Apple has been granted a patent (number 10,133,725) for “learning new words” with the goal of helping its iOS, macOS, watchOS, and tvOS devices learn new words for use in Messages and Siri.
In the patent filing, Apple notes that a user of a client device relies on one or more dictionaries of words for spell checking, suggesting words during typing, and other uses of known words. The tech giant says that such client dictionaries are difficult to keep updated with new words that may become popular through crowdsourced usage of words without compromising privacy.
Current servers can learn the words that users are typing by examining clear text that users have typed when utilizing the servers. For example, some prior art text message services and email services (collectively, messages) receive messages in clear text. Message servers that route messages to client devices can read the clear text and use the words obtained from the clear text of user messages to present advertising to the users.
However Apple notes that the server-learned words remain on the server and do not update an on-device dictionary to include the new words. Also, usage of clear text by servers compromises the privacy of a user. In addition, new words generated on a client device, such as words that are used within documents on the client device and are not transmitted to a server, cannot be learned by the server because the words are localized to the client device.
What’s more, if the client device utilizes an end-to-end encrypted messaging service, such as Messages, then a server can’t learn the words contained in the user message at all and a server can’t update a user client dictionary using crowdsourced data. Apple wants to change this.
Here’s Apple’s summary of the invention: “A client device can determine that a word typed on the client device is a new word that is not contained in a dictionary or asset catalog on the client device. New words can be grouped in classifications such as entertainment, health, finance, etc. A differential privacy system on the client device can comprise a privacy budget for each classification of new words. If there is privacy budget available for the classification, then one or more new terms in a classification can be sent to new term learning server, and the privacy budget for the classification reduced. The privacy budget can be periodically replenished.”
Of course, Apple files for — and is granted — lots of patents by the U.S. Patent & Trademark Office. Many are for inventions that never see the light of day.
