US20060048055A1 - Fault-tolerant romanized input method for non-roman characters - Google Patents

Fault-tolerant romanized input method for non-roman characters Download PDF

Info

Publication number
US20060048055A1
US20060048055A1 US10/928,131 US92813104A US2006048055A1 US 20060048055 A1 US20060048055 A1 US 20060048055A1 US 92813104 A US92813104 A US 92813104A US 2006048055 A1 US2006048055 A1 US 2006048055A1
Authority
US
United States
Prior art keywords
input
component
pinyin
language
conversion
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Abandoned
Application number
US10/928,131
Inventor
Jun Wu
Liren Chen
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Google LLC
Original Assignee
Google LLC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Google LLC filed Critical Google LLC
Priority to US10/928,131 priority Critical patent/US20060048055A1/en
Assigned to GOOGLE INC. reassignment GOOGLE INC. ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: CHEN, LIREN, WU, JUN
Publication of US20060048055A1 publication Critical patent/US20060048055A1/en
Assigned to GOOGLE LLC reassignment GOOGLE LLC CHANGE OF NAME (SEE DOCUMENT FOR DETAILS). Assignors: GOOGLE INC.
Application status is Abandoned legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/20Handling natural language data
    • G06F17/27Automatic analysis, e.g. parsing
    • G06F17/273Orthographic correction, e.g. spelling checkers, vowelisation

Abstract

Fault-tolerant systems and methods to process and correct input spelling errors for non-Roman based languages such as Chinese, Japanese, and Korean (CJK) are disclosed. The method may be applied to a Chinese input method using pinyin. For example, the method may generally include receiving a pinyin input representing characters in Chinese, the input having at least one original pinyin, identifying potentially incorrect pinyins in the input, expanding each potentially incorrect pinyin to at least one additional alternative pinyin, each pair of potentially incorrect and corresponding alternative pinyin having a proximity measurement, converting each pinyin in the input and each alternative pinyin to Chinese characters, computing likelihoods of possible conversions of the pinyin input to Chinese characters, each possible Chinese conversion being a combination of the converted original and/or alternative pinyins of the input, the probabilities being based on the proximity measurement and optionally on a context of the possible Chinese conversion, and determining a most likely Chinese conversion from the possible conversions.

