Text error correction method, device, electronic device and readable storage medium

Abstract

The present disclosure provides a text error correction method, apparatus, electronic device, and readable storage medium, which relate to the technical fields of artificial intelligence such as natural language processing and deep learning. Among them, the text error correction method includes: obtaining a text to be processed and determining the target text in the text to be processed; obtaining at least one candidate phoneme sequence according to a phoneme dictionary and the original phoneme sequence of the target text; selecting a target word from the multiple candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the multiple candidate words; and obtaining an error correction result of the text to be processed according to the target word. The present disclosure can simplify the steps of text error correction, reduce the cost of text error correction, realize the cold start of text error correction, and also achieve a good error correction effect when there are "phonetically similar but misspelled words" in the text, further improving the accuracy of text error correction.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technologies, and particularly to artificial intelligence technologies such as natural language processing and deep learning. Specifically, a text error correction method, apparatus, electronic device, and readable storage medium are provided. Background Art

[0002] In a Chinese environment, an automatic speech recognition system needs to recognize voice signals into Chinese characters. Since there are cases of multiple characters with the same pronunciation or a single character with multiple pronunciations in Chinese characters, the automatic speech recognition system tends to generate a Chinese character sequence that is most likely to appear in the training samples. In addition, since the user may have inaccurate pronunciation or noise is introduced during the channel transmission of the voice signal, resulting in an inability to correctly obtain the Chinese character result from the voice signal, text error correction is required.

[0003] Similarly, in a pure text scenario, there are problems such as spelling mistakes, erroneously added characters, or expression completion, and text error correction is also required. However, when the existing technologies perform text error correction, they usually rely on machine learning technologies, so there are problems of high error correction costs and inability to perform cold start. Summary of the Invention

[0004] According to a first aspect of the present disclosure, a text error correction method is provided, including: obtaining a text to be processed, and determining a target text in the text to be processed; obtaining at least one candidate phoneme sequence according to a phoneme dictionary and an original phoneme sequence of the target text; selecting a target word from the multiple candidate words according to the at least one candidate phoneme sequence and phoneme sequences of the multiple candidate words; and obtaining an error correction result of the text to be processed according to the target word.

[0005] According to a second aspect of the present disclosure, a text error correction apparatus is provided, including: an obtaining unit, configured to obtain a text to be processed and determine a target text in the text to be processed; a processing unit, configured to obtain at least one candidate phoneme sequence according to a phoneme dictionary and an original phoneme sequence of the target text; a selecting unit, configured to select a target word from the multiple candidate words according to the at least one candidate phoneme sequence and phoneme sequences of the multiple candidate words; and an error correction unit, configured to obtain an error correction result of the text to be processed according to the target word.

[0006] According to a third aspect of the present disclosure, an electronic device is provided, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor is enabled to execute the method as described above.

[0007] According to a fourth aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the method described above.

[0008] According to a fifth aspect of the present disclosure, there is provided a computer program product including a computer program which, when executed by a processor, implements the method described above.

[0009] It can be seen from the above technical solutions that the present disclosure can simplify the steps of text error correction, reduce the cost of text error correction, achieve cold start of text error correction, and also achieve good error correction effects in the case of "phonetic similarity and character error" in the text, further improving the accuracy of text error correction.

[0010] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The drawings are used to better understand the solution and do not constitute a limitation to the present disclosure. Among them:

[0012] Figure 1 is a schematic diagram according to the first embodiment of the present disclosure;

[0013] Figure 2 is a schematic diagram according to the second embodiment of the present disclosure;

[0014] Figure 3 is a schematic diagram according to the third embodiment of the present disclosure;

[0015] Figure 4 is a schematic diagram according to the fourth embodiment of the present disclosure;

[0016] Figure 5 is a schematic diagram according to the fifth embodiment of the present disclosure;

[0017] Figure 6 is a schematic diagram according to the sixth embodiment of the present disclosure;

[0018] Figure 7 is a schematic diagram according to the seventh embodiment of the present disclosure;

[0019] Figure 8 is a schematic diagram according to the eighth embodiment of the present disclosure;

[0020] Figure 9 is a block diagram of an electronic device for implementing the text error correction method of the embodiments of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] The following describes exemplary embodiments of the present disclosure with reference to the accompanying drawings. Various details of the embodiments of the present disclosure are included to facilitate understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, descriptions of well-known functions and mechanisms are omitted in the following description for clarity and conciseness.

