Skip to content

Latest commit

 

History

History
69 lines (49 loc) · 2.67 KB

README.md

File metadata and controls

69 lines (49 loc) · 2.67 KB
Raw

How to train the Lex bot

Table of Contents

Postman request

You may find the API.postman_collection.json Postman collection in root folder of this project, which contains a trainLexBot request. To train the Lex bot send this request with a payload containing the same fields as in the payload example, modifying the values according to the original message you want to modify.

Payload example

Considering a message in the smkarticle-bucket-neutral buck with the text iam happy that was labeled as NEUTRAL, but it was determined that it should be positive. Therefore, a request was sent to the messages/train endpoint with the following payload:

{
	"messages": [
		{
			"data": {
				"id": "9671c425-002d-44e1-a1d4-705281ceecd4",
				"name": "message, 2020-12-02T15:38:04.623Z",
				"sent": "2020-12-02T15:38:44.623Z",
				"lexResult": {
					"Payload": {
						"intentName": "NEUTRAL"
					}
				},
				"message": "iam happy",
				"userId": "9671c425-002d-44e1-a1d4-705281ceecd4"
			},
			"classifiedAs": "Positive",
			"keysWords": [],
			"messages": [
				"iam happy"
			],
			"save": 1
		}
	],
	"bucketName": "smkarticle-bucket-neutral"
}

Lex Bot Training from Postman

Payload fields' description:

  • id: The user id.

  • name: The name of the message for which the sentiment should be modified. It can be found in the details of the message stored in smkarticle-bucket-neutral (or smkarticle-bucket) under the Key field.

  • sent: The moment in which the original message was sent. It can be found in the message details, in the Key field next to "message".

  • intentName: The classification that Lex applied to the message.

  • message: The text of the original message.

  • userId: The user id.

  • classifiedAs: The sentiment label to which the message should be associated. E.g. if the message was labelled "NEUTRAL" when it should have been "POSITIVE", this field should be "POSITIVE".

  • keysWords: An array of words to help the bot identify the intent.

  • messages: An array containing the text of the original message.

  • save: A boolean that determines whether the message should be saved or not.

  • bucketName: The name of the bucket where the original message is.