Description

    BACKGROUND OF THE INVENTION
  • 1. Field of the Invention
  • The present invention relates generally to processing non-Roman based languages. More specifically, fault-tolerant systems and methods to process and correct input spelling errors for non-Roman based languages such as Chinese, Japanese, and Korean (CJK) are disclosed.
  • 2. Description of Related Art
  • Spell correction generally includes detecting erroneous words and determining appropriate replacements for the erroneous words. Most spelling errors in alphabetical, i.e., Roman-based, languages such as English are either out of vocabulary words, e.g., “thna” rather than “than,” or valid words improperly used in its context, e.g., “stranger then” rather than “stranger than.” Spell checkers that detect and correct out of vocabulary spelling errors in Roman-based languages are well known.
  • Users of non-Roman based languages such as Chinese, Japanese, and Korean (CJK) often utilize Roman-based (alphabetical) input methods. For example, many Chinese language users use pinyin (phonetic spelling) to input Chinese characters. However, Chinese language users may not know the correct pronunciations (pinyins) of some Chinese characters due to, for example, their dialect and/or accent, and therefore may enter incorrect pinyin inputs.
  • The conventional pinyin input system typically converts a pinyin input and provides a list of candidate Chinese character sets from which the user may select the intended set of Chinese characters. However, the user's intended character set may not be included in the candidate list as most pinyin input methods have a low or no fault tolerance.
  • In addition, non-Roman based languages such as Chinese, Japanese, and Korean (CJK) languages generally have no invalid characters encoded in any computer character set, e.g., UTF-8 character set, such that most spelling errors are valid characters improperly used in context rather than out of vocabulary spelling errors. In Chinese, the correct use of words can generally only be determined in context. Thus an effective spell checker for a non-Roman based language should make use of contextual information to determine which characters and/or words in context are not suitable.
  • Spell correction for non-Roman languages such as CJK languages is also complex and challenging in that there are no standard dictionaries in such languages because the definition of CJK words are not clean. For example, some may regard “Beijing city” in Chinese as one word while others may regard them as two words. In contrast, the English dictionary/wordlist lookup is a key feature in English spell correction and thus English spell correction methods cannot be easily adapted for use in CJK languages. Furthermore, the Chinese language has a high concentration of homographs and homophones as well as invisible (or hidden) word boundaries that create ambiguities that also make efficient and effective Chinese spell correction complex and difficult to implement. As is evident with such differences between Chinese and English, many efficient techniques available for English spell correction are not suitable for Chinese spell correction.
  • Thus what is needed is a computer system and method for effective, efficient and accurate processing and correcting of spelling errors for non-Roman based languages such as Chinese, Japanese and Korean languages.
  • SUMMARY OF THE INVENTION
  • Fault-tolerant systems and methods to process and correct input spelling errors for non-Roman based languages such as Chinese, Japanese, and Korean (CJK) are disclosed. In particular, the systems and methods use the hidden Markov model and a proximity matrix of pinyins with confusing pronunciations. In the fault-tolerant pinyin input method, the proximity between a pair of confusingly similar pinyins may be a positive number if they are pronounced similarly. Otherwise, the value is zero. In one embodiment, the proximity may be a probability or likelihood with a value between 0 and 1. The systems and methods are particularly applicable to web-based search engines and downloadable applications at client sites, e.g., implemented in a toolbar or deskbar, but are applicable to various other applications. It should be appreciated that the present invention can be implemented in numerous ways, including as a process, an apparatus, a system, a device, a method, or a computer readable medium such as a computer readable storage medium or a computer network wherein program instructions are sent over optical or electronic communication lines. The term computer generally refers to any device with computing power such as personal digital assistants (PDAs), cellular telephones, and network switches. Several inventive embodiments of the present invention are described below.
  • The fault-tolerant input method generally includes receiving an input in a format representing characters in a language, the input having at least one original component, identifying potentially incorrect input components, expanding each potentially incorrect component to at least one additional alternative component, each pair of potentially incorrect component and corresponding alternative component having a proximity measurement, converting each input component and each alternative component to the language, computing probabilities of possible conversions of the input to characters in the language, each possible conversion being a combination of the converted original and/or alternative components of the input, the probabilities being based on the proximity measurement and optionally on a context of the possible conversion, and determining a most likely conversion from the possible conversions.
  • As an example, the method may be applied to a Chinese input method using pinyin. For example, the method may include receiving a pinyin input having at least one original pinyin representing Chinese characters, identifying potentially incorrect pinyins, expanding each potentially incorrect pinyin to at least one additional alternative pinyin, each pair of potentially incorrect and corresponding alternative pinyin having a proximity measurement, converting the pinyins in the input and the alternative pinyins to Chinese characters, computing probabilities of possible conversions of the pinyin input to Chinese characters, each possible Chinese conversion being a combination of the converted original and/or alternative pinyins of the input, the probabilities being based on the proximity measurement and optionally on a context of the possible Chinese conversion, and determining a most likely Chinese conversion from the possible conversions.
  • The identifying potentially incorrect input components may be based on pairs of confusingly similar components which in turn may be based on confusing pairs of vowels and consonants. Each potentially incorrect component may be associated with at least one confusingly similar alternative such that the expanding includes expanding each potentially incorrect component to every additional alternative component associated therewith. The converting may be performed on each component individually and/or on multiple components jointly. The determining the most likely conversion may include determining multiple most likely conversions. For example, the most likely conversions may include at least one possible original conversion converted from only the original components of the input. The most likely conversions may also include a possible modified conversion that includes at least one converted alternative component, for example, if the probability of the possible modified conversion exceeds the probability of the possible original conversion by a predetermined threshold.
  • The language may be a non-Roman based language such a Chinese, Japanese, Korean (CJK) or Thai, etc. In the case that the language is Chinese and more typically simplified Chinese, the format may be pinyin. The method can be utilized in various applications such as in a search engine such that the input is a user search query. Various other systems may similarly employ such fault-tolerant input methods such as language input systems.
  • A fault-tolerant input system generally includes a receiver adapted to receive an input having at least one original component in a format representing characters in a language, an identifier adapted to identify potentially incorrect components of the received input, an expander adapted to expand each potentially incorrect component of the received input to at least one additional alternative component in the format, each pair of the potentially incorrect component of the input and the corresponding additional alternative component having a proximity measurement, a converter adapted to convert each component of the received input and each additional alternative component in the format to the language, a computing module adapted to compute probabilities of possible conversions of the received input to characters in the language, each possible conversion being a combination of the converted original and/or alternative components of the received input, the probabilities being based at least in part on the proximity measurement, and an output module adapted to determine a most likely conversion to characters in the language from the possible conversions of the input.
  • A computer program product for use in conjunction with a computer system, the computer program product generally includes a computer readable storage medium on which are stored instructions executable on a computer processor, the instructions including receiving an input in a format representing characters in a language, the input having at least one original component, identifying potentially incorrect input components, expanding each potentially incorrect input component to at least one additional alternative component, each pair of potentially incorrect component and the corresponding alternative component having a proximity measurement, converting each input component and each alternative component in the format to the language, computing probabilities of possible conversions of the input to characters in the language, each possible conversion being a combination of the converted original and/or alternative components of the input, the probabilities being based at least in part on the proximity measurement, and determining a most likely conversion to characters in the language from the possible conversions of the input.
  • A method for generating proximity measurements may generally include collecting user inputs in a format representing characters in a language and rates of user selection of corresponding candidates decoded from the user inputs into the language, identifying user inputs having a low rate of candidate selection as non-selected inputs, determining a percentage of each non-selected input entered without a user selection of any of the decoded candidates being made, assigning a modified input close in confusion proximity to each corresponding non-selected input as a corresponding intended input, the modified input having a higher rate of candidate selection than the corresponding non-selected input, determining the proximity measurement for each pair of modified input and non-selected input based on the number of times the corresponding non-selected input is entered as input and the number of times the corresponding non-selected input is entered as input without a user selection of any of the decoded candidates being made. The confusion proximity may be based on a list of pairs of confusingly similar elements in the format and the modified input differs from the corresponding non-selected input by one or more pairs of the confusingly similar elements.
  • An application implementing the system and method may be implemented on a server site such as on a search engine or may be implemented, e.g., downloaded, on a client site such as a user's computer to provide spell corrections for text inputting into a document or to interface with a remote server such as a search engine.
  • These and other features and advantages of the present invention will be presented in more detail in the following detailed description and the accompanying figures which illustrate by way of example principles of the invention.
  • BRIEF DESCRIPTION OF THE DRAWINGS
  • The present invention will be readily understood by the following detailed description in conjunction with the accompanying drawings, wherein like reference numerals designate like structural elements.
  • FIG. 1 is a table listing an exemplary set of common confusing pronunciations pairs of Chinese vowels and consonants.
  • FIG. 2 is a proximity matrix of confusing pinyins containing the probabilities of various intended pinyins given a potentially incorrect (confusing) input pinyin.
  • FIG. 3 is a flowchart illustrating a process for automatically converting a confusing input pinyin to a set of candidates that may include one or more candidates corresponding to a spelling-corrected pinyin input.
  • FIG. 4 is a table listing confusing pinyins generated from an exemplary pinyin input “zhong guo zhen fu” based on the list of common confusing pronunciations pairs of Chinese vowels and consonants of FIG. 1.
