Today we’re announcing the launch of two new features to ML Kit: Language Identification and Smart Reply.

We have worked closely together with partners like textPlus to make sure Smart Reply is ready for prime time and they have now employed in-app response suggestions with the latest version of their app (screenshot above).

On your own programs you can attain exactly the exact same with the new Smart Reply API. The API provides hints based on the last 10 messages in a conversation, if only one message can be obtained although it still works. It is so message background isn’t kept by us in memory send it.

The identifyLanguage functions requires a piece of a text and its callback provides a BCP-47 speech code. If no speech can be recognized, ML Kit returns a code of und for undetermined. The Language Identification API may also supply a list of languages along with their confidence values.

TextPlus program providing reply suggestions with Smart Reply

ML Kit recognizes text in 110 distinct languages and typically only requires a couple of words to produce an accurate decision. It is fast as well, typically offering a response within 1 to 2 ms across iOS and Android phones.

We’re really excited to expand ML Kit to include Natural Language APIs. Give the two NLP APIs a spin now and let us know what you think! You may always reach us at our Firebase Chat Google Group.

Get started now

val languageIdentification =

After you initialize a Smart Reply instance, phone suggestReplies using a listing of current messages. The callback provides the result that includes a listing of suggestions.

For information on the best way to use the Language Identification API, have a look at the documentation.

val smartReply = FirebaseNaturalLanguage.getInstance().smartReply

Tell me …

A text string’s language is really a subtle yet useful piece of information. A good deal of programs have performance with a dependency on the language: you can imagine features like text translation, spell checking or Smart Reply. Rather than requesting a user to define the language they use, you can use our Language Identification API.

However, when we began development of this API since we realized, the suggestion model that is center is not all that’s required to provide a solution that programmers can use in their own programs. We added a version to discover issues, so that we avoid making suggestions in response or in cases of. We included language identification, to ensure we don’t provide hints for languages that the core model isn’t trained on. Support is being launched with by the Smart Reply feature first.

There is popping up in programs A new attribute to supply the consumer with an array of suggested responses, either as actions on a notification or within the app itself. This really can help a consumer or a means.

Similar to the Smart Reply API, you are able to identify the language with a function call (using Kotlin in this case ):

NLP is a category of ML that deals with generating and analyzing other kinds of natural language data, speech, and text. We are eager to start out with two APIs: one which produces reply tips in conversation applications, and also one which helps you identify text’s language.

Adding Smart Reply for your app is done using a simple function call (using Kotlin in this example):

Create reply tips based on past messages

Posted by Christiaan Prins and Max Gubin
You may notice that these features both are different. Therefore, we are excited to enlarge ML Kit with alternatives for Natural Language Processing (NLP)!

.addOnSuccessListener identifiedLanguage ->
Log.i(TAG,”Identified language: $identifiedLanguage”)

.addOnFailureListener e ->
Log.e(TAG,”Language identification error”, e)

Although as a programmer, you get it incorporated in your program and easily can simply pick up this API, it may be interesting to show a bit on how it works under the hood.

Since ML Kit grows we anticipate adding APIs and classes that allows you to offer smarter experiences for your customers. With that, please keep a look out for some fascinating ML Kit announcements at Google I/O.

For details about the best way best to use the Smart Reply API, check out the documentation.

SmartReply.suggestReplies(dialog )
.addOnSuccessListener