Open Sentence/Speech Understanding
With Open sentence/Speech Understanding API users received function name and value that used in the function when you input text.
Resources : REST, SDK(Android™,Server side JAVA)
The utterance understanding API takes input of an uttered sentence in text format and returns a function as the most-likely sought-afer feedback.
You can ask a question using a text sentence in a natural, conversational tone.
A response includes the name of a function, which may be accompanied by characteristic words extracted from the input sentence.
" 'I'm hungry' ." --> " 'Gourmet search' " is returned.
" 'I want to eat ramen.' " --> " 'Gourmet search,' 'gourmet genre name: ramen: food name' " is returned.
" I want to call Mr. Yamada." --> "'phone,' 'phoneTo: Yamada: last name' " is returned.
" I want to go to Tameike-sanno from Yokohama by train." --> " 'Train transfer information,' 'stationFrom: Yokohama (station name), ' 'stationTo: Tameike-sanno (station name)' " is returned.
This API corresponds Japanese and English. If you want to use Japanese version, you should specify "language=ja" as request parameter. In other hands, you want to use English version, specify "language=en" as well.
Between Japanese and English version, there are different command list and slot list. English version doesn't have "types".
This API does not return a real "answer" to an input sentence.
If you want the application to display actual train transfer information in response to an input sentence (e.g., "I want to go to Tameike-sanno from Yokohama by train"),
you need to make a further inqury into a search engine that provides train transfer information by entering the words "Yokohama" and "Tameike-sanno" returned from the API .
In using the utterence understanding API, what you do is basically sending a "text" message and receiving a "judgment result" from the server.
Please refer to the sample codes for requests and responses together with this section.
This section describes both the transmission and reception procedures , using the terms actually shown on the interface.
For details, please refer to the list of parameters.
1. Data sent from the application to the ser ver
They are important while what is required as mandatory is only the "utterance text".
*APIKEY and language designation are some of other mandatory data, but you can largely let them stay fixed.
2. Data sent from the server to the application
The important data are the following 5 items, to each of which a sample is given.
"commandName" : Train transfer information
"slotName" : stationTo
"slotValue" : Tameike-sanno
"valueType" : station
"CommandID" and "commandName":
The ID and the name are specified for each function. The ID and the name are in a one-to-one relationship.
For details, please refer to the list of commands.
"slotName", "slotValue" , and "valueType" :
A box of words set for each function is called a slot.
For example, train transfer information has "stationFrom(get-on station)" and "stationTo(get-off station)" slots.
slotName means the name of a slot.
slotValue is a word actually contained in a slot.
In the above example, it means that a slot named "stationTo" has a word "Tameike-sanno" in it.
valueType means a type of a word.
In the above example, it means that the word "Tameike-sanno" belongs to a type called "station."
Each command uses particular slots. And each slot contains particular types.
For details, please refer to the list of types.
For extracting words that belong to the slots other than those specified from a stencence, please refer to "extractedWords."
All the words extracted are output with their respective types.
*Word extraction only Japanese version
"wordsValue": "Tameike-sanno station",
"wordsType": "station", "upoi", "snoun"
In the above case, the extracted words are tied to the types of the station name, the locaiton name, and the noun.
For details of types, please refer to the list of types.