System, computer program and method for improving text input in shorthand-on-keyboard interface (improvement of text input in shorthand-on-keyboard interface on keyboard)
16 claims: 6 independent, 10 dependent
- 1グラフィカル・キーボード・インタフェースを介して入力された入力信号を認識するための単語認識システムであって、 前記システムは、 一般的に用いられる単語を含むコア辞書と、 前記コア辞書に含まれない単語を含む拡張辞書と、 前記入力信号に関連付けられた単語を認識するための認識モジュールであって、前記認識モジュールは、前記入力信号に関連付けられ、かつ前記コア辞書および前記拡張辞書から選択された候補単語のN-ベスト・リストを発生する、認識モジュールと、 少なくとも各候補単語と前記入力信号との間の類似性 尺度 に基づいて、前記候補単語のN-ベスト・リストをランク付けするための事前ランク付けモジュールと、 前記入力信号に関連付けられた出力単語を前記コア辞書から出力するための選択モジュールと、 前記候補単語のランク付けされたN-ベスト・リストを提示するためのユーザ選択インタフェースと、 前記候補単語のランク付けされたN-ベスト・リストからの候補単語の選択を受信し、前記入力信号に関連付けられた前記選択された候補単語を前記コア辞書に載せることを許可するためのモジュールと を含む、システム。
- 2前記ユーザ選択インタフェースが、前記コア辞書からの候補単語および前記拡張辞書からの候補単語を、容易に区別するために異なる知覚的特徴でリスト化する、請求項1に記載のシステム。
- 3前記事前ランク付けモジュールが、前記コア辞書からの最も高いランクの単語を、前記候補単語のN-ベスト・リストにおける最も高いランクの単語として出力する、請求項1に記載のシステム。
- 4グラフィカル・キーボード・インタフェースを介して入力された入力信号を認識するための単語認識方法であって、前記方法は、コンピュータにより実行され、 前記方法は、 一般的に用いられる単語をコア辞書にストアするステップと、 前記コア辞書に含まれない単語を拡張辞書にストアするステップと、 前記入力信号に関連付けられた単語を認識するステップと、 前記入力信号に関連付けられ、かつ前記コア辞書および前記拡張辞書から選択された候補単語のN-ベスト・リストを発生するステップと、 少なくとも各候補単語と前記入力信号との間の類似性 尺度 に基づいて、前記候補単語のN-ベスト・リストをランク付けするステップと、 前記入力信号に関連付けられた出力単語を前記コア辞書から出力するステップと、 前記候補単語のランク付けされたN-ベスト・リストを提示するステップと、 前記候補単語のランク付けされたN-ベスト・リストからの候補単語の選択を受信し、前記入力信号に関連付けられた前記選択された候補単語を前記コア辞書に載せることを許可するステップと を含む、方法。
- 5ユーザ入力インタフェースを介して入力された入力信号を認識するための、コンピュータが使用可能な媒体にストアされたコンピュータ・プログラムであって、当該コンピュータに、 一般的に用いられる単語をコア辞書にストアするステップと、 前記コア辞書に含まれない単語を拡張辞書にストアするステップと、 前記入力信号に関連付けられた単語を認識するステップと、 前記入力信号に関連付けられ、かつ前記コア辞書および前記拡張辞書から選択された候補単語のN-ベスト・リストを発生するステップと、 少なくとも各候補単語と前記入力信号との間の類似性 尺度 に基づいて、前記候補単語のN-ベスト・リストをランク付けするステップと、 前記入力信号に関連付けられた出力単語を前記コア辞書から出力するステップと、 前記候補単語のランク付けされたN-ベスト・リストを提示するステップと、 前記候補単語のランク付けされたN-ベスト・リストからの候補単語の選択を受信し、前記入力信号に関連付けられた前記選択された候補単語を前記コア辞書に載せることを許可するステップと を実行させるための、コンピュータ・プログラム。
- 6グラフィカル・キーボード・インタフェースを介して入力された入力信号に係るテキストの語幹および接辞を組み合わせることを可能にするためのシステムであって、 前記システムは、 一般的に用いられる単語を含むコア辞書と、 前記コア辞書に含まれない単語を含む拡張辞書と、 前記入力信号に関連付けられた単語を認識するための認識モジュールと、 前記入力信号に関連付けられた出力単語を前記コア辞書から出力するための選択モジュールと、 前記コア辞書から選択された、前記入力信号に関連付けられた候補単語、および、前記拡張辞書から選択された、前記入力信号に関連付けられた候補単語を提示するユーザ選択インタフェースと、 候補単語の選択を受信し、前記入力信号に関連付けられた前記選択された候補単語を前記コア辞書に載せることを許可するためのモジュールと、 前記入力信号を入力接辞として認識するための連結モジュールであって、更に、候補単語を隣接候補単語として認識する、連結モジュールと、 前記入力接辞を含む辞書内の1組の単語を検索するための、複合単語出力モジュールであって、前記入力接辞を含む前記1組の単語において最も高いランクの辞書単語を出力する、複合単語出力モジュールと、 前記入力接辞を含む前記1組の単語における各辞書単語を前記候補単語および前記入力接辞を含むストリングと比較する類似性関数に従って、前記入力接辞を含む前記1組の単語をランク付けするための、ランク付けモジュールと を含む、システム。
- 7前記入力接辞が接尾辞である、請求項6に記載のシステム。
- 8前記複合単語出力モジュールが前記接尾辞および前記最も高いランクの辞書単語を結合させる、請求項7に記載のシステム。
- 9前記入力接辞が接頭辞である、請求項6に記載のシステム。
- 10前記複合単語出力モジュールが前記接頭辞および前記最も高いランクの辞書単語を結合させる、請求項9に記載のシステム。
- 11前記類似性関数が距離関数を含む、請求項9に記載のシステム。
- 12前記隣接候補単語が、前記入力接辞の前に付く候補単語または前記入力接辞の後に付く候補単語のいずれか1つを含む、請求項9に記載のシステム。
- 13前記入力信号に係るテキストが前記入力接辞として認識されない場合、前記複合単語出力モジュールは、前記入力信号に係るテキストおよび前記隣接候補単語の連結の結果として生じるストリングを生成し、前記辞書における前記ストリングの発生頻度を求め、前記ストリングの発生頻度を前記入力信号に係るテキストおよび前記隣接候補単語のそれぞれの発生頻度と比較し、 前記ストリングの発生頻度が前記入力信号に係るテキストおよび前記隣接候補単語のそれぞれの発生頻度を超えた場合、前記複合単語 出力 モジュールは、前記入力信号に係るテキストおよび前記隣接候補単語を連結単語として連結し、前記ストリングを前記連結単語によって置換する、請求項12に記載のシステム。
- 14前記入力信号に係るテキストおよび前記隣接候補単語の発生頻度に対する前記ストリングの発生頻度の比較が重み付け比較である、請求項13に記載のシステム。
- 15グラフィカル・キーボード・インタフェースを介して入力された入力信号に係るテキストの語幹および接辞を組み合わせることを可能にするための方法であって、前記方法は、コンピュータにより実行され、 前記方法は、 一般的に用いられる単語をコア辞書にストアするステップと、 前記コア辞書に含まれない単語を拡張辞書にストアするステップと、 前記入力信号に関連付けられた単語を認識するステップと、 前記入力信号に関連付けられた出力単語を前記コア辞書から出力するステップと、 前記コア辞書から選択された、前記入力信号に関連付けられた候補単語、および、前記拡張辞書から選択された、前記入力信号に関連付けられた候補単語を提示するステップと、 候補単語の選択を受信し、前記入力信号に関連付けられた前記選択された候補単語を前記コア辞書に載せることを許可するステップと、 前記入力信号を入力接辞として認識するステップと、 前記入力信号が前記入力接辞として認識された場合、候補単語を隣接候補単語として認識するステップと、 前記入力接辞を含む辞書内の1組の単語を検索するステップと、 前記入力接辞を含む前記1組の単語における各辞書単語を、前記候補単語および前記入力接辞を含むストリングと比較することによる類似性関数に従って、前記入力接辞を含む前記1組の単語をランク付けするステップと、 前記入力接辞を含む前記1組の単語において最も高いランクの辞書単語を出力するステップとを含む、方法。
- 16グラフィカル・キーボード・インタフェースを介して入力された入力信号に係るテキストの語幹および接辞を組み合わせることを可能にするための、コンピュータが使用可能な媒体にストアされたコンピュータ・プログラムであって、当該コンピュータに、 一般的に用いられる単語をコア辞書にストアするステップと、 前記コア辞書に含まれない単語を拡張辞書にストアするステップと、 前記入力信号に関連付けられた単語を認識するステップと、 前記入力信号に関連付けられた出力単語を前記コア辞書から出力するステップと、 前記コア辞書から選択された、前記入力信号に関連付けられた候補単語、および、前記拡張辞書から選択された、前記入力信号に関連付けられた候補単語を提示するステップと、 候補単語の選択を受信し、前記入力信号に関連付けられた前記選択された候補単語を前記コア辞書に載せることを許可するステップと、 前記入力信号を入力接辞として認識し、候補単語を隣接候補単語として更に認識するステップと、 前記入力接辞を含む辞書内の1組の単語を検索するステップと、 前記入力接辞を含む前記1組の単語における各辞書単語を、前記候補単語および前記入力接辞を含むストリングと比較する類似性関数に従って、前記入力接辞を含む前記1組の単語をランク付けするステップと、 前記入力接辞を含む前記1組の単語において最も高いランクの辞書単語を出力するステップとを実行させるための、コンピュータ・プログラム。