[0022] Figure 1 is a schematic diagram according to the first embodiment of the present disclosure. As Figure 1 shown, the text error correction method of this embodiment specifically includes the following steps:

[0023] S101. Obtain the text to be processed and determine the target text in the text to be processed;

[0024] S102. Obtain at least one candidate phoneme sequence according to the phoneme dictionary and the original phoneme sequence of the target text;

[0025] S103. Select the target word from the multiple candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the multiple candidate words;

[0026] S104. Obtain the error correction result of the text to be processed according to the target word.

[0027] For the text error correction method of this embodiment, after determining the target text in the text to be processed, first obtain at least one candidate phoneme sequence according to the phoneme dictionary and the original phoneme sequence of the target text, then select the target word from the multiple candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the multiple candidate words, and finally obtain the error correction result of the text to be processed according to the target word. This embodiment performs text error correction based on the similarity of sounds, does not rely on any training data, simplifies the steps of text error correction, reduces the cost of text error correction, can achieve the cold start of text error correction, and can also achieve good error correction effects in the case of "phonetically similar but character wrong" in the text, thereby improving the accuracy of text error correction.

[0028] When this embodiment executes S101 to obtain the text to be processed, the text input at the input end can be used as the text to be processed, or the text selected by the input end from the network can be used as the text to be processed, or the voice input at the input end can be recognized and the recognized text can be used as the text to be processed.

[0029] That is to say, the text error correction method of this embodiment is applicable to a variety of different scenarios, and can be applied to a pure text scenario or a speech recognition scenario.

[0030] After this embodiment executes S101 to obtain the text to be processed, it executes the step of determining the target text in the obtained text to be processed. Among them, the target text determined by this embodiment when executing S101 can be the text to be processed itself, or a part of the text to be processed, such as keywords or entity words in the text to be processed.

[0031] When this embodiment executes S101 to determine the target text in the text to be processed, an optional implementation method that can be adopted is: input the text to be processed into an entity extraction model, and use the output result of the entity extraction model as the target text in the text to be processed. That is, this embodiment uses an entity extraction model and uses the entity words extracted by the entity extraction model from the text to be processed as the target text.

[0032] After this embodiment executes S101 to determine the target text in the text to be processed, it executes S102 to obtain at least one candidate phoneme sequence corresponding to the target text according to the phoneme dictionary and the original phoneme sequence of the determined target text.

[0033] Among them, the phoneme dictionary used by this embodiment when executing S102 is pre-set. The phoneme dictionary contains multiple original phonemes and the similar phonemes corresponding to each original phoneme; the number of similar phonemes corresponding to an original phoneme can be one or multiple.

[0034] For example, if the original phonemes included in the phoneme dictionary in this embodiment are respectively "a", "ang", "ing", "l", and "n". Among them, the similar phonemes corresponding to the original phoneme "a" can be "an" and "ang"; the similar phonemes corresponding to the original phoneme "ang" can be "an"; the similar phonemes corresponding to the original phoneme "ing" can be "in"; the similar phonemes corresponding to the original phoneme "l" can be "n"; the similar phonemes corresponding to the original phoneme "n" can be "l".

[0035] The original phoneme sequence or candidate phoneme sequence in this embodiment is composed of at least one phoneme subsequence, and each phoneme subsequence corresponds to a Chinese character; the phoneme subsequence is composed of at least one phoneme, and the phoneme represents the smallest unit of pronunciation; in order to simplify the processing process, the phonemes in this embodiment can be phonemes without tones.

[0036] When this embodiment executes S102, for each character in the target text, it can obtain the phoneme subsequence corresponding to the character, and then obtain the original phoneme sequence of the target text according to the obtained phoneme subsequences of each character; the original phoneme sequence obtained by this embodiment when executing S102 can be one or multiple.

[0037] For example, if the target text is "小行", in consideration of polyphones, the original phoneme sequence of "小行" obtained by executing S102 in this embodiment may be "xi ao, x ing" and "xi ao, hang". Among them, in the original phoneme sequence "xi ao, xing", the phoneme subsequence "xi ao" consists of three phonemes, corresponding to the Chinese character "小" in the target text; the phoneme subsequence "x ing" consists of two phonemes, corresponding to the Chinese character "行" in the target text.

[0038] When executing S102 in this embodiment to obtain at least one candidate phoneme sequence according to the phoneme dictionary and the original phoneme sequence of the target text, an optional implementation method that can be adopted is: for each phoneme in the original phoneme sequence, obtain a similar phoneme corresponding to the phoneme in the phoneme dictionary; after replacing the phoneme in the original phoneme sequence with the similar phoneme corresponding to it, use the phoneme replacement result and the original phoneme sequence as at least one candidate phoneme sequence.