  • FIG. 5 is a table listing conversions to Chinese characters or words generated from the list of confusing pinyins in FIG. 4 for the exemplary pinyin input “zhong guo zhen fu.”
  • FIG. 6 is a lattice diagram illustrating the confusing pinyins, some of their conversions to Chinese characters and words, and the transitions between each pair of the Chinese character/word conversions for the exemplary pinyin input “zhong guo zhen fu” of FIGS. 4 and 5.
  • FIG. 7 is two illustrative user interfaces each presenting a candidate list of three most likely candidate conversions from the exemplary pinyin input “zhong guo zhen fu” of FIGS. 4-6, including one candidate conversion corresponding to a spelling-corrected pinyin input.
  • FIG. 8 is a graph illustrating character conversion accuracy rate relative to confusing pinyin input error rate with and without fault-tolerant input processing.
  • FIG. 9 is a graph illustrating query conversion accuracy rate relative to confusing pinyin input error rate with and without fault-tolerant input processing.
  • FIG. 10 is a flowchart illustrating a process for training a fault-tolerant input processor.
  • DESCRIPTION OF SPECIFIC EMBODIMENTS
  • Fault-tolerant systems and methods to process and correct input spelling errors for non-Roman based languages such as Chinese, Japanese, and Korean (CJK) are disclosed. The fault-tolerant input systems and methods described herein generally relate to processing, detecting, and correcting spelling errors by employing probabilities that may be derived from user input entries and associated user selections such as query logs. It is noted that for purposes of clarity only, the examples presented herein are generally presented in terms of processing, detecting and correcting Chinese pinyin inputs. However, the systems and methods for spelling error detection and correction may be similarly applicable for other non-Roman based languages such as Japanese, Korean, Thai, etc. The following description is presented to enable any person skilled in the art to make and use the invention. Descriptions of specific embodiments and applications are provided only as examples and various modifications will be readily apparent to those skilled in the art. The general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the invention. Thus, the present invention is to be accorded the widest scope encompassing numerous alternatives, modifications and equivalents consistent with the principles and features disclosed herein. For purpose of clarity, details relating to technical material that is known in the technical fields related to the invention have not been described in detail so as not to unnecessarily obscure the present invention.
  • Users of non-Roman based languages such as Chinese, Japanese, and Korean (CJK) often utilize Roman-based (alphabetical) input methods. For example, many Chinese language users use pinyin (phonetic spelling) for inputting Chinese characters such as a search query for a search engine. Pinyin is a phonetic input method used mainly for inputting simplified Chinese character. As referred to herein, pinyin generally refers to phonetic representation of Chinese characters, with or without representation of the tones associated with the Chinese characters. In particular, “pinyin” as used herein refers to all phonetic notations for Chinese, simplified or traditional, include zhuyin fuhao (Bopomofo), i.e., “The Notation of Annotated Sounds.”
  • Some Chinese language users may not know the correct pronunciations (pinyins) of some Chinese characters due to, for example, their dialect and/or accent, and therefore may enter incorrect pinyin inputs. However, despite an incorrect pinyin input, the fault-tolerant pinyin input processing system converts the incorrect pinyin input and aims to provide a list of candidate Chinese character sets that includes the user's intended Chinese character set.
  • Most often, pinyin spelling errors attributable to incorrect pronunciations are based on a set of confusing pronunciation pairs of vowels and consonants. In other words, a large portion of pinyin spelling errors attributable to incorrect pronunciations may result from the user interchanging one for the other of a confusing pair of vowels or consonants. The table of FIG. 1 lists an exemplary set of common confusing pairs of pronunciations of Chinese vowels and consonants. Pronunciations of the confusing consonants and/or vowels listed in FIG. 1 can be similar and confusing to many Chinese language users such that pinyin inputs that contain one or more of the confusing consonants and/or vowels may be misspelled. It is to be understood that additional and/or alternative confusing consonants and/or vowels may exist and may be taken into account by the fault-tolerant input system and method as described herein and that FIG. 1 is merely an exemplary set that may be employed in the systems and methods described herein. As shown, common confusing consonant pairs may include retroflexion and non-retroflexion, such as sh/s, ch/c, and the like, as well as various other confusing consonant pairs such as h/f, l/n, and the like. Common confusing vowels may include front and back nasal sound such as an/ang, en/eng, in/ing, and the like.
  • To identify and correct pinyin input errors, a probability of an intended pinyin given an input pinyin may be utilized to facilitate in determining whether the input pinyin is likely to be incorrect and if so, the most likely corrected or intended pinyin and/or the most likely pinyin-to-character/word conversions based on the corrected or intended pinyin. A collection of probabilities of an intended pinyin P′ given an input pinyin P, i.e., p(P′|P), may be provided in a proximity matrix, also referred to herein as fuzzy pinyin matrix. FIG. 2 is an example of a normalized proximity or fuzzy pinyin matrix of confusing pinyin pairs. A confusing pinyin, as referred to herein, generally refers to a pinyin containing at least one of the identified common confusing consonant or vowel pairs such as those listed in FIG. 1. In other words, a confusing pinyin is one that may be mispronounced and thus misspelled according to the set of common confusing pairs of pronunciations such as that shown in FIG. 1. All other pinyins are non-confusing pinyins, i.e., pinyins that do not contain any of the identified common confusing consonant or vowel pairs such as those listed in FIG. 1.
  • Referring again to FIG. 2, each cell of the proximity matrix indicates how likely when a user intended pinyin P′ when the user entered pinyin P. The matrix contains the probabilities of various possible intended pinyins P′ given a confusing input pinyin P. The empty cells represent a value of 0, i.e., no confusion or ambiguity between the corresponding pinyin pair. To illustrate, given an input pinyin P “lin,” the user's intended pinyin P′ may be “lin” (the original input pinyin, i.e., P′=P), “ling” (in-ing confusing pair from FIG. 1), “nin” (l-n confusing pair), or “ning” (both l-n and in-ing confusing pairs). Each of the possible intended pinyins P′ corresponds to an estimated or otherwise computed probability that the user intended pinyin P′ given the input pinyin P.
  • The proximity matrix may be built based on some knowledge or assumptions of Chinese pronunciations and optionally on statistics specifying the frequency of the mistakes that Chinese language users make. Such knowledge and/or statistics may be obtained from, for example, search query logs. For purposes of reducing computational complexity only, once the fuzzy pinyin matrix is established, misspellings can be assumed to be context independent although such an assumption may not be true.
  • The matrix of FIG. 2 is a small subset of all possible confusing pinyin pairs and only illustrates some of the likelihoods or probabilities that may be computed or otherwise assigned to the corresponding confusing pinyin pairs and different likelihoods or probabilities may be similarly computed or otherwise assigned. In particular, assuming that the table of FIG. 1 lists all confusing pairs of Chinese vowels and consonants, roughly half of the approximately 408 Chinese pinyins can be confused with another confusing Chinese pinyin. The estimated probability of an intended pinyin P′ being entered as pinyin P may be used to determine if the input pinyin is likely to be misspelled and if so, the most likely correct or intended pinyin or candidate characters. Contextual probabilities may also be used in determining a set of likely candidate Chinese characters to increase the accuracy of such determinations. The estimated probability can be used to determine the probability of the corresponding Chinese characters. Thus the more likely a given pinyin (original input or alternate) is the intended pinyin, the higher the probability of the corresponding Chinese characters. Merely as example, if each of the intended pinyins “san” and “shan” has a likelihood of approximately 1% and 3%, respectively, of being mispronounced and misspelled as “shang,” the Chinese characters corresponding to the pinyin “san” would be assigned a weight that is one-third that for the Chinese characters corresponding to “shan.” Thus the perplexity stays nearly the same as the that of not considering the confusing pinyins. Perplexity generally relates to the average number of possible characters given the context. The greater the perplexity, the more uncertain the character can be.
  • FIG. 3 is a flowchart illustrating a process 20 for automatically converting a pinyin input that may include one or more confusing pinyins to candidate Chinese character sets that may include one or more Chinese character sets corresponding to a spelling-corrected pinyin input. At block 22, a user pinyin input is received. At block 24, each confusing pinyin, if any, in the pinyin input is identified and expanded to include the other possible intended pinyins. At block 26, the most likely conversion candidates are determined. In performing blocks 24 and 26, a fuzzy pinyin proximity matrix, such as one shown in FIG. 2, may be employed to identify possible intended pinyins for a given confusing pinyin and their corresponding probabilities. In addition, in determining the most likely conversion candidates at block 26, contextual probabilities may also be employed. At block 28, N (e.g., 3) most likely conversion candidates are presented to the user.
  • Various suitable mechanisms for converting each fuzzy pinyin to corresponding Chinese characters and/or words such as performed in block 26 may be implemented. For example, various decoders are suitable for converting pinyin to Hanzi (Chinese characters). In one embodiment, a Viterbi decoder using hidden Markov models may be implemented. The training for the hidden Markov models may be achieved, for example, by collecting empirical counts or by computing an expectation and performing an iterative maximization process. The Viterbi algorithm is a useful and efficient algorithm to decode the source input according to the output observations of a Markov communication channel. The Viterbi algorithm has been successfully implemented in various applications for natural language processing, such as speech recognition, optical character recognition, machine translation, speech tagging, parsing and spell checking. However, it is to be understood that instead of the Markov assumption, various other suitable assumptions may be made in implementing the decoding algorithm. In addition, the Viterbi algorithm is merely one suitable decoding algorithm that may be implemented and various other suitable decoding algorithms such as a finite state machine, a Bayesian network, a decision plane algorithm (a high dimension Viterbi algorithm) or a Bahl-Cocke-Jelinek-Raviv (BCJR) algorithm (a two pass forward/backward Viterbi algorithm) may be implemented.
  • FIGS. 4-6 illustrate an example of the process 20 for automatically converting an exemplary confusing pinyin input “zhong guo zhen fu” to candidate Chinese character sets that may include one or more Chinese character sets corresponding to a spelling-corrected pinyin input. Specifically, FIG. 4 is a table expanding each pinyin in the confusing pinyin input “zhong guo zhen fu” to its corresponding confusing or fuzzy pinyins (i.e., the original pinyin and any other confusing pinyins) based on the list of common confusing pronunciations pairs of Chinese vowels and consonants. For example, a lookup of the proximity matrix such as that shown in FIG. 2 may be performed in order to expand each pinyin in the confusing pinyin input to its corresponding confusing or fuzzy pinyins. As shown, each of “zhong,” “zhen,” and fu” is expanded to multiple confusing pinyins. In contrast, “guo” corresponds only to itself as it is generally not confused with any other pinyin, i.e., no ambiguity.
  • FIG. 5 is a table listing illustrating some of the possible conversions to Chinese characters or words generated from the list of expanded fuzzy or confusing pinyins of FIG. 4 for the exemplary pinyin input “zhong guo zhen fu.” As shown, in the absence of context, each Chinese pinyin can map to multiple Chinese characters. For example, the pinyin “fu” can be mapped to amplitude, home, luck, husband, etc. while the pinyin “hu” maps to family name, lake, tiger, etc. In addition, where two or more pinyins can form a Chinese word, the multiple pinyins can also be converted to the Chinese word. For example, the pinyins “zhong guo” may be converted to Chinese or China.