Independent claims16
53 paragraphs, as filed
0001The present invention generally relates to dictionary-based text input and text prediction systems. More specifically, the present invention is a shorthand-on-keyboard, which is an efficient way to enter words by drawing geometric patterns on a graphically represented on-screen keyboard. ) Is used for text input.
0002A short hand on a graphical keyboard (a simple method of communication) (hereinafter referred to as "short hand on keyboard") or a short hand on a keyboard as a chart (sokgraph) without a physical keyboard. , Typically represents an input method and system for efficiently inputting text using a pen for handwriting input. The shorthand-on-keyboard allows the user to enter words and commands into the computer by following letters or function keys on the graphical keyboard. Experienced users can partially or completely memorize geometric patterns of words and commands that are frequently used on keyboard layouts, and draw these patterns based on memory, for example, using a digital pen. Can be done.
0003All text input systems based on word-level recognition, such as shorthand-on-keyboards and handwriting / speech recognition and text prediction systems, rely on some form of dictionary that defines the word set recognized by these systems. .. The user's input is collated against the choices in the dictionary. Words that are not in the dictionary are usually not automatically recognized. In such cases, a special mode must be provided. For example, with a shorthand on keyboard, the user can first look up the candidate list (N-best list). If none of the choices on the candidate list are the desired words, the user determines if the pattern drawn was incorrect. If the pattern drawn is correct, the user recognizes that the desired word is not in the dictionary. The user then enters this new word in the dictionary by entering the individual letters of the word. Ideally, the dictionary contains just enough words that a particular user needs to write. Dictionaries that are too large or too small can cause problems for the user.
0004Larger dictionaries can cause some problems. This is because the number of choices other than the correct answer may increase for each user input, so that the recognition accuracy tends to be low. In any language, there tends to be a core vocabulary set that is common to everyone. Beyond this core set, vocabulary tends to be specialized for a particular individual. For example, an engineer may create an email containing highly technical terms and abbreviations for a particular area or business area. For other users, these specialized terms are irrelevant and can cause noise in the recognition process and reduce the certainty of the recognition process.
0005Smaller dictionaries are more reliable in that they are usually more likely to recognize user input correctly when the desired word is in the dictionary. Smaller dictionaries give more flexibility and tolerance to user input, allowing for rougher and more inaccurate input than the ideal form of desired input choice. Yet another advantage of small dictionaries is that the search space is small. As a result, it is possible to reduce the search waiting time for a small dictionary. This is especially important for mobile devices where processing power is severely limited.
0006However, if a small dictionary does not contain the words that the user needs, the user can experience annoyance. Uncertainty arises for the user because the user does not know whether or not the word appears in the dictionary before input. In a conventional system, it is possible that a word is not recognized either if the word is entered incorrectly or if the word is not in the dictionary. This can make it difficult for the user to determine why the word is not recognized. In general, to find out if a word is in the dictionary, the user has to try the word repeatedly. When it is certain that the word is not in the dictionary, the user adds the word to the dictionary by typing as described above through the interface provided by the recognition system. As dictionaries get smaller, users often have to add words to the dictionary.