[0039] That is to say, since the provided phoneme dictionary contains different phonemes and their corresponding similar phonemes, this embodiment can achieve the purpose of obtaining all possible phoneme sequences of the target text based on the original phoneme sequence of the target text, thereby ensuring that the obtained candidate phoneme sequences correspond to different pronunciations of the target text and improving the comprehensiveness of the obtained candidate phoneme sequences.

[0040] For example, if the target text in this embodiment is "小行", and if the original phoneme sequence is "xi ao, xing"; if in the phoneme dictionary, the similar phoneme corresponding to the original phoneme "ing" is "in", then this embodiment executes S102 to obtain the similar phoneme "in" to replace "ing" in the original phoneme sequence, thereby obtaining the phoneme replacement result of "xi ao, x in", and then taking "x iao, x ing" and "xi ao, x in" as candidate phoneme sequences.

[0041] It is understandable that if the present embodiment fails to obtain similar phonemes when executing S102, the original phoneme sequence may be directly used as a candidate phoneme sequence.

[0042] After executing S102 to obtain at least one candidate phoneme sequence, this embodiment executes S103 to select a target word from the plurality of candidate words according to the obtained at least one candidate phoneme sequence and the phoneme sequences of the plurality of candidate words.

[0043] Among them, the multiple candidate words used in executing S103 in this embodiment can be preset by the input end, that is, the input end can customize the candidate words according to actual needs, thereby realizing personalized text error correction.

[0044] In some technical fields, such as finance, technology, medicine and other technical fields, there are a large number of professional terms, and it is somewhat difficult to correct errors in professional terms. Therefore, when implementing S103, this embodiment may further include the following: determining the technical field to which the target text belongs; obtaining a plurality of terms corresponding to the determined technical field as a plurality of candidate terms. This embodiment may obtain a plurality of candidate terms through a thesaurus corresponding to different technical fields.

[0045] In order to further improve the error correction efficiency, the phoneme sequence of the candidate terms used when implementing S103 in this embodiment may be pre-set; when pre-setting the phoneme sequence of the candidate terms in this embodiment, only the case of polyphonic characters may be considered.

[0046] When implementing S103 in this embodiment, the method of obtaining the candidate phoneme sequence of the target text described above may also be used. According to the original phoneme sequence of the candidate terms, at least one obtained candidate phoneme sequence is used as the phoneme sequence of the candidate terms. The specific process will not be elaborated here.

[0047] Specifically, when implementing S103 to select a target term from a plurality of candidate terms according to at least one obtained candidate phoneme sequence and the phoneme sequences of the plurality of candidate terms, an optional implementation method that can be adopted is: obtaining the similarity between the target text and the plurality of candidate terms according to at least one obtained candidate phoneme sequence and the phoneme sequences of the plurality of candidate terms; using the candidate term corresponding to the maximum value of the obtained similarity as the target term.

[0048] Since a candidate term may also have multiple phoneme sequences, when implementing S103 in this embodiment, one can be respectively selected from at least one candidate phoneme sequence and the multiple phoneme sequences of a candidate term. Then, according to the candidate phoneme sequence and the phoneme sequence selected each time, multiple similarity calculations are performed to obtain the similarity between the target text and the candidate term.

[0049] After implementing S103 to select a target term from a plurality of candidate terms in this embodiment, S104 is implemented to obtain the error correction result of the text to be processed according to the selected target term.

[0050] When implementing S104 in this embodiment, the selected target term may be used to replace the target text in the text to be processed, and then the replacement result is used as the error correction result of the text to be processed; this embodiment may also directly output the selected target term as the error correction result of the text to be processed.

[0051] That is to say, this embodiment replaces the target text in the text to be processed with the selected target term, achieving the purpose of determining the target term among a plurality of candidate terms through sound similarity, and can improve the efficiency and accuracy of text error correction.

[0052] Figure 2 It is a schematic diagram according to the second embodiment of the present disclosure. As Figure 2 shown, when performing S103 "obtaining the similarity between the target text and multiple candidate words according to at least one candidate phoneme sequence and the phoneme sequences of multiple candidate words" in this embodiment, the following specific steps are included:

[0053] S201. For each candidate word, select a candidate phoneme sequence and a phoneme sequence of this candidate word;

[0054] S202. When it is determined that the target text and this candidate word have the same length value, count the number of occurrences of the phonemes included in the phoneme subsequence of the Chinese character at each position in the candidate phoneme sequence in the phoneme subsequence of the Chinese character at the corresponding position in the phoneme sequence of this candidate word;

[0055] S203. Obtain the first similarity between the target text and this candidate word according to the counted number of occurrences and the target length value.