  • FIG. 6 is a lattice diagram illustrating the confusing/fuzzy pinyins, some of their conversions to Chinese characters or words, and the transitions between each pair of the Chinese character/word conversions for the exemplary pinyin input “zhong guo zhen fu” of FIGS. 4 and 5. As shown, each fuzzy pinyin node is associated with a probability p(P′|P) of the fuzzy pinyin being the intended pinyin P′ given the input pinyin P. For example, the probabilities p(zong|zhong) and p(zhong|zhong) of the fuzzy pinyins “zong” and “zhong” being the intended pinyin given the input pinyin “zhong” is 0.15 and 0.85, respectively. Note that because the input pinyin “guo” corresponds only to itself, its assigned probability is 1.0.
  • FIG. 6 also illustrates some of the conversions of the fuzzy pinyins to Chinese characters or words as well as some of the transitions between each possible pair of the Chinese character/word conversions. Merely for purposes of clarity, not all of the conversions of the fuzzy pinyins to Chinese characters or words are shown and not all of the transitions of the conversions are shown. A probability for each link or transition may be assigned based on context, for example. Thus the probability or likelihood of each possible path, i.e., from the conversion of the first to the last pinyin, may be determined. In one embodiment, the probability score for a given path may be determined by multiplying the probability of each node in the path as well as the probability of each link in the path.
  • FIG. 7 shows two illustrative user interfaces each presenting a candidate list of N (e.g., 3) likely conversion candidates from the exemplary pinyin input “zhong guo zhen fu” of FIGS. 4-6. Note that various other suitable user interfaces may be utilized. The likely N conversion candidates presented to the user may be selected from both candidates generated from the original input and candidates, if any, generated using the fault-tolerant methods. A decision as to whether to present the spelling-corrected conversion candidate may be based on, for example, whether the probability score of the spelling-corrected candidate exceeds that of the most and/or the second most likely original input candidate by a minimum predetermined threshold. As shown, the two most likely candidate conversions 30, 32 are converted from the original pinyin input “zhong guo zhen fu” while the third candidate conversion 34 is converted from a spelling-corrected pinyin input of “zhong guo zheng fu.” In particular, the candidate conversion 30 corresponds to the China/Chinese—oscillation—amplitude path while the candidate conversion 30 corresponds to the China/Chinese—vibration—rich path in the lattice shown in FIG. 6. The spelling-corrected candidate conversion 34 corresponds to the Chinese-government path in the lattice shown in FIG. 6.
  • The spelling-corrected or fuzzy conversion candidate 34 may be presented differently from the original pinyin conversion candidates 30, 32 so as to indicate to and alert the user that the fuzzy conversion 34 corresponds to a spelling-corrected pinyin input rather than the original pinyin input. Merely as examples, the corrected candidate conversion 34 may be presented in a different color and/or a different font (font size, underlined, bold, and/or italicized, etc.).
  • FIG. 8 is a graph illustrating character conversion accuracy rate relative to a confusing pinyin input error rate with (upper curve) and without (lower curve) fault-tolerant input processing. The confusing pinyin input error rate (x-axis) refers to the percentage of confusing pinyins that are incorrectly entered. The graph of FIG. 8 may be generated from, for example, a set of user inputs such as user queries. As shown, the conversion accuracy with 0% confusing pinyin input error rate can be as high as approximately 96%. However, if any of the pinyins in the character input is incorrect, i.e., confused with another similarly pronounced pinyin, the character conversion accuracy decreases. The lower curve illustrates a relatively significant decrease in character conversion accuracy without the use of the fault-tolerant (fuzzy) pinyin input processing as the confusing pinyin input error rate increases. For example, the character conversion accuracy decreases to approximately 62% at a 100% confusing pinyin input error rate. Note that even if all the confusing pinyins in an input are incorrect, the input may contain non-confusing pinyins such that the conversion accuracy is not 0% even at a 100% confusing pinyin input error rate. The upper curve illustrates a much lower decrease in character conversion accuracy with the use of fault-tolerant (fuzzy) pinyin input processing even as the confusing pinyin input error rate increases. For example, the character conversion accuracy remains relatively unchanged even at 100% confusing pinyin input error rate.
  • FIG. 9 is a graph illustrating query conversion accuracy rate relative to a confusing pinyin input error rate with (upper curve) and without (lower curve) fault-tolerant input processing. If one word or character in a given query is incorrectly converted, the entire query is considered as incorrect. For example, a query containing 4 characters of which 1 is non-confusing and 3 are confusing and one of the 3 confusing pinyin is incorrect would have a 33% confusing pinyin input error rate. As such, the conversion accuracy at the query level is lower than that at the character level as shown and described above with reference to FIG. 8. The confusing pinyin input error rate (x-axis) refers to the percentage of confusing pinyins in the query input that are incorrectly entered. As shown, the conversion accuracy with 0% confusing pinyin input error rate for a query can be as high as approximately 85%. However, if any of the pinyins in the query input is incorrect, i.e., confused with another similarly pronounced pinyin, the query conversion accuracy decreases. The lower curve illustrates a relatively significant decrease in query conversion accuracy without the use of the fault-tolerant (fuzzy) pinyin input processing as the confusing pinyin input error rate increases. For example, the query conversion accuracy decreases to approximately 25% at a 100% confusing pinyin input error rate. The upper curve illustrates a much lower decrease in query conversion accuracy with the use of fault-tolerant (fuzzy) pinyin input processing even as the confusing pinyin input error rate increases. For example, the query conversion accuracy decreases only slightly to approximately 80% even at 100% confusing pinyin input error rate.
  • FIG. 10 is a flowchart illustrating a process 40 for training the fault-tolerant input system and method and more specifically, for generating the proximity matrix such as that shown in FIG. 2. In particular, the process of generating the proximity matrix may utilize data collected from user pinyin inputs and associated user selections in order to generate fuzzy pinyin probabilities. By using a set of inputs to train the proximity matrix, the likelihood of a spelling error in the pinyin input can be determined and processed to enhance the efficiency and effectiveness of the fault-tolerant input system.
  • The process for training the fault-tolerant input system and method and more specifically, for generating the proximity matrix may utilize data collected from user pinyin inputs and associated user selections in order to generate fuzzy pinyin probabilities. For example, in the context of a search engine, the user queries and the converted queries selected by the users are tracked and processed. For example, at block 42, data on user pinyin inputs and the rates of selection by users of each corresponding conversion/decoding candidate is collected. By using a set of actual user inputs to train the proximity matrix, the likelihood of a spelling error in the pinyin input can be determined and processed to enhance the efficiency and effectiveness of the fault-tolerant input system.
  • When a pinyin input containing at least one misspelled pinyin is entered, the user typically does not select any of the conventionally converted candidates, i.e., converted without fault tolerance, as generally none of the candidates would correspond to the user's intended input. In contrast, when the user correctly enters a pinyin input, one of the converted Chinese candidates is likely to be selected. Thus at block 44, the non-selected pinyin inputs that have low candidate selection rates are identified as those that may contain a misspelled pinyin P, i.e., a pinyin that is likely to be different from the intended pinyin P′. In addition, at block 46, the percentage or rate of pinyin input entered by users without a selection from the conversion candidates being made is determined as the proximity measurement (probability) p(P′|P).
  • The process 40 also determines all possible correct (intended) pinyin inputs for the non-selected incorrect pinyin inputs. In particular, at block 48, for each non-selected pinyin input, the intended pinyin input is determined as the pinyin input with relatively high (or highest) user selection rates that are close in confusion proximity (i.e., close in editing distance as determined in terms of the pairs of confusing pinyin elements such as those shown in FIG. 1) to the corresponding non-selected pinyin input. In other words, the confusion proximity is determined based on the list of pairs of confusingly similar pinyin elements. Thus the misspelled pinyin P (contained in the original pinyin input) and the intended pinyin P′ (contained in the intended pinyin input) differ from each other by one or more pairs of the confusingly similar pinyin elements.
  • At block 50, the proximity measurement is associated with each pair of confusing pinyins P and P′. For example, the non-selected pinyin input and the intended pinyin input may be compared to determine the incorrect pinyin P and the intended pinyin P′. The proximity measurement associated with the confusing pair of pinyins P and P′ thus relates to the likelihood that the users entered pinyin input P and intended pinyin P′, i.e., the frequency that an input pinyin P should be substituted by the pinyin P′. Specifically, the proximity measurement p(P′|P) (as determine in block 46) of a given confusing pinyin pair P and P′ is determined as n(P→P′)/n(P) where n(P) is the number of times that pinyin input P is entered by users and n(P→P′) is the number of times that the pinyin input should be substituted by its corresponding fuzzy pinyin P′, i.e., the number of times that the users did not make a selection from the list of candidate conversions.
  • As is evident, the order of the blocks illustrated for process 40 is merely illustrative and may be modified. Further, it is noted that the proximity matrix may remain constant once generated or may be modified periodically over time based on new user pinyin input and selection data. One alternative process for generating the proximity matrix, such as when there is an insufficient amount of data to generate meaningful probabilities, is to set the proximity (probability) between each pair of confusing pinyins to an estimated constant. For example, it may be assumed that the likelihood or probability that a user enters pinyin “shang” instead of the intended pinyin “shan” is 0.03. Note that for a given confusing input pinyin P, the sum of the likelihood or probability that the input pinyin is the intended pinyin, i.e., P′=P, and the likelihoods or probabilities that the input pinyin is a variant of the intended pinyin should total approximately 1.0.
  • It is noted that the illustrative fault-tolerant input systems and methods as described herein are particularly applicable in the context of a web search engine and to a search engine for a database containing organized data. However, it is to be understood that the fault-tolerant input systems and method may be adapted and employed for various other applications for spelling error detection and correction, particularly for entries in a non-Romanized language. For example, the system and method may be adapted for a CJK text input application, e.g., word processing application, that detects and corrects spelling errors.
  • The fault-tolerant input systems and method are particularly well suited for use with non-Roman based languages and can be highly effective in both detecting spelling errors and in inferring the correct Chinese characters from pinyin inputs with spelling errors. Such fault-tolerant input system and method allow Chinese users to input Chinese words using approximate pronunciations rather than requiring the use of precise pinyins. The method may also utilize context to reduce uncertainty and/or the perplexity.
  • While the exemplary embodiments of the present invention are described and illustrated herein, it will be appreciated that they are merely illustrative and that modifications can be made to these embodiments without departing from the spirit and scope of the invention. Thus, the scope of the invention is intended to be defined only in terms of the following claims as may be amended, with each claim being expressly incorporated into this Description of Specific Embodiments as an embodiment of the invention.

