Apple working to improve iPhone autocorrect

The U.S. Patent and Trademark Office published three patent applications filled by Apple and confirming that the company continues to research technologies that would contribute to improved accuracy of the autocorrect feature on the iPhone.

Application No. 12/976834, titled "Combining timing and geometry information for typing correction," describes a process for using keystroke geometry and timing to better detect a user's intended input.  Any typed string could be analyzed for use in either autocorrection or autocompletion.

"For example, if the string “theere” is entered and the time between the keystrokes “ee” is less than an associated baseline by more than a threshold amount, the replacement candidate “there” may be assigned a higher score," the application read.

The system would also take into account the location of letters on the keyboard and the geometry of the typed words in order to catch mistakes.

The second autocorrect-related patent application (No. 12/976849) is entitled "Using parts-of-speech tagging and named entity recognition for spelling correction."

The proposed process involves understanding the context surrounding typed words in order to tag their parts of speech and identify named entities, such as differentiating between the company "Apple" and reference to the fruit. That process would involve parsing words, phrases and sentences and would make use of clues like capitalization, suffixes and prefixes and other "contextual features."

The third patent application is titled "Using statistical language models for contextual lookup" (No. 12/976864) describes the use of the aforementioned filling for language processing to search queries. Such process should simplify search in the Internet or documents.