Scribd Logo
Upload

Welcome to Scribd!

  • 20 views

    Iphone Spelling Correction

    Uploaded by

    Amr Ibrahim Mohammed Sheta
    - The iPhone uses a hidden Markov model for spelling correction as the user types using a virtual keyboard. The hidden information is what the user intended to type, while the observed information is what they actually typed. - It calculates the probabilities of possible words based on the likelihood of hitting neighboring keys and suggests the most probable real word. - This allows it to detect and correct errors in real-time, as long as it can perform the necessary calculations faster than the user can type.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Iphone Spelling Correction

    Uploaded by

    Amr Ibrahim Mohammed Sheta
    20 views11 pages
    - The iPhone uses a hidden Markov model for spelling correction as the user types using a virtual keyboard. The hidden information is what the user intended to type, while the observed information is what they actually typed. - It calculates the probabilities of possible words based on the likelihood of hitting neighboring keys and suggests the most probable real word. - This allows it to detect and correct errors in real-time, as long as it can perform the necessary calculations faster than the user can type.

    Original Description:

    A description of iPhone Spelling Correction

    Original Title

    iPhone Spelling Correction

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    - The iPhone uses a hidden Markov model for spelling correction as the user types using a virtual keyboard. The hidden information is what the user intended to type, while the observed information is what they actually typed. - It calculates the probabilities of possible words based on the likelihood of hitting neighboring keys and suggests the most probable real word. - This allows it to detect and correct errors in real-time, as long as it can perform the necessary calculations faster than the user can type.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    20 views11 pages

    Iphone Spelling Correction

    Uploaded by

    Amr Ibrahim Mohammed Sheta
    - The iPhone uses a hidden Markov model for spelling correction as the user types using a virtual keyboard. The hidden information is what the user intended to type, while the observed information is what they actually typed. - It calculates the probabilities of possible words based on the likelihood of hitting neighboring keys and suggests the most probable real word. - This allows it to detect and correct errors in real-time, as long as it can perform the necessary calculations faster than the user can type.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 11

    Hidden 

    Markov Models and Spelling 
    Correction on the iPhone
    Dr. Jeffrey L. Popyack
    Department of Computer Science
    How the iPhone Does Spelling 
    Correction
    iPhone User Interface
    • The iPhone uses a touch‐screen 
    user interface.  Users type by 
    tapping a virtual keyboard.
    • In order to be a useful device, 
    humans need to be able to enter 
    text quickly and accurately.
    • Human Factors:  Humans make 
    mistakes.  In general, the size of 
    the iPhone’s virtual keyboard is 
    too small for human fingers to 
    enter text quickly and accurately.
    iPhone User Interface
    • Artificial Intelligence:  
    – Can the iPhone’s computational 
    power be used to do spelling 
    correction while the user is typing?
    – Can an algorithm be developed that 
    can detect and correct errors?
    – How many calculations can be made 
    between individual key‐clicks?
    – Can the necessary calculations be 
    done faster than the user can type?
    Hidden Markov Model
    A Hidden Markov Model can describe a situation where 
    events occur but cannot be accurately observed.  If the 
    relationship between the observed info and the hidden info is

    Hidden? Hidden? Hidden?

    • Observed • Observed • Observed

    known, the hidden info can be deduced with some likelihood 
    of success.  This can be done to estimate hidden values or 
    predict future hidden values.
    Hidden Markov Model
    In this case, the “hidden” info is what the 
    user meant to type, and the “observed” info 
    is what they did type.

    T H E

    • Observed:  • Observed:  • Observed: 


    T G R

    iPhone designers recognize that a certain percentage of the time, 
    the user will tap a neighboring key instead of the desired key.  
    Suppose they hit the correct key 60% of the time.
    Hidden Markov Model
    T G R
    .6 .6 .6
    R ‐ .1 T ‐ .08 E ‐ .1
    Y ‐ .08
    F ‐ .1 D ‐ .1
    F ‐ .08
    G ‐ .1 F ‐ .1 Possible Words:
    H ‐ .08 TGR – Probability = .6 * .6 * .6 = .216
    Y ‐ .1 T ‐ .1 TGE – Probability = .6 * .6 * .1 = .036
    V ‐ .08 …
    THE – Probability = .6 * .08 * .1 = .0048
    …
    The most likely word is the one with the 
    RYE – Probability = .1 * .08 * .1 = .001 
    highest probability that is in the dictionary.
    What Did the User Mean?
    • MPY

    • WOWEE
    What Did the User Mean?
    • MPY
    NOT
    • WOWEE
    What Did the User Mean?
    • MPY
    NOT
    • WOWEE
    SPADE
    Notes
    • We used the probability “.6” for the likelihood that the 
    user typed the correct letter.  In reality, this is a value 
    that would be determined through careful research.  It 
    might vary for different letters.
    • The iPhone interface suggests a word as a replacement 
    for a given word that is typed, and will automatically 
    replace the typed word with the suggested word unless 
    the user cancels the action.  A different process would 
    be appropriate if the suggested word is often wrong.
    • Certain words in the dictionary may be used much 
    more frequently than others ‐

    You might also like