0007There are traditional solutions to the dictionary size problem. A commonly used method is to use a large dictionary and then use high-level language rules such as word-level trigram models to eliminate very unlikely candidates. The drawback of language model techniques is generally the overhead of generating and efficiently using large language models. In addition, the language model can cause errors and accidentally remove the desired word. This is especially true if the language model is generic rather than customized to a particular user. In fact, efficient customization of language models is difficult. Moreover, language models are difficult to integrate with already highly accurate recognition techniques such as shorthand-on-keyboards.
0008One alternative traditional approach is to generate a personalized dictionary for the user by utilizing the user-generated written text, such as written email and other documents. Although this technique creates a dictionary that is more suited to a particular user, the user-generated previous document data may be too small to cover all the desired words. Moreover, in fact, it is difficult to write computer program code that can open and read all the various email and document formats that users may be using. This technique often requires the user to locate and select a previous document, which is inconvenient for the user. Also, customized dictionaries can be difficult to carry between different devices.
<p num="0009"> While these traditional solutions are appropriate for the intended purpose, they allow dictionaries with a relatively small number of irrelevant choices for the user's desired input, and most users often It is desirable to find a solution that provides easy access to almost every word a user may need, including more specialized words that are not used. In general, it is desirable to include all the words that a user may need in a very large dictionary. However, a very large dictionary means that, given the same collation threshold, more words match the pattern drawn on the keyboard, which reduces the signal-to-noise ratio in the input system. Therefore, increasing the size of the dictionary reduces flexibility and certainty for the user. Therefore, there is a need for a dictionary configuration for a shorthand-on-keyboard system that balances ease of use with flexibility and certainty.</p><p num="0010"> Another problem with traditional shorthand-on-keyboard input methods is that you have to enter text accurately at the word level, one word at a time. Some words are long. For newer users, it can be cognitively difficult to draw long words in one move with a shorthand-on-keyboard. This difficulty is especially acute in some European languages, where longer words are more complex than English. Furthermore, if a general affix can be drawn as a movement separate from the stem of a word, the user can find the input more convenient. For example, when writing the word "working" using a shorthand on keyboard, the user wants to draw a pattern of work on the graphical keyboard, then draw an ing, and combine the two into one word. I have something to think about. For this reason, there is a need for an effective system and method that automatically combines partial words on the keyboard (sokgraph) into a single word intended by the user.</p><p num="0011"> Therefore, there is a need to improve text input in shorthand-on-keyboard interfaces.</p>
<p num="0012"> The present invention discloses systems, computer programs, and related methods (collectively referred to herein as "systems" or "systems") for improving text input in a shorthand-on-keyboard interface. .. The system includes a core dictionary and an extended dictionary. The core dictionary contains words commonly used in a language. The core dictionary typically contains approximately 5,000 to 15,000 words, depending on the intended use of the system. The extended dictionary contains words that are not included in the core dictionary. The extended dictionary contains about 30,000 to 100,000 words.</p><p num="0013"> The core dictionary allows the system to focus on words commonly used as the highest ranked candidate words when identifying movements, providing more reliable recognition performance associated with smaller dictionaries. Can be done. In this system, only words from the core dictionary can be output directly. Additional candidate words are available from the extended dictionary, which allows the user to find lesser-known words on the candidate list, but only by menu selection. The system enhances word recognition accuracy without sacrificing word selection from large dictionaries. The core dictionary provides greater flexibility and tolerance for rough and inaccurate user input compared to the ideal form of desired input choice.</p><p num="0014"> The system further includes a recognition module, a pre-ranking module, and a ranking module. The recognition module generates an N-best list of candidate words that correspond to the input pattern. The pre-ranking module ranks N-best candidate words according to predetermined criteria. The ranking module adjusts the ranking of the N-best list of candidate words to place words from the core dictionary higher than words from the extended dictionary for ranked word candidates. Generate a list. Only the words listed in the core dictionary are output by this system. This system only lists candidate words in the extended dictionary in the N-best list. These words must be selected by the user for output. Once selected by the user from the N-best list, words from the extended dictionary are allowed to appear in the core dictionary.</p><p num="0015"> More specifically, in a preferred embodiment, only the words listed in the core dictionary are output by the recognition system. Words in the extended dictionary can only be included in the N-best list and must be explicitly selected by the user for output. Once selected, words in the extended dictionary are also allowed to appear in the core dictionary.</p><p num="0016"> This system reduces the overhead given to the user when the word indicated by the user is not included in the vocabulary of the core dictionary. It is not uncertain whether the word is in the dictionary or the system misrecognized the input, but the user can scan the N-best list and select the desired candidate word. ..</p><p num="0017"> The system further includes a concatenation module and a compound word module. The concatenation module allows the user to enter long word parts separately. The system automatically combines words and word parts that are partial "sokgraphs" into a single word intended by the user. The word part may be a stem such as "work" and an affix such as "ing" "pre". The compound word module combines one or more common short words that form one long word by concatenation, such as the English word "short + hand". It is more common in some European languages, such as Swedish or German, that several short words are concatenated into one compound word.</p><p num="0018"> The system allows user interaction to adjust the concatenation of words 1 and 2 and the separation of combined words. When the user clicks on a concatenated word, for example "smoke free", the user is presented with the menu option "Split into smoke and free" or an equivalent option. Alternatively, pen trace movements, such as downward movements across the word "smokefree", can be defined as split commands. For concatenable words that do not work because of low confidence, word 1 and word 2 have embedded menu options. If the user clicks on word 1, the option "move to the right" or an equivalent option is available. If the user clicks on word 2, the option "move to the left" or an equivalent option is available. Alternatively, define the movement of a pen, such as a circle across both word 1 and word 2, as a command to connect the two words as one concatenated long word.<u style="single">For example, the present invention provides the following items.</u><u style="single">(Item 1)</u><u style="single">A word recognition system for recognizing input signals input via a shorthand on keyboard interface.