Claims (38)

1. A fault-tolerant input processing method, comprising:
receiving an input in a format representing characters in a language, the input having at least one original component;
identifying potentially incorrect components of the input;
expanding each potentially incorrect component of the input to at least one additional alternative component in the format, each pair of the potentially incorrect component of the input and the corresponding additional alternative component having a proximity measurement;
converting each component of the input and each additional alternative component in the format to the language;
computing likelihoods of possible conversions of the input to characters in the language, each possible conversion being a combination of the converted original and/or alternative components of the input, the likelihoods being based at least in part on the proximity measurement; and
determining a most likely conversion to characters in the language from the possible conversions of the input.
2. The method of claim 1, wherein the identifying potentially incorrect components of the input is based on pairs of confusingly similar components.
3. The method of claim 1, wherein the identifying potentially incorrect components of the input is based on confusing pairs of vowels and consonants.
4. The method of claim 1, wherein each potentially incorrect component is associated with at least one additional confusingly similar alternative component and wherein the expanding includes expanding each potentially incorrect component to every additional alternative component associated with the potentially incorrect component.
5. The method of claim 1, wherein the converting of each component is one of converting each component individually and converting multiple components jointly.
6. The method of claim 1, wherein the computing likelihoods is also based on a context of the possible conversion.
7. The method of claim 1, wherein the determining the most likely conversion includes determining more than one most likely conversion.
8. The method of claim 1, wherein the most likely conversion includes a first possible conversion comprising only converted original components of the input and a second possible conversion comprising at least one converted alternative component of the input if the likelihood of the second possible conversion exceeds the likelihood of the first possible conversion by a predetermined threshold.
9. The method of claim 1, wherein the language is a non-Roman based language.
10. The method of claim 1, wherein the language is Chinese and the format is pinyin.
11. The method of claim 1, wherein the input is a user search query.
12. A fault-tolerant input system, comprising:
a receiver,adapted to receive an input having at least one original component in a format representing characters in a language;
an identifier adapted to identify potentially incorrect components of the received input;
an expander adapted to expand each potentially incorrect component of the received input to at least one additional alternative component in the format, each pair of the potentially incorrect component of the input and the corresponding additional alternative component having a proximity measurement;
a converter adapted to convert each component of the received input and each additional alternative component in the format to the language;
a computing module adapted to compute likelihoods of possible conversions of the received input to characters in the language, each possible conversion being a combination of the converted original and/or alternative components of the received input, the likelihoods being based at least in part on the proximity measurement; and
an output module adapted to determine a most likely conversion to characters in the language from the possible conversions of the input.
13. The system of claim 12, wherein the identifier is further adapted to identify based on pairs of confusingly similar components.
14. The system of claim 12, wherein the identifier is further adapted to identify based on confusing pairs of vowels and consonants.
15. The system of claim 12, wherein each potentially incorrect component is associated with at least one additional confusingly similar alternative component and wherein the expander is further adapted to expand each potentially incorrect component to every additional alternative component associated with the potentially incorrect component.
16. The system of claim 12, wherein the converter is further adapted to convert each component by one of converting each component individually and converting multiple components jointly.
17. The system of claim 12, wherein the computing module is further adapted to compute likelihoods based on a context of the possible conversion.
18. The system of claim 12, wherein the output module is further adapted to determine more than one most likely conversion.
19. The system of claim 12, wherein the most likely conversion includes a first possible conversion comprising only converted original components of the input and a second possible conversion comprising at least one converted alternative component of the input if the likelihood of the second possible conversion exceeds the likelihood of the first possible conversion by a predetermined threshold.
20. The system of claim 12, wherein the language is a non-Roman based language.
21. The system of claim 12, wherein the language is Chinese and the format is pinyin.
22. The system of claim 12, wherein the system is a search engine and the received input is a search query.
23. A computer program product for use in conjunction with a computer system, the computer program product comprising a computer readable storage medium on which are stored instructions executable on a computer processor, the instructions including:
receiving an input in a format representing characters in a language, the input having at least one original component;
identifying potentially incorrect components of the input;
expanding each potentially incorrect component of the input to at least one additional alternative component in the format, each pair of the potentially incorrect component of the input and the corresponding additional alternative component having a proximity measurement;
converting each component of the input and each additional alternative component in the format to the language;
computing probabilities of possible conversions of the input to characters in the language, each possible conversion being a combination of the converted original and/or alternative components of the input, the probabilities being based at least in part on the proximity measurement; and
determining a most likely conversion to characters in the language from the possible conversions of the input.
24. The computer program product of claim 23, wherein the identifying potentially incorrect components of the input is based on pairs of confusingly similar components.
25. The computer program product of claim 23, wherein the identifying potentially incorrect components of the input is based on confusing pairs of vowels and consonants.
26. The computer program product of claim 23, wherein each potentially incorrect component is associated with at least one additional confusingly similar alternative component and wherein the expanding includes expanding each potentially incorrect component to every additional alternative component associated with the potentially incorrect component.
27. The computer program product of claim 23, wherein the converting of each component is one of converting each component individually and converting multiple components jointly.
28. The computer program product of claim 23, wherein the computing probabilities is also based on a context of the possible conversion.
29. The computer program product of claim 23, wherein the determining the most likely conversion includes determining more than one most likely conversion.
30. The computer program product of claim 23, wherein the most likely conversion includes a first possible conversion comprising only converted original components of the input and a second possible conversion comprising at least one converted alternative component of the input if the probability of the second possible conversion exceeds the probability of the first possible conversion by a predetermined threshold.
31. The computer program product of claim 23, wherein the language is a non-Roman based language.
32. The computer program product of claim 23, wherein the language is Chinese and the format is pinyin.
33. The computer program product of claim 23, wherein the input is a user search query.
34. A method for generating proximity measurements, comprising;
collecting user inputs in a format representing characters in a language and rates of user selection of corresponding candidates decoded from the user inputs into the language;
identifying user inputs having a low rate of candidate selection as non-selected inputs;
assigning a modified input close in confusion proximity to each corresponding non-selected input as a corresponding intended input, the modified input having a higher rate of candidate selection than the corresponding non-selected input; and
determining the proximity measurement for each pair of modified input and non-selected input based on a percentage of the corresponding non-selected input is entered as input without a user selection of any of the decoded candidates being made.
35. The method of claim 34, wherein the confusion proximity is based on a list of pairs of confusingly similar elements in the format and the modified input differs from the corresponding non-selected input by one or more pairs of the confusingly similar elements.
36. The method of claim 34, wherein the language is a non-Roman based language and the format is a Romanized representation of the language.
37. The method of claim 34, wherein the language is Chinese and the format is pinyin.
38. The method of claim 34, wherein the user inputs are user search queries.
US10/928,131 2004-08-25 2004-08-25 Fault-tolerant romanized input method for non-roman characters Abandoned US20060048055A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US10/928,131 US20060048055A1 (en) 2004-08-25 2004-08-25 Fault-tolerant romanized input method for non-roman characters

Applications Claiming Priority (5)

Application Number Priority Date Filing Date Title
US10/928,131 US20060048055A1 (en) 2004-08-25 2004-08-25 Fault-tolerant romanized input method for non-roman characters
PCT/US2005/029141 WO2006026156A2 (en) 2004-08-25 2005-08-15 Fault-tolerant romanized input method for non-roman characters
CN 200580036712 CN101133411B (en) 2004-08-25 2005-08-15 Fault-tolerant romanized input method for non-roman characters
US11/866,991 US7810030B2 (en) 2004-08-25 2007-10-03 Fault-tolerant romanized input method for non-roman characters
US12/897,096 US9069753B2 (en) 2004-08-25 2010-10-04 Determining proximity measurements indicating respective intended inputs

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US11/866,991 Continuation US7810030B2 (en) 2004-08-25 2007-10-03 Fault-tolerant romanized input method for non-roman characters

Publications (1)

Publication Number Publication Date
US20060048055A1 true US20060048055A1 (en) 2006-03-02

Family

ID=35734924

Family Applications (3)

Application Number Title Priority Date Filing Date
US10/928,131 Abandoned US20060048055A1 (en) 2004-08-25 2004-08-25 Fault-tolerant romanized input method for non-roman characters
US11/866,991 Active 2026-03-18 US7810030B2 (en) 2004-08-25 2007-10-03 Fault-tolerant romanized input method for non-roman characters
US12/897,096 Active 2028-02-24 US9069753B2 (en) 2004-08-25 2010-10-04 Determining proximity measurements indicating respective intended inputs

Family Applications After (2)

Application Number Title Priority Date Filing Date
US11/866,991 Active 2026-03-18 US7810030B2 (en) 2004-08-25 2007-10-03 Fault-tolerant romanized input method for non-roman characters
US12/897,096 Active 2028-02-24 US9069753B2 (en) 2004-08-25 2010-10-04 Determining proximity measurements indicating respective intended inputs

Country Status (3)

Country Link
US (3) US20060048055A1 (en)
CN (1) CN101133411B (en)
WO (1) WO2006026156A2 (en)

Cited By (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080052064A1 (en) * 2006-08-25 2008-02-28 Nhn Corporation Method for searching for chinese character using tone mark and system for executing the method
US20080183673A1 (en) * 2007-01-25 2008-07-31 Microsoft Corporation Finite-state model for processing web queries
US20080319947A1 (en) * 2007-06-25 2008-12-25 Sap Ag Mixed initiative semantic search
US20090281788A1 (en) * 2008-05-11 2009-11-12 Michael Elizarov Mobile electronic device and associated method enabling identification of previously entered data for transliteration of an input
US20090278853A1 (en) * 2008-05-12 2009-11-12 Masaharu Ueda Character input program, character input device, and character input method
KR100942218B1 (en) 2007-08-31 2010-02-11 리서치 인 모션 리미티드 Handheld electronic device and associated method enabling phonetic text input in a text disambiguation environment and outputting an improved lookup window
US20100033487A1 (en) * 2004-12-03 2010-02-11 Sony Computer Entertainment Inc. Display device, control method for the same, and information storage medium
US20100286979A1 (en) * 2007-08-01 2010-11-11 Ginger Software, Inc. Automatic context sensitive language correction and enhancement using an internet corpus
US20110153324A1 (en) * 2009-12-23 2011-06-23 Google Inc. Language Model Selection for Speech-to-Text Conversion
US20130090916A1 (en) * 2011-10-05 2013-04-11 Daniel M. Wang System and Method for Detecting and Correcting Mismatched Chinese Character
US20140012569A1 (en) * 2012-07-03 2014-01-09 National Taiwan Normal University System and Method Using Data Reduction Approach and Nonlinear Algorithm to Construct Chinese Readability Model
US9015036B2 (en) 2010-02-01 2015-04-21 Ginger Software, Inc. Automatic context sensitive language correction using an internet corpus particularly for small keyboard devices
US20150213333A1 (en) * 2014-01-28 2015-07-30 Samsung Electronics Co., Ltd. Method and device for realizing chinese character input based on uncertainty information
US9135544B2 (en) 2007-11-14 2015-09-15 Varcode Ltd. System and method for quality management utilizing barcode indicators
US9400952B2 (en) 2012-10-22 2016-07-26 Varcode Ltd. Tamper-proof quality management barcode indicators
US9471566B1 (en) * 2005-04-14 2016-10-18 Oracle America, Inc. Method and apparatus for converting phonetic language input to written language output
US9646277B2 (en) 2006-05-07 2017-05-09 Varcode Ltd. System and method for improved quality management in a product logistic chain
US20170277276A1 (en) * 2015-06-11 2017-09-28 Lianhui ZHOU Method for Inputting Chinese Phrase
US10083231B1 (en) * 2017-11-30 2018-09-25 International Business Machines Corporation Fuzzy term partition identification
US10176451B2 (en) 2007-05-06 2019-01-08 Varcode Ltd. System and method for quality management utilizing barcode indicators

