The Agent class

class agent.Agent

The Agent class provides the methods to connect to DialogFlow APIs and perform basic intent queries.

language

Agent’s language code (e.g. "en-US").

Type:string

project_id

DialogFlow’s API project ID.

Type:string

session_client

SessionsClient class to manage API sessions.

Type:dialogflow.SessionsClient

get_intent_score(sentence)

Given a sentence, detects a related intent with a certain confidence score.

Parameters:sentence (string) – Input sentence to be processed Returns:A string containing the detected intent and the correspondent confidence score. Return type:string, float

Examples

>>> agent = Agent()
>>> agent.get_intent_score("It's hot")
("overheat", 0.8965118527412415)

The Client class

class client.Client

The Client class is the backends which receives the User input and performs situation identification and reaction selection through an abductive inference process (given contextual data).

To start interacting with the client, run the following command:

An example interaction is:

agent

DialogFlow agent.

Type:Agent

context

Dictionary containing context table and reactions information. For example:

{
    "reactions": [
        "I'll open the window",
        "I'll bring you a bottle of water",
        "I'll turn on the TV"
    ],
    "queries": [
        "Should I open the window?",
        "Should I bring you a bottle of water?",
        "Should I turn on the TV?"
    ],
    "default": ["I am sorry, I can't do a lot for you."],
    "r1": {
        "overheat": 0.1,
        "bug-in-room": 0.07,
        "boredom": 0
    },
    "r2": {
        "overheat": 0.02,
        "bug-in-room": 0.02,
        "boredom": 0
    },
    "r3": {
        "overheat": 0,
        "bug-in-room": 0,
        "boredom": 0.07
    }
}

Type:dict

ask_question(question)

Poses a Y/N question to the user.

Note

This function loops until the user does not give a Y/N answer.

Parameters:question (string) – The question to be displayed. Returns:Returns the answer as ‘y’ or ‘n’. Return type:string

deep_situation_analysis(sentence)

Given a sentence, breaks it and send the single words to DialogFlow. Then, it returns the intent of the word with the maximum score.

Parameters:sentence (string) – Input sentence to be processed Returns:A string containing the detected situation. Return type:string

main()

Starts the client workflow.

std_situation_analysis(sentence)

Given a sentence, simply returns DialogFlow identified intent

Parameters:sentence (string) – Input sentence to be processed Returns:A string containing the detected situation. Return type:string

Examples

>>> client = Client()
>>> client.std_situation_analysis("It's hot")
"overheat"