</u><u style="single"> A core dictionary containing commonly used words and</u><u style="single"> An extended dictionary containing words not included in the core dictionary and</u><u style="single"> A recognition module for recognizing the word associated with the input signal,</u><u style="single"> A selection module for outputting the output word associated with the input signal from the core dictionary, and</u><u style="single"> When a candidate word associated with the input signal is selected by the user, a module for allowing the candidate word to be placed in the core dictionary, and</u><u style="single">Including the system.</u><u style="single">(Item 2)</u><u style="single">The system of item 1, further comprising a user selection interface that presents candidate words associated with the input signal from at least one of the core dictionary and the extended dictionary for selection by the user.</u><u style="single">(Item 3)</u><u style="single">The system according to item 2, wherein the user selection interface lists candidate words from the core dictionary and candidate words from the extended dictionary with different perceptual features for easy distinction.</u><u style="single">(Item 4)</u><u style="single">The system according to item 1, wherein the recognition module generates an N-best list of candidate words from the core dictionary and the extended dictionary.</u><u style="single">(Item 5)</u><u style="single">The system according to item 4, further comprising a pre-ranking module for ranking the N-best list of said candidate words according to at least one criterion.</u><u style="single">(Item 6)</u><u style="single">The system according to item 5, wherein the pre-ranking module outputs the highest ranked word from the core dictionary as the highest ranked word in the N-best list of the candidate words.</u><u style="single">(Item 7)</u><u style="single">A word recognition method for recognizing input signals input via a shorthand on keyboard interface.</u><u style="single"> Steps to store commonly used words in the core dictionary,</u><u style="single"> Steps to store words not included in the core dictionary in the extended dictionary,</u><u style="single"> The step of recognizing the word associated with the input signal and</u><u style="single"> A step of outputting the output word associated with the input signal from the core dictionary, and</u><u style="single"> When a candidate word associated with the input signal is selected by the user, a step of allowing the candidate word to be placed in the core dictionary, and</u><u style="single">Including methods.</u><u style="single">(Item 8)</u><u style="single">7. The method of item 7, further comprising the step of presenting candidate words associated with the input signal from at least one of the core dictionary and the extended dictionary for selection by the user.</u><u style="single">(Item 9)</u><u style="single">8. The method of item 8, further comprising listing candidate words from the core dictionary and candidate words from the extended dictionary with different perceptual features for easy distinction.</u><u style="single">(Item 10)</u><u style="single">A computer program product that has program code stored on a computer-usable medium for recognizing input signals input through a user input interface.</u><u style="single"> A core dictionary containing commonly used words and</u><u style="single"> An extended dictionary containing words not included in the core dictionary and</u><u style="single"> The program code for recognizing the word associated with the input signal and</u><u style="single"> A program code for outputting the output word associated with the input signal from the core dictionary, and</u><u style="single"> When the candidate word associated with the input signal is selected by the user, the program code for permitting the candidate word to be placed in the core dictionary and the program code.</u><u style="single">Computer program products, including.</u><u style="single">(Item 11)</u><u style="single">The system according to item 1, wherein the stem and affix of the text related to the input signal can be combined.</u><u style="single"> A concatenation module for recognizing the input signal as an input affix, and a concatenation module for recognizing a candidate word as an adjacent candidate word.</u><u style="single"> A compound word output module for searching a set of words in a dictionary containing the input affix, which outputs the highest ranked dictionary word in the set of words including the input affix. Module and</u><u style="single"> To rank the set of words containing the input affix according to a similarity function that compares each dictionary word in the set of words containing the input affix with the candidate word and the string containing the input affix. , Ranking module and</u><u style="single">Including the system.</u><u style="single">(Item 12)</u><u style="single">The system according to item 11, wherein the input affix is a suffix.</u><u style="single">(Item 13)</u><u style="single">The system of item 12, wherein the compound word output module combines the suffix and the highest ranked dictionary word.</u><u style="single">(Item 14)</u><u style="single">The system of item 11, wherein the input affix is a prefix.</u><u style="single">(Item 15)</u><u style="single">The system of item 14, wherein the compound word output module combines the prefix with the highest ranked dictionary word.</u><u style="single">(Item 16)</u><u style="single">14. The system of item 14, wherein the similarity function includes a distance function.</u><u style="single">(Item 17)</u><u style="single">The system of item 14, wherein the adjacent candidate word comprises either one of the candidate words that precedes the input affix or the candidate word that follows the input affix.</u><u style="single">(Item 18)</u><u style="single">When the text related to the input signal is not recognized as the input verb, the compound word module generates a string generated as a result of concatenation of the text related to the input signal and the adjacent candidate word, and the generation of the string in the dictionary. The frequency is calculated, the frequency of occurrence of the string is compared with the frequency of occurrence of the text related to the input signal and the frequency of occurrence of the adjacent candidate word, and the frequency of occurrence of the string is the frequency of occurrence of the input text and the adjacent candidate word. The system of item 17, wherein the compound word module concatenates the input text and the adjacent candidate words as concatenated words and replaces the string with the concatenated word when exceeded.</u><u style="single">(Item 19)</u><u style="single">The system according to item 18, wherein the comparison of the occurrence frequency of the string with respect to the occurrence frequency of the text related to the input signal and the adjacent candidate word is a weighted comparison.</u><u style="single">(Item 20)</u><u style="single">The method according to item 7, wherein the stem and affix of the text related to the input signal can be combined.</u><u style="single"> The step of recognizing the input signal as an input affix,</u><u style="single"> When the input signal is recognized as the input affix, the step of recognizing the candidate word as an adjacent candidate word and</u><u style="single"> A step to search for a set of words in the dictionary containing the input affix,</u><u style="single"> A step of ranking the set of words containing the input affix according to a similarity function by comparing each dictionary word in the set of words containing the input affix with the candidate word and the string containing the input affix. When,</u><u style="single"> A step of outputting the highest ranked dictionary word in the set of words including the input affix, and</u><u style="single">A method that further comprises.</u><u style="single">(Item 21)</u><u style="single">19. The method of claim 19, further comprising combining the suffix and the highest ranked dictionary word.</u><u style="single">(Item 22)</u><u style="single">The computer program product according to item 10, wherein the stem and affix of the text related to the input signal can be combined.</u><u style="single"> Program code for recognizing the input signal as an input affix and further recognizing a candidate word as an adjacent candidate word.</u><u style="single"> Program code for searching a set of words in the dictionary containing the input affix,</u><u style="single"> To rank the set of words containing the input affix according to a similarity function by comparing each dictionary word in the set of words containing the input affix with the candidate word and the string containing the input affix. Program code and</u><u style="single"> Program code for outputting the highest ranked dictionary word in the set of words including the input affix, and</u><u style="single">Computer program products, including.</u></p><p num="0019"> The various features of the invention and the methods for achieving them will be described in more detail with reference to the following description, claims and drawings. In the drawings, reference numbers are repeatedly used as appropriate to show the correspondence between the referenced elements.</p>