Families Citing this family (93)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8645137B2 (en) 2000-03-16 2014-02-04 Apple Inc. Fast, language-independent method for user authentication by voice
US7565624B2 (en) 2006-06-30 2009-07-21 Research In Motion Limited Method of learning character segments during text input, and associated handheld electronic device
US8395586B2 (en) * 2006-06-30 2013-03-12 Research In Motion Limited Method of learning a context of a segment of text, and associated handheld electronic device
CN100424703C (en) 2006-08-23 2008-10-08 北京搜狗科技发展有限公司 Method for obtaining newly encoded character string, input method system and word stock generation device
US8176419B2 (en) * 2007-12-19 2012-05-08 Microsoft Corporation Self learning contextual spell corrector
US9330720B2 (en) 2008-01-03 2016-05-03 Apple Inc. Methods and apparatus for altering audio output signals
US8996376B2 (en) 2008-04-05 2015-03-31 Apple Inc. Intelligent text-to-speech conversion
US20100030549A1 (en) 2008-07-31 2010-02-04 Lee Michael M Mobile device having human language translation capability with positional feedback
US8589149B2 (en) * 2008-08-05 2013-11-19 Nuance Communications, Inc. Probability-based approach to recognition of user-entered data
US9959870B2 (en) 2008-12-11 2018-05-01 Apple Inc. Speech recognition involving a mobile device
US10241752B2 (en) 2011-09-30 2019-03-26 Apple Inc. Interface for a virtual digital assistant
US9858925B2 (en) 2009-06-05 2018-01-02 Apple Inc. Using context information to facilitate processing of commands in a virtual assistant
US9318108B2 (en) 2010-01-18 2016-04-19 Apple Inc. Intelligent automated assistant
US8682667B2 (en) 2010-02-25 2014-03-25 Apple Inc. User profiling for selecting user specific voice input processing information
CN102830808A (en) * 2010-07-01 2012-12-19 杨春华 Pinyin input method utilizing standard component code as complementary code
JP5694537B2 (en) 2010-09-02 2015-04-01 オプティメディカ・コーポレイションOptimedica Corporation The patient interface for ophthalmic diagnostic and interventional procedures
US9262612B2 (en) 2011-03-21 2016-02-16 Apple Inc. Device access using voice authentication
CN102156551B (en) * 2011-03-30 2014-04-23 北京搜狗科技发展有限公司 Method and system for correcting error of word input
CN102226896A (en) * 2011-05-13 2011-10-26 中山大学 Information fault tolerance input method based on fuzzy keying
US10057736B2 (en) 2011-06-03 2018-08-21 Apple Inc. Active transport based notifications
US10241644B2 (en) 2011-06-03 2019-03-26 Apple Inc. Actionable reminder entries
US8706472B2 (en) * 2011-08-11 2014-04-22 Apple Inc. Method for disambiguating multiple readings in language conversion
CN102955770B (en) * 2011-08-17 2017-07-11 深圳市世纪光速信息技术有限公司 One kind of automatic identification method and system for phonetic
US8994660B2 (en) 2011-08-29 2015-03-31 Apple Inc. Text correction processing
CN102368271A (en) * 2011-11-01 2012-03-07 无敌科技(西安)有限公司 Chinese content spelling correcting system and method with fault-tolerant capability
US8976118B2 (en) 2012-01-20 2015-03-10 International Business Machines Corporation Method for character correction
US9317606B1 (en) * 2012-02-03 2016-04-19 Google Inc. Spell correcting long queries
US10134385B2 (en) 2012-03-02 2018-11-20 Apple Inc. Systems and methods for name pronunciation
US9483461B2 (en) 2012-03-06 2016-11-01 Apple Inc. Handling speech synthesis of content for multiple languages
US9280610B2 (en) 2012-05-14 2016-03-08 Apple Inc. Crowd sourcing information to fulfill user requests
US9721563B2 (en) 2012-06-08 2017-08-01 Apple Inc. Name recognition system
US9495129B2 (en) 2012-06-29 2016-11-15 Apple Inc. Device, method, and user interface for voice-activated navigation and browsing of a document
US9135912B1 (en) * 2012-08-15 2015-09-15 Google Inc. Updating phonetic dictionaries
US9576574B2 (en) 2012-09-10 2017-02-21 Apple Inc. Context-sensitive handling of interruptions by intelligent digital assistant
US9547647B2 (en) 2012-09-19 2017-01-17 Apple Inc. Voice-based media searching
US8713433B1 (en) * 2012-10-16 2014-04-29 Google Inc. Feature-based autocorrection
US8612213B1 (en) 2012-10-16 2013-12-17 Google Inc. Correction of errors in character strings that include a word delimiter
CN104969289A (en) 2013-02-07 2015-10-07 苹果公司 Voice trigger for a digital assistant
US9368114B2 (en) 2013-03-14 2016-06-14 Apple Inc. Context-sensitive handling of interruptions
AU2014233517B2 (en) 2013-03-15 2017-05-25 Apple Inc. Training an at least partial voice command system
WO2014144579A1 (en) 2013-03-15 2014-09-18 Apple Inc. System and method for updating an adaptive speech recognition model
US9317499B2 (en) * 2013-04-11 2016-04-19 International Business Machines Corporation Optimizing generation of a regular expression
US9298694B2 (en) * 2013-04-11 2016-03-29 International Business Machines Corporation Generating a regular expression for entity extraction
WO2014197334A2 (en) 2013-06-07 2014-12-11 Apple Inc. System and method for user-specified pronunciation of words for speech synthesis and recognition
US9582608B2 (en) 2013-06-07 2017-02-28 Apple Inc. Unified ranking with entropy-weighted information for phrase-based semantic auto-completion
WO2014197336A1 (en) 2013-06-07 2014-12-11 Apple Inc. System and method for detecting errors in interactions with a voice-based digital assistant
WO2014197335A1 (en) 2013-06-08 2014-12-11 Apple Inc. Interpreting and acting upon commands that involve sharing information with remote devices
US10176167B2 (en) 2013-06-09 2019-01-08 Apple Inc. System and method for inferring user intent from speech inputs
AU2014278592B2 (en) 2013-06-09 2017-09-07 Apple Inc. Device, method, and graphical user interface for enabling conversation persistence across two or more instances of a digital assistant
JP2016521948A (en) 2013-06-13 2016-07-25 アップル インコーポレイテッド System and method for emergency call initiated by voice command
CN104750672B (en) * 2013-12-27 2017-11-21 重庆新媒农信科技有限公司 One kind of Chinese word error correction method applied to the apparatus and the search
US9620105B2 (en) 2014-05-15 2017-04-11 Apple Inc. Analyzing audio input for efficient speech and music recognition
US9502031B2 (en) 2014-05-27 2016-11-22 Apple Inc. Method for supporting dynamic grammars in WFST-based ASR
US9760559B2 (en) 2014-05-30 2017-09-12 Apple Inc. Predictive text input
US10170123B2 (en) 2014-05-30 2019-01-01 Apple Inc. Intelligent assistant for home automation
US9715875B2 (en) 2014-05-30 2017-07-25 Apple Inc. Reducing the need for manual start/end-pointing and trigger phrases
US9430463B2 (en) 2014-05-30 2016-08-30 Apple Inc. Exemplar-based natural language processing
US10078631B2 (en) 2014-05-30 2018-09-18 Apple Inc. Entropy-guided text prediction using combined word and character n-gram language models
WO2015184186A1 (en) 2014-05-30 2015-12-03 Apple Inc. Multi-command single utterance input method
US9734193B2 (en) 2014-05-30 2017-08-15 Apple Inc. Determining domain salience ranking from ambiguous words in natural speech
US9633004B2 (en) 2014-05-30 2017-04-25 Apple Inc. Better resolution when referencing to concepts
US9842101B2 (en) 2014-05-30 2017-12-12 Apple Inc. Predictive conversion of language input
US9785630B2 (en) 2014-05-30 2017-10-10 Apple Inc. Text prediction using combined word N-gram and unigram language models
US9338493B2 (en) 2014-06-30 2016-05-10 Apple Inc. Intelligent automated assistant for TV user interactions
US9377871B2 (en) 2014-08-01 2016-06-28 Nuance Communications, Inc. System and methods for determining keyboard input in the presence of multiple contact points
US9818400B2 (en) 2014-09-11 2017-11-14 Apple Inc. Method and apparatus for discovering trending terms in speech requests
US9646609B2 (en) 2014-09-30 2017-05-09 Apple Inc. Caching apparatus for serving phonetic pronunciations
US10074360B2 (en) 2014-09-30 2018-09-11 Apple Inc. Providing an indication of the suitability of speech recognition
US9886432B2 (en) 2014-09-30 2018-02-06 Apple Inc. Parsimonious handling of word inflection via categorical stem + suffix N-gram language models
US9668121B2 (en) 2014-09-30 2017-05-30 Apple Inc. Social reminders
US10127911B2 (en) 2014-09-30 2018-11-13 Apple Inc. Speaker identification and unsupervised speaker adaptation techniques
US9711141B2 (en) 2014-12-09 2017-07-18 Apple Inc. Disambiguating heteronyms in speech synthesis
US9865280B2 (en) 2015-03-06 2018-01-09 Apple Inc. Structured dictation using intelligent automated assistants
US9721566B2 (en) 2015-03-08 2017-08-01 Apple Inc. Competing devices responding to voice triggers
US9886953B2 (en) 2015-03-08 2018-02-06 Apple Inc. Virtual assistant activation
US9899019B2 (en) 2015-03-18 2018-02-20 Apple Inc. Systems and methods for structured stem and suffix language models
US9842105B2 (en) 2015-04-16 2017-12-12 Apple Inc. Parsimonious continuous-space phrase representations for natural language processing
US10083688B2 (en) 2015-05-27 2018-09-25 Apple Inc. Device voice control for selecting a displayed affordance
US10127220B2 (en) 2015-06-04 2018-11-13 Apple Inc. Language identification from short strings
US10101822B2 (en) 2015-06-05 2018-10-16 Apple Inc. Language input correction
US10186254B2 (en) 2015-06-07 2019-01-22 Apple Inc. Context-based endpoint detection
CN104991657A (en) * 2015-06-11 2015-10-21 周连惠 Chinese and Japanese katakana integrated input method and input method system
US9697820B2 (en) 2015-09-24 2017-07-04 Apple Inc. Unit-selection text-to-speech synthesis using concatenation-sensitive neural networks
US10049668B2 (en) 2015-12-02 2018-08-14 Apple Inc. Applying neural network language models to weighted finite state transducers for automatic speech recognition
US10223066B2 (en) 2015-12-23 2019-03-05 Apple Inc. Proactive assistance based on dialog communication between devices
US9934775B2 (en) 2016-05-26 2018-04-03 Apple Inc. Unit-selection text-to-speech synthesis based on predicted concatenation parameters
CN106020506A (en) * 2016-05-30 2016-10-12 北京小米移动软件有限公司 Information input method and device
US9972304B2 (en) 2016-06-03 2018-05-15 Apple Inc. Privacy preserving distributed evaluation framework for embedded personalized systems
US10049663B2 (en) 2016-06-08 2018-08-14 Apple, Inc. Intelligent automated assistant for media exploration
US10192552B2 (en) 2016-06-10 2019-01-29 Apple Inc. Digital assistant providing whispered speech
US10067938B2 (en) * 2016-06-10 2018-09-04 Apple Inc. Multilingual word prediction
DK179415B1 (en) 2016-06-11 2018-06-14 Apple Inc Intelligent device arbitration and control
US10043516B2 (en) 2016-09-23 2018-08-07 Apple Inc. Intelligent automated assistant

