --
The list of options to show to the user.
(dict) --
Represents an option to be shown on the client platform (Facebook, Slack, etc.)
text (string) --
Text that is visible to the user on the button.
value (string) --
The value sent to Amazon Lex when a user chooses the button. For example, consider button text "NYC." When the user chooses the button, the value sent can be "New York City."