Feature-based autocorrection

#This application claims the benefit of U.S. Provisional Application No. 61/714,589, filed Oct. 16, 2012, the entire content of which is hereby incorporated by reference.

#Computing devices (e.g., mobile phones, tablet computers, etc.) may provide a graphical keyboard as part of a graphical user interface for composing text using a presence-sensitive display. For instance, a presence-sensitive display of a computing device may output a graphical, or soft, keyboard that permits the user to enter data by tapping (or otherwise indicating) keys displayed at the presence-sensitive display. The graphical keyboard may enable a user of the computing device to compose and edit text included in, for example, an e-mail, a text message, a document, etc.

#Some computing devices also include an auto-correction system to detect and correct spelling and grammar errors within user-entered text. Some auto-correction systems, however, may fail to detect non-spelling or non-grammar errors within the text and/or may make erroneous automatic corrections. Some auto-correction systems may fail to detect such errors because these systems perform exact word matching for misspelled dictionary words within the text. Consequently, such auto-detection systems may not detect an error so long as the entered text includes a valid word (e.g., a word found within a dictionary employed by the system). In addition, if a user intentionally types a non-dictionary word, the auto-correction system may automatically (and contrary to the user's intent) replace the misspelled word with a dictionary word.

#FIG. 1 is a conceptual diagram illustrating an example computing device that is configured to detect errors within text and suggest corrections, in accordance with one or more aspects of the present disclosure.

#FIG. 2 is a block diagram illustrating an example computing device, in accordance with one or more aspects of the present disclosure.

#FIGS. 3A-3B are conceptual diagrams illustrating example graphical user interfaces for presenting suggested corrections to text, in accordance with one or more aspects of the present disclosure.

#A computing device (e.g., a laptop, a mobile phone, a tablet computer, a table top computer, etc.) may receive input at a presence-sensitive display that outputs a graphical keyboard. In response to receiving the input, the computing device may generate text for display at the presence-sensitive display. The text generated for display by the computing device may include intentional and unintentional spelling and grammar mistakes. In general, techniques of this disclosure may enable a computing device to detect a potential spelling error of a word within text generated responsive to user input and output suggested words that may correct a spelling error included in the user input.

#In some examples, the computing device can detect a spelling error and determine suggested words to correct the error by utilizing several systems (e.g., a spatial model, a language model, etc.). Using these systems, the computing device may determine a probability that a word within the generated text is a misspelling of at least one of the suggested words. If the probability satisfies a threshold, the computing device may output one or more of the suggested words that a user can optionally ignore, or select to fix the spelling error within the word.

#In this way the computing device may more efficiently generate text based on user input than other devices that use conventional word-matching techniques. Consequently, words suggested by the computing device upon detection of a spelling error may generally be more accurate than those suggested by other devices. As such, a user of the computing device may spend less time fixing auto-correction errors than a user of other devices, and may more quickly and easily select accurate suggested words for correction of a spelling error. By enabling more efficient text input at the graphical keyboard, the computing device may also perform fewer operations to process and generate text from such input, and thus consume less processing power than other computing devices.