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5477448A (en) * 1994-06-01 1995-12-19 Mitsubishi Electric Research Laboratories, Inc. System for correcting improper determiners
US5903861A (en) * 1995-12-12 1999-05-11 Chan; Kun C. Method for specifically converting non-phonetic characters representing vocabulary in languages into surrogate words for inputting into a computer
US6014615A (en) * 1994-08-16 2000-01-11 International Business Machines Corporaiton System and method for processing morphological and syntactical analyses of inputted Chinese language phrases
US6047300A (en) * 1997-05-15 2000-04-04 Microsoft Corporation System and method for automatically correcting a misspelled word
US6401060B1 (en) * 1998-06-25 2002-06-04 Microsoft Corporation Method for typographical detection and replacement in Japanese text
US6636849B1 (en) * 1999-11-23 2003-10-21 Genmetrics, Inc. Data search employing metric spaces, multigrid indexes, and B-grid trees
US6848080B1 (en) * 1999-11-05 2005-01-25 Microsoft Corporation Language input architecture for converting one text form to another text form with tolerance to spelling, typographical, and conversion errors

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4833610A (en) * 1986-12-16 1989-05-23 International Business Machines Corporation Morphological/phonetic method for ranking word similarities
US5572423A (en) * 1990-06-14 1996-11-05 Lucent Technologies Inc. Method for correcting spelling using error frequencies
US6131102A (en) * 1998-06-15 2000-10-10 Microsoft Corporation Method and system for cost computation of spelling suggestions and automatic replacement
US7107204B1 (en) * 2000-04-24 2006-09-12 Microsoft Corporation Computer-aided writing system and method with cross-language writing wizard
US7149970B1 (en) * 2000-06-23 2006-12-12 Microsoft Corporation Method and system for filtering and selecting from a candidate list generated by a stochastic input method
CA2357382A1 (en) * 2001-09-17 2003-03-17 Soma Networks, Inc. Software update method, apparatus and system
US7143033B2 (en) * 2002-04-03 2006-11-28 The United States Of America As Represented By The Secretary Of The Navy Automatic multi-language phonetic transcribing system
US7113950B2 (en) * 2002-06-27 2006-09-26 Microsoft Corporation Automated error checking system and method
US7315982B2 (en) * 2003-02-26 2008-01-01 Xerox Corporation User-tailorable romanized Chinese text input systems and methods
US7395203B2 (en) * 2003-07-30 2008-07-01 Tegic Communications, Inc. System and method for disambiguating phonetic input

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5477448A (en) * 1994-06-01 1995-12-19 Mitsubishi Electric Research Laboratories, Inc. System for correcting improper determiners
US6014615A (en) * 1994-08-16 2000-01-11 International Business Machines Corporaiton System and method for processing morphological and syntactical analyses of inputted Chinese language phrases
US5903861A (en) * 1995-12-12 1999-05-11 Chan; Kun C. Method for specifically converting non-phonetic characters representing vocabulary in languages into surrogate words for inputting into a computer
US6047300A (en) * 1997-05-15 2000-04-04 Microsoft Corporation System and method for automatically correcting a misspelled word
US6401060B1 (en) * 1998-06-25 2002-06-04 Microsoft Corporation Method for typographical detection and replacement in Japanese text
US6848080B1 (en) * 1999-11-05 2005-01-25 Microsoft Corporation Language input architecture for converting one text form to another text form with tolerance to spelling, typographical, and conversion errors
US6636849B1 (en) * 1999-11-23 2003-10-21 Genmetrics, Inc. Data search employing metric spaces, multigrid indexes, and B-grid trees

