Padatious
Padatious is a machine-learning, neural-network based intent parser.
Padatious is a machine-learning, neural-network based intent parser. It is an alternative to the Adapt intent parser. Unlike Adapt, which uses small groups of unique words, Padatious is trained on the sentence as a whole.
Padatious has a number of key benefits over other intent parsing technologies.
- With Padatious, Intents are easy to create
- The machine learning model in Padatious requires a relatively small amount of data
- Machine learning models need to be trained. The model used by Padatious is quick and easy to train.
- Intents run independently of each other. This allows quickly installing new skills without retraining all other skill intents.
- With Padatious, you can easily extract entities and then use these in Skills. For example, "Find the nearest gas station" ->
{ "place":"gas station"}
In speech recognition and voice assistance, an intent is the task the user intends to accomplish. A user can accomplish the same task in multiple ways. The role of the intent parser is to extract from the user's speech key data elements that specify their intent in more detail. This data can then be passed to other services, such as Skills to help the user accomplish their intended task.
Example: Julie wants to know about today's weather in her current location, which is Melbourne, Australia.
hey mycroft, what's today's weather like?hey mycroft, what's the weather like in Melbourne?hey mycroft, weather
Each of these examples has very similar intent. The role of Padatious is to determine intent programmatically.
In the example above, we would extract data elements like:
- weather - we know that Julie wants to know about the weather, but she has not been specific about the type of weather, such as wind, precipitation, snowfall or the risk of fire danger from bushfires. Melbourne, Australia rarely experiences snowfall, but falls under bushfire risk every summer.
- location - Julie has stipulated her location as Melbourne, but she does not state that she means Melbourne, Australia. How do we distinguish this from Melbourne, Florida, United States?
- date - Julie has been specific about the timeframe she wants weather data for - today. But how do we know what today means in Julie's timezone. Melbourne, Australia is between 14-18 hours ahead of the United States. We don't want to give Julie yesterday's weather, particularly as Melbourne is renowned for having changeable weather.
Padatious uses a series of example sentences to train a machine learning model to identify an intent.
The examples are stored in a Skill's
vocab[lang] directory, in files ending in the file extension .intent. For example, if you were to create a tomato Skill to respond to questions about a tomato, you would create the filevocab/en-us/what.is.intentThis file would contain examples of questions asking what a tomato is.
What would you say a tomato is?What's a tomato?Describe a tomatoWhat defines a tomato
and
vocab/en-us/do.you.like.intentwith examples of questions about mycroft's opinion about tomatoes:
- Are you fond of tomatoes?
- Do you like tomatoes?
- What are your thoughts on tomatoes?
- Are you fond of
{type}tomatoes? - Do you like
{type}tomatoes? - What are your thoughts on
{type}tomatoes?
Note the {type} in above examples these are wild-cards where matching content is forwarded to the skill's intent handler.
Each file should contain at least 4 examples for good modeling.
In the above example,
{type} will match anything. While this makes the intent flexible, it will also match if we say something like Do you like eating tomatoes?. It would think the type of tomato is eating which doesn't make much sense. Instead, we can specify what type of things the {type} of tomato should be. We do this by defining the type entity file here:vocab/en-us/type.entitywhich would contain something like:
red
reddish
green
greenish
yellow
yellowish
ripe
unripe
pale
Now, we can say things like Do you like greenish red tomatoes? and it will tag type: as greenish red.
A skill using Padatious is no different than previous skills except that
self.register_intent_file() is used instead of self.register_intent(). To register a .entity file, use self.register_entity_file().For example, the Tomato Skill would be written as:
from mycroft import MycroftSkill
class TomatoSkill(MycroftSkill):
def __init__(self):
MycroftSkill.__init__(self)
def initialize(self):
self.register_intent_file('what.is.intent', self.handle_what_is)
self.register_intent_file('do.you.like.intent', self.handle_do_you_like)
def handle_what_is(self, message):
self.speak('A tomato is a big red thing')
def handle_do_you_like(self, message):
tomato_type = message.data.get('type')
if tomato_type is not None:
self.speak("Well, I'm not sure if I like " + tomato_type + " tomatoes.")
else:
self.speak('Of course I like tomatoes!')
The
register_intent_file(intent_file, handler) methods arguments are:- intent_file: the filename of above mentioned intent files with the .intent as argument.
- handler: the method/function that the examples in the intent_file should map to
The corresponding decorator is also available:
@intent_handler()In the handler method the wild card words can be fetched from the message using
def handler(self, message):
word = message.data.get('your_keyword') # if not present will return None
Sometimes you might find yourself writing a lot of variations of the same thing. For example, to write a skill that orders food, you might write the following intent:
Order some {food}.
Order some {food} from {place}.
Grab some {food}.
Grab some {food} from {place}.
Rather than writing out all combinations of possibilities, you can embed them into a single line by writing each possible option inside parentheses with | in between each part. For example, that same intent above could be written as:
(Order | Grab) some {food} (from {place} | )
Nested parentheses are supported to create even more complex combinations, such as the following:
(Look (at | for) | Find) {object}.
Which would expand to:
Look at {object}
Look for {object}
Find {object}
Let's say you are writing an Intent to call a phone number. You can make it only match specific formats of numbers by writing out possible arrangements using
# where a number would go. For example, with the following intent:Call {number}.
Call the phone number {number}.
the number.entity could be written as:
+### (###) ###-####
+## (###) ###-####
+# (###) ###-####
(###) ###-####
###-####
###-###-####
###.###.####
### ### ####
##########
Let's say you wanted to create an intent to match places:
Directions to {place}.
Navigate me to {place}.
Open maps to {place}.
Show me how to get to {place}.
How do I get to {place}?
This alone will work, but it will still get a high confidence with a phrase like "How do I get to the boss in my game?". We can try creating a
.entity file with things like:New York City
#### Georgia Street
San Francisco
The problem is, now anything that is not specifically a mix of New York City, San Francisco, or something on Georgia Street won't match. Instead, we can specify an unknown word with :0. This would would be written as:
:0 :0 City
#### :0 Street
:0 :0
Now, while this will still match quite a lot, it will match things like "Directions to Baldwin City" more than "How do I get to the boss in my game?"
NOTE: Currently, the number of :0 words is not fully taken into consideration so the above might match quite liberally, but this will change in the future.
NOTE: This section is of use if you are using Padatious on a project other than Mycroft. If you're developing Skills for Mycroft, you don't need to worry about this
program.py:
from padatious.intent_container import IntentContainer
container = IntentContainer('intent_cache')
container.load_file('hello', 'hello.intent')
container.load_file('goodbye', 'goodbye.intent')
container.train()
data = container.calc_intent('Hello there!')
print(data.name)
hello.intent:
Hi there!
Hello.
goodbye.intent:
See you!
Goodbye!
You can then run this in Python using:
python3 program.py
NOTE: This section is of use if you are using Padatious on a project other than Mycroft. If you're developing Skills for Mycroft, you don't need to worry about this
Padatious is designed to be run in Linux. Padatious requires the following native packages to be installed:
- Python development headers
pip3swig
To install these packages on a Ubuntu system, run this command:
sudo apt-get install libfann-dev python3-dev python3-pip swig
Next, install Padatious via
pip3:pip3 install padatious
Padatious also works in Python 2 if you are unable to upgrade.