There are two types of annotations: natural language annotations
and implementation annotations.
Natural language annotations, as the name suggests, are the annotations related to
Almond's natural language modules and
it will be translated to different languages based on users' profile.
It's denoted by #_[<key>=<value>].
Implementation annotations are used for describing certain implementation characteristics,
denoted by #[<key>=<value>] (without the underscore used by natural language annotations).
The name annotation specifies the name of the device shown in user’s device list.
By default it will be the same as the name of the device shown in Thingpedia (specified in the metadata).
If you use config module from @org.thingpedia.config.form, the configuration parameters can be referred
using the syntax $param or ${param}.
See Authentication & Discovery for more details.
The description annotation specifies the description of the device shown in user’s device list.
By default it will be the same as the description of the device shown in Thingpedia (specified in the metadata).
If you use config module from @org.thingpedia.config.form, the configuration parameters can be referred
using the syntax $param or ${param}.
See Authentication & Discovery for more details.
The version annotation specifies the version of the device in Thingpedia.
This will be automatically generated based on the submission history of the device.
The version annotation specifies the version of the device in Thingpedia.
This will be automatically generated based on the submission history of the device.
This annotation is only for Almond integrators. It marks the device as a builtin system device, which will not be shown in Thingpedia.
The doc annotation provides a short description of the function,
as a reference for the developers and reviewers.
(deprecated)
The canonical annotation provides the canonical form of the function name.
It is used for training in a previous version of Almond.
Before a program is executed, Almond will confirm with the user
to make sure it understands the command correctly when needed.
The confirmation sentence comes from the confirmation annotation of the function.
Use noun phrases for queries and verb phrases in the imperative form for actions.
Note that the confirmation sentence should include all the required input parameter,
using the syntax $param or ${param}.
(deprecated)
The confirmation_remote annotation is used to confirm commands that refer to other user's
devices (remote_commands). The owner of the device can be referred by $__person.
Measure(ms) (a time interval)The poll-interval annotation specifies how often a query will be fired when it is monitored.
It can only be used for monitorable queries.
It takes a time interval, such as 5min, 1h.
If the query supports push notification, set the polling interval to 0ms.
The formatted annotation defines how the results will be presented to the user in Almond.
It can only be used for queries.
If omitted, Almond will simply list the value of all the output parameters one by one
in the order they are declared in the function signature.
See Natural Language Support for Devices in Thingpedia for detailed instructions.
The url annotation provides the URL of the API endpoint or RSS feed for declarative Thingpedia devices.
The canonical annotation provides the canonical form of the parameter name.
It defaults to the parameter name by replacing _ with space.
It helps to generate more natural synthetic sentences for training.
The prompt annotation provides the slot filling question when
the value of an input parameter is missing in the command.
If omitted, Almond will ask “What's the value of ?”
The json_key annotation can only be used for generic rest devices.
It specifies the corresponding field name from the returned JSON for the parameter.
If the parameter name is the same with the field name, this annotation can be omitted.
The string_values annotation specifies the dataset of example values for the parameter.
This can only be used for parameters of type String or Entity (hashtags, user names, etc).
See Available String Datasets for a list of available datasets.
You should upload your own dataset in order to get the best natural language support,
and the name of the dataset should be <device-id>:<param-name>.
The utterances annotation provides different ways of expressing the corresponding command.
It is used to generate the training data for the device, and the first utterance in the list
will also be used in Thingpedia Cheatsheet.
See Natural Language Support for Devices in Thingpedia for detailed instructions.