0020<figref num="1">It is the schematic of the exemplary operating environment in which the word pattern recognition system of this invention can be used.</figref><figref num="2">It is a block diagram of the high-level architecture of the word pattern recognition system of FIG.</figref><figref num="3">It is a process flow chart which shows the operation method of the word pattern recognition system of FIG. 1 and FIG. 2 when ranking candidate words according to the position in a core dictionary or an extended dictionary.</figref><figref num="4">It is a figure which shows the N-best list generated by the word pattern recognition system of FIG. 1 and FIG. 2 which displays a word from a core dictionary and a word from an extended dictionary in different ways.</figref><figref num="5">It is a diagram showing the N-best list generated by the word pattern recognition system of FIGS. 1 and 2 that groups and ranks words from the core dictionary higher than words from the extended dictionary.</figref><figref num="6">The behavior of the word pattern recognition system in Figures 1 and 2 when recognizing word candidates as suffixes or prefixes and combining the recognized prefixes or prefixes with the recognized words in the original language in an appropriate way. A process flow chart showing the method.</figref><figref num="7">It is a process flow chart which shows the operation method of the word pattern recognition system of FIG. 1 and FIG. 2 when combining a plurality of words into a compound word.</figref><figref num="8A">It is a figure which shows the menu of the word pattern recognition system of FIG. 1 and FIG. 2, and this menu enables a user to divide a compound word into a stem and a suffix.</figref><figref num="8B">It is a figure which shows the menu of the word pattern recognition system of FIG. 1 and FIG. 2, and this menu enables a user to divide a compound word into a stem and a suffix.</figref><figref num="8C">It is a figure which shows the menu of the word pattern recognition system of FIG. 1 and FIG. 2, and this menu enables a user to divide a compound word into a stem and a suffix.</figref><figref num="9">It is a figure which shows the movement of a pen performed by a user on the complex word presented by the word pattern recognition system of FIG. 1 and FIG. 2, and the complex word is divided into a stem and a suffix by the movement of the pen.</figref><figref num="10A">It is a figure which shows the menu of the word pattern recognition system of FIG. 1 and FIG. 2, and by applying this menu to a stem, a user can combine a stem and a suffix into a complex word.</figref><figref num="10B">It is a figure which shows the menu of the word pattern recognition system of FIG. 1 and FIG. 2, and by applying this menu to a stem, a user can combine a stem and a suffix into a complex word.</figref><figref num="10C">It is a figure which shows the menu of the word pattern recognition system of FIG. 1 and FIG. 2, and by applying this menu to a stem, a user can combine a stem and a suffix into a complex word.</figref><figref num="11">It is a figure which shows the menu of the word pattern recognition system of FIG. 1 and FIG. 2, and by applying this menu at the time of suffix, a user can combine a stem and a suffix into a complex word.</figref><figref num="12">It is a figure which shows the movement of a pen performed by a user on the stem and the suffix presented by the word pattern recognition system of FIG. 1 and FIG. 2, and the stem and the suffix are combined into a complex word by this movement of the pen.</figref>
0021The following definitions and descriptions provide background information about the technical field of the invention and are intended to facilitate its understanding without limiting the scope of the invention.
0022Dictionary: A set of elements that define recognizable elements that can be collated against user input in a recognition system.
0023PDA: Personal Digital Assistant: Pocket-sized personal computer. PDA usually stores phone numbers, appointments, and to-do lists. Some PDAs have a small keyboard, while others have only a special pen used to input and output on a virtual keyboard.
0024sokgraph: Shorthand on keyboard as a graph. Pattern representation of words on a virtual keyboard.
0025Virtual Keyboard: A computer-simulated keyboard with interactive touch screen capabilities that can be used to replace or supplement the keyboard with keystrokes. Virtual keys are usually entered continuously by a pen for handwriting input. It is also called a graphical keyboard, an on-screen keyboard, or a pen keyboard for handwriting input.
0026Figure 1 may use systems, computer programs, and related methods to improve text input in a shorthand-on-keyboard interface (word pattern recognition system 10 or "system 10") in accordance with the present invention. An exemplary overall environment that can be shown. System 10 includes software program code or computer programs that are typically embedded in or installed on the computer. The computer on which System 10 is installed can be a mobile device such as a PDA 15 or a mobile phone 20. System 10 can also be installed on devices such as tablet computers 25, touch screen monitors 30, electronic whiteboards 35, and digital pens 40.
0027System 10 can be installed on any device that uses a virtual keyboard or a similar interface for input, such as auxiliary device 45. System 10 can be stored on a suitable storage medium such as a diskette, CD, hard drive, or similar device.
0028System 10 identifies a word from the shape and position of the pen movement that the user forms on the graphical keyboard. The system 10 transmits the identified word to a software receiver such as an application or operating system.
0029Figure 2 shows the high-level hierarchy of System 10. System 10 includes dictionary 205. Dictionary 205 includes core dictionary 210 and extended dictionary 215. The core dictionary 210 contains words commonly used in a source language. The core dictionary typically contains about 5,000 to 15,000 words, depending on the intended use of System 10. The extended dictionary 215 contains words that are not included in the core dictionary 215. Extended dictionary 215 contains approximately 30,000 to 100,000 words.
0030System 10 further includes a recognition module 220, a pre-ranking module 225, and a selection / ranking module 230. The recognition module 220 generates an N-best list of candidate words corresponding to the input pattern 235. The pre-ranking module 225 ranks the N-best candidate words according to predetermined criteria. The ranking module 230 adjusts the ranking of the N-best list of candidate words to rank the words obtained from the core dictionary 210 higher than the words obtained from the extended dictionary 215. Generates list 240 of word candidates. As explained earlier, the words obtained from the extended dictionary are not output, only the words from the core dictionary are output.
0031System 10 further includes a concatenation module 245 and a compound word module 250. Concatenation module 245 concatenates words selected from the list 240 of ranked word candidates. For example, "ing" is concatenated with "code" to form "coding". The compound word module 250 combines words selected from the list of ranked word candidates 240 into larger words. Output word 255 is a word selected from the list of ranked word candidates 240 and processed by concatenation module 245 and compound word module 250 as needed. Only the words listed in the core dictionary 210 are given by system 10 as output words 255. System 10 lists the candidate words in the extended dictionary 215 only in the N-best list. These words must be selected by the user to be the output word 255. Once selected by the user, system 10 allows words from extended dictionary 215 to be placed in core dictionary 210.