[0056] That is to say, based on the characteristics of Chinese pinyin in this embodiment, when the target text and the candidate word are sequences of equal length, according to the candidate phoneme sequence of the target text and the phoneme sequence of the candidate word, the calculation of phoneme similarity is realized at the Chinese character granularity, and the first similarity between the target text and the candidate word is obtained, thereby improving the accuracy when calculating the similarity of sequences of equal length.

[0057] It can be understood that if there is a situation where the current candidate word corresponds to multiple phoneme sequences, when calculating the similarity between the target text and the current candidate word in this embodiment, the calculation can be performed according to the candidate phoneme sequence and each phoneme sequence, and then the maximum value of the calculated similarities is used as the first similarity between the target text and the current candidate word.

[0058] Among them, the target length value used when this embodiment performs S203 is specifically the length value of the sequence with the shorter length in the candidate phoneme sequence and the phoneme sequence.

[0059] When this embodiment performs S203 to obtain the first similarity between the target text and this candidate word according to the counted number of occurrences and the target length value, the following calculation formula can be used:

[0060]

[0061] In the formula: Sim1 represents the first similarity between the target text and the candidate word; M1 represents the length value of the candidate phoneme sequence of the target text, M2 represents the length value of the phoneme sequence of the candidate word; k represents the kth Chinese character; N represents the number of Chinese characters in the target text or the candidate word; It indicates that the m-th phoneme belongs to the phoneme subsequence corresponding to the c-th Chinese character; I represents a counting function, and when the condition in the function holds, the calculation result is 1, otherwise it is 0. It indicates that the m-th phoneme in the phoneme subsequence corresponding to the k-th Chinese character in the target text and in the j1-th candidate phoneme sequence belongs to the phoneme subsequence corresponding to the k-th Chinese character in the phoneme sequence of the i-th candidate word.

[0062] Figure 3 It is a schematic diagram according to the third embodiment of the present disclosure. As Figure 3 shown, when performing S103 "obtaining the similarity between the target text and multiple candidate words according to at least one candidate phoneme sequence and the phoneme sequences of multiple candidate words" in this embodiment, the following specific steps are included:

[0063] S301: For each candidate word, select a candidate phoneme sequence and a phoneme sequence of this candidate word;

[0064] S302: When it is determined that the target text and this candidate word have different length values, take the one with the smaller length value as the first entity, and the other as the second entity;

[0065] S303: Slide the first entity on the second entity, and obtain the corresponding similarity at each slide according to the phoneme sequences of the two entities at each slide;

[0066] S304: Obtain the second similarity between the target text and this candidate word according to the corresponding similarity at each slide.

[0067] That is to say, due to various reasons in the actual scenario resulting in the problem that the lengths of the target text and the candidate word are not equal, therefore, in this embodiment, when the target text and the candidate word are non-equal-length sequences, the first entity and the second entity are distinguished according to the length size, and then by sliding the first entity on the second entity, the second similarity between the target text and the candidate word is obtained, thereby improving the accuracy when calculating the similarity of non-equal-length sequences.

[0068] It can be understood that when performing S303 in this embodiment to obtain the similarity corresponding to each slide according to the phoneme sequences of the two entities at each slide, since the two entities have the same length at each slide, therefore, this embodiment can use the method for calculating the similarity of equal-length sequences to obtain the similarity corresponding to each slide.

[0069] When obtaining the second similarity between the target text and the candidate word according to the similarity corresponding to each sliding in S304 in this embodiment, the average value of the similarities corresponding to each sliding can be used as the second similarity between the target text and the candidate word, or the maximum value of the similarities corresponding to each sliding can be used as the second similarity between the target text and the candidate word.

[0070] Figure 4 It is a schematic diagram according to the fourth embodiment of the present disclosure. As Figure 4 shown, when performing S103 "obtaining the similarity between the target text and multiple candidate words according to at least one candidate phoneme sequence and the phoneme sequences of multiple candidate words" in this embodiment, the following steps are specifically included:

[0071] S401. For each candidate word, select a candidate phoneme sequence and a phoneme sequence of the candidate word;

[0072] S402. Sequentially match the phoneme subsequences corresponding to each Chinese character in the candidate phoneme sequence with the phoneme subsequences corresponding to each Chinese character in the phoneme sequence of the candidate word;

[0073] S403. When it is determined that there is a phoneme subsequence in the phoneme sequence of the candidate word that completely matches the phoneme subsequence of the current Chinese character in the candidate phoneme sequence, use the number of phonemes in the phoneme subsequence as the counting result of the current Chinese character;

[0074] S404. According to the counting result of each Chinese character in the target text and the length value of the target text, obtain the third similarity between the target text and the candidate word.