Cited By (56)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100033487A1 (en) * 2004-12-03 2010-02-11 Sony Computer Entertainment Inc. Display device, control method for the same, and information storage medium
US7773089B2 (en) * 2004-12-03 2010-08-10 Sony Computer Entertainment Inc. Displaying character strings based on font information
US9471566B1 (en) * 2005-04-14 2016-10-18 Oracle America, Inc. Method and apparatus for converting phonetic language input to written language output
US10037507B2 (en) 2006-05-07 2018-07-31 Varcode Ltd. System and method for improved quality management in a product logistic chain
US9646277B2 (en) 2006-05-07 2017-05-09 Varcode Ltd. System and method for improved quality management in a product logistic chain
US8271265B2 (en) * 2006-08-25 2012-09-18 Nhn Corporation Method for searching for chinese character using tone mark and system for executing the method
US20080052064A1 (en) * 2006-08-25 2008-02-28 Nhn Corporation Method for searching for chinese character using tone mark and system for executing the method
US20080183673A1 (en) * 2007-01-25 2008-07-31 Microsoft Corporation Finite-state model for processing web queries
US8024319B2 (en) 2007-01-25 2011-09-20 Microsoft Corporation Finite-state model for processing web queries
US10176451B2 (en) 2007-05-06 2019-01-08 Varcode Ltd. System and method for quality management utilizing barcode indicators
US7987176B2 (en) * 2007-06-25 2011-07-26 Sap Ag Mixed initiative semantic search
US20080319947A1 (en) * 2007-06-25 2008-12-25 Sap Ag Mixed initiative semantic search
US8229944B2 (en) 2007-06-25 2012-07-24 Sap Ag Mixed initiative semantic search
US9026432B2 (en) 2007-08-01 2015-05-05 Ginger Software, Inc. Automatic context sensitive language generation, correction and enhancement using an internet corpus
US20100286979A1 (en) * 2007-08-01 2010-11-11 Ginger Software, Inc. Automatic context sensitive language correction and enhancement using an internet corpus
US8914278B2 (en) * 2007-08-01 2014-12-16 Ginger Software, Inc. Automatic context sensitive language correction and enhancement using an internet corpus
KR100942218B1 (en) 2007-08-31 2010-02-11 리서치 인 모션 리미티드 Handheld electronic device and associated method enabling phonetic text input in a text disambiguation environment and outputting an improved lookup window
US9558439B2 (en) 2007-11-14 2017-01-31 Varcode Ltd. System and method for quality management utilizing barcode indicators
US9135544B2 (en) 2007-11-14 2015-09-15 Varcode Ltd. System and method for quality management utilizing barcode indicators
US9836678B2 (en) 2007-11-14 2017-12-05 Varcode Ltd. System and method for quality management utilizing barcode indicators
US8463597B2 (en) * 2008-05-11 2013-06-11 Research In Motion Limited Mobile electronic device and associated method enabling identification of previously entered data for transliteration of an input
US20090281788A1 (en) * 2008-05-11 2009-11-12 Michael Elizarov Mobile electronic device and associated method enabling identification of previously entered data for transliteration of an input
US8725491B2 (en) 2008-05-11 2014-05-13 Blackberry Limited Mobile electronic device and associated method enabling identification of previously entered data for transliteration of an input
US20090278853A1 (en) * 2008-05-12 2009-11-12 Masaharu Ueda Character input program, character input device, and character input method
US8307281B2 (en) * 2008-05-12 2012-11-06 Omron Corporation Predicting conversion candidates based on the current context and the attributes of previously selected conversion candidates
US10089566B2 (en) 2008-06-10 2018-10-02 Varcode Ltd. Barcoded indicators for quality management
US9710743B2 (en) 2008-06-10 2017-07-18 Varcode Ltd. Barcoded indicators for quality management
US9626610B2 (en) 2008-06-10 2017-04-18 Varcode Ltd. System and method for quality management utilizing barcode indicators
US10049314B2 (en) 2008-06-10 2018-08-14 Varcode Ltd. Barcoded indicators for quality management
US9317794B2 (en) 2008-06-10 2016-04-19 Varcode Ltd. Barcoded indicators for quality management
US9996783B2 (en) 2008-06-10 2018-06-12 Varcode Ltd. System and method for quality management utilizing barcode indicators
US9384435B2 (en) 2008-06-10 2016-07-05 Varcode Ltd. Barcoded indicators for quality management
US9646237B2 (en) 2008-06-10 2017-05-09 Varcode Ltd. Barcoded indicators for quality management
US10157040B2 (en) 2009-12-23 2018-12-18 Google Llc Multi-modal input on an electronic device
US20110153324A1 (en) * 2009-12-23 2011-06-23 Google Inc. Language Model Selection for Speech-to-Text Conversion
US9047870B2 (en) 2009-12-23 2015-06-02 Google Inc. Context based language model selection
US20110153325A1 (en) * 2009-12-23 2011-06-23 Google Inc. Multi-Modal Input on an Electronic Device
US20110161080A1 (en) * 2009-12-23 2011-06-30 Google Inc. Speech to Text Conversion
US9031830B2 (en) 2009-12-23 2015-05-12 Google Inc. Multi-modal input on an electronic device
US8751217B2 (en) 2009-12-23 2014-06-10 Google Inc. Multi-modal input on an electronic device
US20110161081A1 (en) * 2009-12-23 2011-06-30 Google Inc. Speech Recognition Language Models
US9251791B2 (en) 2009-12-23 2016-02-02 Google Inc. Multi-modal input on an electronic device
US9495127B2 (en) 2009-12-23 2016-11-15 Google Inc. Language model selection for speech-to-text conversion
US9015036B2 (en) 2010-02-01 2015-04-21 Ginger Software, Inc. Automatic context sensitive language correction using an internet corpus particularly for small keyboard devices
US20130090916A1 (en) * 2011-10-05 2013-04-11 Daniel M. Wang System and Method for Detecting and Correcting Mismatched Chinese Character
US8725497B2 (en) * 2011-10-05 2014-05-13 Daniel M. Wang System and method for detecting and correcting mismatched Chinese character
US20140012569A1 (en) * 2012-07-03 2014-01-09 National Taiwan Normal University System and Method Using Data Reduction Approach and Nonlinear Algorithm to Construct Chinese Readability Model
US9965712B2 (en) 2012-10-22 2018-05-08 Varcode Ltd. Tamper-proof quality management barcode indicators
US9400952B2 (en) 2012-10-22 2016-07-26 Varcode Ltd. Tamper-proof quality management barcode indicators
US9633296B2 (en) 2012-10-22 2017-04-25 Varcode Ltd. Tamper-proof quality management barcode indicators
US10242302B2 (en) 2012-10-22 2019-03-26 Varcode Ltd. Tamper-proof quality management barcode indicators
US20150213333A1 (en) * 2014-01-28 2015-07-30 Samsung Electronics Co., Ltd. Method and device for realizing chinese character input based on uncertainty information
US10242296B2 (en) * 2014-01-28 2019-03-26 Samsung Electronics Co., Ltd. Method and device for realizing chinese character input based on uncertainty information
US20170277276A1 (en) * 2015-06-11 2017-09-28 Lianhui ZHOU Method for Inputting Chinese Phrase
US10042433B2 (en) * 2015-06-11 2018-08-07 Lianhui ZHOU Method for inputting chinese phrase
US10083231B1 (en) * 2017-11-30 2018-09-25 International Business Machines Corporation Fuzzy term partition identification

Also Published As

Publication number Publication date
US20080028303A1 (en) 2008-01-31
CN101133411B (en) 2014-08-27
US20110022952A1 (en) 2011-01-27
WO2006026156A3 (en) 2006-10-19
WO2006026156A2 (en) 2006-03-09
CN101133411A (en) 2008-02-27
US7810030B2 (en) 2010-10-05
US9069753B2 (en) 2015-06-30

Similar Documents

Publication Publication Date Title
Kukich Techniques for automatically correcting words in text
Choudhury et al. Investigation and modeling of the structure of texting language
US8463598B2 (en) Word detection
JP5599662B2 (en) System and method for converting a Chinese character pronunciation sequence of native language using statistical methods
US5845306A (en) Context based system for accessing dictionary entries
CN100492350C (en) Language input architecture for converting one text form to another text form with modeless entry
CN101568918B (en) Web-based collocation error proofing
US8457946B2 (en) Recognition architecture for generating Asian characters
US6347300B1 (en) Speech correction apparatus and method
US5485372A (en) System for underlying spelling recovery
CN101467125B (en) Processing of query terms
Han et al. Automatically constructing a normalisation dictionary for microblogs
Liu et al. A broad-coverage normalization system for social media language
US5477448A (en) System for correcting improper determiners
US7447627B2 (en) Compound word breaker and spell checker
US5535121A (en) System for correcting auxiliary verb sequences
Han et al. Lexical normalisation of short text messages: Makn sens a# twitter
JP5997217B2 (en) Method of removing ambiguity plurality of readings in the language conversion
US7149970B1 (en) Method and system for filtering and selecting from a candidate list generated by a stochastic input method
US20050216253A1 (en) System and method for reverse transliteration using statistical alignment
KR101292404B1 (en) Method and system for generating spelling suggestions
US8504350B2 (en) User-interactive automatic translation device and method for mobile device
US7818332B2 (en) Query speller
US6848080B1 (en) Language input architecture for converting one text form to another text form with tolerance to spelling, typographical, and conversion errors
US5521816A (en) Word inflection correction system

Legal Events

Date Code Title Description
AS Assignment

Owner name: GOOGLE INC., CALIFORNIA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:WU, JUN;CHEN, LIREN;REEL/FRAME:016549/0097

Effective date: 20040825

AS Assignment

Owner name: GOOGLE LLC, CALIFORNIA

Free format text: CHANGE OF NAME;ASSIGNOR:GOOGLE INC.;REEL/FRAME:044142/0357

Effective date: 20170929