0032The system 10 adapts the recognition of input pattern 235 by the recognition module 220 to the user's vocabulary while maintaining the maximum signal-to-noise ratio in the recognition system. System 10 allows the core dictionary 210 and the extended dictionary 215 to participate in the recognition process of recognition module 220. However, only the words in the core dictionary 210 go directly into the output of the recognition module 220. These words are the default output. Words in the extended dictionary 215 that match the input pattern 235 are listed in the "N-best" list and simply wait for the user to select. When the user selects one of these candidate words from the N-best list and replaces it with the default output, the selected word is allowed to appear in the core dictionary 210. After a word is allowed to appear in the core dictionary 210, if the allowed word matches input pattern 235, this word can go directly into the output of the recognition module.
0033FIG. 3 shows System 10's method 300 in generating an N-best list of candidates that match input pattern 235. The user gestures a word on the shorthand-on-keyboard interface (step 305). The recognition module 220 generates an N-best list of word candidates (step 310). The pre-ranking module 225 has a confidence value or<u style="single">Similarity</u>Rank the N-best list of word candidates from the core dictionary 210 and the extended dictionary 215 according to criteria such as scale (step 315).
0034The ranking module 230 determines whether the highest ranked word in the N-best list of candidate words was obtained from the core dictionary 210 (decision step 320). In the case of yes, the ranking module 230 outputs the N-best list of ranked word candidates as a list of ranked word candidates 240 (step 325). If the highest ranked candidate in the N-best list of candidate words is not in the core dictionary 210, the ranking module 230 searches the N-best list of candidate words and obtains it from the core dictionary 210. Find the position of the highest ranked word candidate (step 330).
0035If no word candidate obtained from the core dictionary 210 is found in the N-best list of candidate words (judgment step 335), the ranking module 230 ranks the N-best list of ranked word candidates. Output as list 240 of attached word candidates. Otherwise, the ranking module 230 moves the found word suggestions to the highest ranked position in the N-best list of word suggestions (step 335). The ranking module outputs the N-best list of ranked word candidates as a list of ranked word candidates 240 (step 340).
0036To allow the user to select a candidate word that is not the highest rank, the user interface component displays the next best candidate list (N-best list), from which the user enters input pattern 235. You can see alternative candidate words that match closely. In one embodiment, the position of a candidate word on this list is determined by the rank associated with that candidate word, regardless of whether the candidate word is in the core dictionary 210 or the extended dictionary 215. .. However, the highest ranked word must always be in the core dictionary, unless none of the words in the core dictionary match the user's input. In another embodiment, candidate words are grouped by dictionary source. That is, the candidate words from the core dictionary 210 are grouped, and the candidate words from the extended dictionary 215 are grouped.
0037Candidate word sources can optionally be indicated by emphasizing different perceptual features associated with the candidate word to facilitate recognition of the source of the candidate word, eg, from a core or extended dictionary. Illustrative perceptual features include, for example, colors, background shading, bold fonts, italic pairs of fonts, and the like. If the user does not choose any word, System 10 outputs the highest ranked word in the N-best list of candidate words from the core dictionary. If the user does not select a word, System 10 outputs the highest ranked word in the N-best list of candidate words from the core dictionary.
0038Words obtained from the extended dictionary 215 can be accessed from the N-best list of candidate words. This significantly increases the error tolerance of System 10 when generating the highest ranked candidates. This is because the highest ranked candidate displayed by the system comes from the smaller core dictionary 210. In the rare situation where the desired word is not found in the core dictionary 210, the user activates the N-best list to select the desired candidate.
0039FIG. 4 shows the N-best list 400 of exemplary candidate words generated by the ranking module 230. Candidate words from the core dictionary 210 include candidate words 1, 405, candidate words 2, 410, and candidate words 3, 415, which are collectively referred to as core candidate words 420. Candidate words from the extended dictionary 215 include candidate words 4, 425, candidate 5, 430, candidate 6, 435, candidate 7, 440, and candidate 8, 445, which are collectively referred to as extended candidate word 450. The core candidate word 420 and the extended candidate word 450 are displayed with different emphasis.
0040In this example, the core candidate word 420 is shown in bold and the extended candidate word 450 is shown in italics. Any form of emphasis can be used to distinguish between the core candidate word 420 and the extended candidate word 450, for example, text color, color background, shading, etc. Candidate words in the N-best list 400 of exemplary candidate words are positioned according to the rank given by the recognition module 220. However, the top word candidate position 455 is reserved for words obtained from the core dictionary 210. This is not the case if none of the words from the core dictionary match the user's input. In this case, the word from the extended dictionary can be placed at the word candidate position 455 at the top.
0041FIG. 5 illustrates an embodiment, the exemplary N-best list 500 containing candidate words ranked according to the source and according to the ranking criteria given by the recognition module 200. Similar to FIG. 4, the core candidate word 420 and the extended candidate word 450 are displayed with different highlights. In this example, the core candidate word 420 is shown in bold and the extended candidate word 450 is shown in italics.
0042The system 10 significantly reduces the overhead inflicted on the user when the word indicated by the user is not included in the vocabulary of the core dictionary 210. It is not uncertain whether the word is contained in the core dictionary 210 or the system misrecognized the input, but the user should scan the N-best list to select the desired candidate word. Can be done.
0043For those familiar with the latest technology, it will be clear that splitting words into separate dictionaries is one practice that is also a simple conceptual model. Alternatively, the dictionary 205 can be conceptualized as a core dictionary layer and an extended dictionary layer and ranked by frequency or priori probability. When words from the extended dictionary layer are selected from the N-best candidate interface, the frequency or prior probabilities of the selected words are adjusted relative to thresholds or other criteria to make the selected words belong to the core layer. It has the effect of adjusting so as to.
0044In addition, the system 10 allows the user to enter long word parts separately. System 10 automatically combines partial "sokgraphs" into a single word intended by the user. The word part may be a stem such as "work" and an affix such as "ing", or two or more common short words that are concatenated like the English word "short + hand". May form long words. It is more common in some European languages, such as Swedish or German, that several short words are concatenated into one compound word.
0045Concatenation is based on the individually recognized parts of the concatenated word. In the stem and affix example, the user first shows the input pattern 235 in action for the word that represents the stem, and then shows the affix input pattern 235 in action. For example, in the word "coding", the user first writes "code" in action and then "ing" in action. For input tracing on the keyboard, the recognition module 220 finds the best matches and strings these matches.
0046<maths num="1"><img id="000002" he="7" wi="25" file="JP5738245B2_D0001.tif" img-format="tif" img-content="drawing" /></maths>
0047Output to N-best list with. Here, the rank i of the string means the reliability of the recognition module 220 in the selected string matching input pattern 235. The string with rank i = 1 is the top-level choice of recognition module 220. The recognition module 220 stores the latest N-best list in a temporary buffer. A buffered N-best list of regular words (stems), S<sub>0</sub>Is shown.
0048In one embodiment, the suffixes are stored in a list called concatenable suffixes. The geometric pattern sokgraphs on the graphical keyboard is represented in the same way as the common word sokgraph. For example, with the suffix "ing", the sokgraph is a continuous trace that starts with the i key, moves to the n key, and ends with the g key. The system recognizes the input pattern 235 for the "ing" sokgraph in the same way as any other sokgraph. However, the suffix "ing" is stored in the list of concatenable suffixes. Alternatively, both the suffix and the regular word can be stored in the same dictionary, but the identifier distinguishes the suffix from the regular word. In one embodiment, concatenable suffixes are stored in a reference table, where each suffix, such as "ing", is associated with a set of pointers to the input in the dictionary that ends with that suffix.
0049Figure 6 shows System 100's method 600 for combining connectable suffixes with stem words. The user performs an action on a word on the shorthand-on-keyboard interface (step 605). Concatenation module 245 gets the highest ranked word for the output N-best list 240 of word candidates (step 610). The concatenation module 245 determines whether or not the acquired word is a concatenable suffix, for example, by comparing the acquired word with the concatenable suffix list (determination step 615). If the retrieved word is not a concatenable suffix, concatenation module 245 does nothing (step 620).
0050If the retrieved word is a concatenable suffix, concatenation module 245 finds concatenation candidates ending in the identified concatenable suffix (step 625). The concatenation module 245 removes the concatenable suffix from each concatenation candidate (step 630). Words ending with the current suffix (eg "ing") are S<sub>1 (i)</sub>(For example, coding or working), and the rest without the suffix is S<sub>2 (i)</sub>(For example, "code" or "work").
0051The concatenation module 245 calculates the string edit distance between the concatenated candidate with the suffix removed and the concatenable suffix (specifically, an Morgan edit error using the Wagner-Fisher algorithm) (step 635). Then the rest S<sub>2 (i)</sub>The buffered N-top option S in the best list<sub>0 (i)</sub>Collate against. S<sub>0</sub>Contains the entire word, not a fragment of the word (eg S)<sub>0(1)</sub>= code), collation is not accurate. System 10 uses edit distance (the least number of edit actions selected from inserting, deleting, or replacing a single character) to match two strings and S<sub>0(1)</sub>Closest to S<sub>2 (i)</sub>Find the string in (i = 1, N) and call it S<sub>2min</sub>It is expressed as. The concatenation module 245 classifies concatenation candidates according to the associated edit distance (step 640). The concatenation module 245 returns the concatenation candidates with the minimum edit distance (step 645).
0052In an alternative embodiment, word frequency or previous probabilities or high-level language rules are used to rank concatenation candidates that share the same editing distance.
0053S<sub>1 (i)</sub>S in<sub>2min</sub>Returns the word corresponding to as a selected concatenation candidate. For example, "code" for "cod" (the part without the "coding" suffix ) is closer to "code" for "work" (the part without the "working" suffix) in terms of editing distance . In one embodiment, it is also possible to set the threshold as the minimum allowable edit distance mismatch.
0054In another embodiment, the suffix is not linked to all words ending in the suffix. Instead, if it recognizes a suffix, System 10 scans dictionary 205, finds words that end with the recognized suffix, removes the last part from the found word, and collates the removed rest with the previous word. , Select the closest match for the concatenation as described above. The difference between these two embodiments lies in the trade-off between computational time and memory space. Scanning a dictionary means that a separate pointer list is not needed, thus reducing the storage demand for dictionaries in the medium that the software code is accessing. On the other hand, scanning a dictionary takes longer than finding the position of a word than a system containing a dictionary indexed by a separate pointer list.
0055System 10 treats prefixes and stems in the same way as stems and suffixes. Concatenation module 245 first recognizes prefix-based words from the output of list 240 of ranked word candidates, either from a separate prefix list or a common dictionary with prefix identifiers. The concatenation module 245 then recognizes the word after the prefix. Concatenation module 245 matches all words containing that prefix, removes the prefix of the matching word, and returns the closest match for concatenation.
0056Concatenating two short words into one long word is not definitive. For example, in Swedish, "smoke free" and "smoke free" are both allowed, but their meanings are opposite ("smoking is not allowed" vs. "smoking is allowed"). The compound word module 250 handles the concatenation of two words using a statistical and interactive method. To support this method, System 10 puts in dictionary 205 the frequency of all words (based on the total number of occurrences of each word in the document data of the text) and the frequency of all bigrams (two side by side). Stores statistical information, including (based on the total number of words).
0057FIG. 7 shows System 10's method 700 for combining words into complex words. Method 700 examines consecutive word pairs (word 1, word 2) (step 705). The compound word module 250 determines whether or not a combination of consecutive words (word 1 + word 2 = word 3) is found in the dictionary 205 (decision step 710). If the combined word or word 3 is not found, the compound word module 250 does nothing (step 715). If a match (word 3 = word 1 + word 2) is found, the compound word module 250 compares the frequency of word 3 with the bigram (word 1, word 2) (step 720). If the frequency of word 3 is greater than the frequency of the bigram (word 1, word 2) compared to a given threshold, or the ratio of the frequency of word 3 to the frequency of the bigram (word 1, word 2) is given. If it is greater than the threshold of (judgment step 725), the compound word module replaces word 1 and word 2 with word 3 (step 730). In all other cases, no action is taken (step 715). Alternatively, the comparison of the frequency of word 3 and the frequency of bigrams (word 1, word 2) is a weighted comparison.
0058System 10 provides a user interface that allows user interaction to coordinate coupling and isolation. 8A, 8B, and 8C show that the combined word is separated into two separate words or parts of the word. The exemplary screen 805 displays the exemplary concatenated word "coding" 810 to the user. The user selects the displayed concatenated word "coding", for example by clicking on the word "coding" 810 (Fig. 8A). Selection of the word "coding" 810 displays menu option 815. This includes, for example, the selectable instructions "split into" code "and" ing "" or equivalent options (Figure 8B). When the user selects the instruction shown in menu option 815, system 10 splits the displayed concatenated word "coding" 810 into stem "code" 820 and suffix "ing" 825 (Figure 8C).
0059Figure 9 shows an alternative example of pen trace movement 905 used to split the concatenated word "coding" 810. Screen 805 displays the concatenated word "coding" 810 to the user. The user performs a pen trace movement 905 on the concatenated word "coding" 810. System 10 divides the displayed concatenated word "coding" 810 into stem "code" 820 and suffix "ing" 825, as shown in FIG. 8C.
0060For connectable words that do not work due to low reliability, word 1 and word 2 have menu options embedded, as shown in Figures 10A to 10C. For example, screen 805 displays word 1 "code" 1005 and word 2 "ing" 1010 to the user, as shown in FIG. 10A. Selecting the word 1 "code" 1005 displays the option menu 1015, which contains the selectable instruction "move to the right" or an equivalent option (Figure 10B). When the user selects the command "move to the right" shown in the option menu 1015, system 10 concatenates word 1 "code" 1005 and word 2 "ing" 1010, and the concatenated word "coding" 1020. (Fig. 10C).
0061FIG. 11 shows an exemplary option menu 1105 that is displayed when the user selects word 2 ing 1010. When the user selects the command "move left" shown in option menu 1105, system 10 concatenates word 1 "code" 1005 and word 2 "ing" 1010 and concatenates as shown in Figure 10C. Form the word "coding" 1020.
0062FIG. 12 shows an alternative example of pen trace movement 1205 used to concatenate word 1 code 1005 and word 2 ing 1010. Pen trace movement 1205 includes, for example, a circle across word 1 "code" 1005 and word 2 "ing" 1010. System 10 recognizes the command represented by the pen trace movement 1205, concatenates word 1 "code" 1205 and word 2 "ing" 1010, and concatenates the concatenated word "coding" as shown in Figure 10C. Form 1020.
17 sheets
Sheet 1 Sheet 2 Sheet 3 Sheet 4 Sheet 5 Sheet 6 Sheet 7 Sheet 8 Sheet 9 Sheet 10 Sheet 11 Sheet 12 Sheet 13 Sheet 14 Sheet 15 Sheet 16 Sheet 17
Every citation, both ways
| Document | Relation | Office |
|---|---|---|
| JP04264668A | Cites | Japan |
| JP06131328A | Cites | Japan |
| JP2000200267A | Cites | Japan |
| JP2001034394A | Cites | Japan |
| JP2003256415A | Cites | Japan |
| JP2005078134A | Cites | Japan |
| US6018708A | Cites | United States of America |
17 members in 4 offices
Members17
| Document | Office | Kind | |
|---|---|---|---|
| US2007094024A1 | United States of America | A1 | |
| WO2007045597A1 | World Intellectual Property Organization (WIPO) | A1 | |
| CN101292214A | China | A | |
| JP2009512923A | Japan | A | |
| CN101292214B | China | B | |
| US2011071834A1 | United States of America | A1 | |
| US8311796B2 | United States of America | B2 | |
| JP2012256353A | Japan | A | |
| JP2012256354A | Japan | A | |
| US2013006639A1 | United States of America | A1 | |
| US2013234947A1 | United States of America | A1 | |
| US8543384B2 | United States of America | B2 | |
| JP5400200B2 | Japan | B2 | |
| US8712755B2 | United States of America | B2 | |
| US2014278374A1 | United States of America | A1 | |
| JP5738245B2This record | Japan | B2 | |
| US9256580B2 | United States of America | B2 |
25 legal events, as the office reported them to INPADOC
Over the term
Point at a mark for the eventEvents
| Event | Code | |
|---|---|---|
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Certificate of patent or registration of utility modelJAPANESE INTERMEDIATE CODE: R150R150 | R150 | |
| First payment of annual fees (during grant procedure)JAPANESE INTERMEDIATE CODE: A61A61 | A61 | |
| Written decision to grant a patent or to grant a registration (utility model)JAPANESE INTERMEDIATE CODE: A01A01 | A01 | |
| Decision of grant or rejection writtenTRDD | TRDD | |
| Written submission of copy of amendment under article 19 pctJAPANESE INTERMEDIATE CODE: A524A524 | A524 | |
| Written permission of extension of timeJAPANESE INTERMEDIATE CODE: A602A602 | A602 | |
| Written request for extension of timeJAPANESE INTERMEDIATE CODE: A601A601 | A601 | |
| Notification of reasons for refusalJAPANESE INTERMEDIATE CODE: A131A131 | A131 | |
| Request for written amendment filedJAPANESE INTERMEDIATE CODE: A523A521 | A521 | |
| Written permission of extension of timeJAPANESE INTERMEDIATE CODE: A602A602 | A602 | |
| Written request for extension of timeJAPANESE INTERMEDIATE CODE: A601A601 | A601 | |
| Written permission of extension of timeJAPANESE INTERMEDIATE CODE: A602A602 | A602 | |
| Written request for extension of timeJAPANESE INTERMEDIATE CODE: A601A601 | A601 | |
| Notification of reasons for refusalJAPANESE INTERMEDIATE CODE: A131A131 | A131 | |
| Report on retrievalJAPANESE INTERMEDIATE CODE: A971007A977 | A977 | |
| Notification of change in applicantJAPANESE INTERMEDIATE CODE: A711A711 | A711 |
Numbers
- Publication
- 5738245
- Application
- 178643
Titles2
- Japanese
- ショートハンド・オン・キーボード・インタフェースにおいてテキスト入力を改善するためのシステム、コンピュータ・プログラムおよび方法(キーボード上のショートハンド・オン・キーボード・インタフェースにおけるテキスト入力の改良)
- English
- Systems, computer programs and methods for improving text input in the shorthand on keyboard interface (improvements in text input in the shorthand on keyboard interface on the keyboard)
Classification
- CPC, 4
- G06F40/237
- G06F3/0237
- G06F3/04883
- G06F40/10
- IPC, 5
- G06F3 0488
- G06F40 00
- G06F40 237
- G06K9 62
- G06K9 68