[0075] That is to say, in this embodiment, the similarity is calculated by sliding at the Chinese character granularity, so that the third similarity between the target text and the candidate word can be obtained whether the target text and the candidate word are of equal length, further improving the calculation efficiency and calculation accuracy of the similarity.

[0076] It can be understood that after completing the matching of the current Chinese character in the target text in S402 in this embodiment, the starting point for the Chinese character after the current Chinese character in the target text to match the Chinese character in the candidate word is specifically the Chinese character after the Chinese character that matches the current Chinese character in the candidate word; if the matching of the current Chinese character in the target text cannot be completed in S402 in this embodiment, the starting point for the Chinese character after the current Chinese character in the target text to match the Chinese character in the candidate word is specifically the first Chinese character in the candidate word.

[0077] When implementing S404 to obtain the third similarity between the target text and the candidate word according to the count result of each Chinese character in the target text and the length value of the target text, the division result between the sum of the count results of each Chinese character and the length value of the target text can be used as the third similarity between the two.

[0078] It can be understood that since the method for calculating similarity provided in this embodiment can be applied to the scenario where the target text and the candidate word are non-equal-length sequences, this embodiment can also adopt the above method for calculating the similarity of non-equal-length sequences, and obtain the second similarity through the sequence sliding method. Then, the maximum value of the second similarity obtained through sequence sliding and the third similarity obtained through character sliding in this embodiment is used as the similarity between the target text and the candidate word.

[0079] Figure 5 It is a schematic diagram according to the fifth embodiment of the present disclosure. Figure 5 It shows a schematic diagram for calculating the first similarity between the target text and the candidate word when the target text and the candidate word are equal-length sequences in this embodiment: Figure 5 In the left side of, "Xiaoming" is the target text. If the candidate phoneme sequence is "xi ao, m ing", and on the right side, "Daomiao" is the candidate word. If the phoneme sequence is "d ao, m i ao"; for the first Chinese character "Xiao" in the target text, "ao" in its phoneme subsequence "x i ao" appears in the phoneme subsequence "d ao" of the first Chinese character "Dao" in the candidate word; for the second Chinese character "Ming" in the target text, "m" in its phoneme subsequence "m ing" appears in the phoneme subsequence "m iao" of the second Chinese character "Miao" in the candidate word. Then the total number of occurrences is 2. Since the length values of both the candidate phoneme sequence and the phoneme sequence are 5, the first similarity between the target text and the candidate word is "2 / 5 (0.4)".

[0080] Figure 6 It is a schematic diagram according to the sixth embodiment of the present disclosure. Figure 6 It shows a schematic diagram for calculating the second similarity between the target text and the candidate word when the target text and the candidate word are non-equal-length sequences in this embodiment: Figure 6 In the left side of, "Xiaomingming" is the target text, and "Daomiao" is the candidate word. Then, the candidate word "Daomiao" with a smaller length value is used as the first entity, and "Xiaomingming" is used as the second entity. Thus, "Daomiao" is slid from the beginning to the end on "Xiaomingming", and the similarities between the equal-length sequences "Xiaoming" and "Daomiao" (0.4) and between the equal-length sequences "Mingming" and "Daomiao" (0.2) are calculated in turn. Then, the second similarity between the target text and the candidate word is obtained based on the two calculated similarities.

[0081] Figure 7 is a schematic diagram according to a seventh embodiment of the present disclosure. Figure 7 A schematic diagram showing the calculation of the third similarity between the target text and the candidate word in this embodiment when the target text and the candidate word are sequences of equal length: Figure 7 The target text is "Xiao Mingming's room" in the upper middle. If the candidate phoneme sequence is "xi ao, m ing, m ing, de, f ang, j ian", Figure 7 For the word “小明明房间” in the middle and lower part, if the phoneme sequence is “xi ao, m ing, m ing, f ang, ji an”; for the first Chinese character in the target text, take the phoneme subsequence “xi ao” of the character “小” and match it with the phoneme subsequence of the first Chinese character “小” below. If the phoneme subsequences are completely matched, “3” is used as the counting result of the character “小” in the target text. If they do not match, match it with the phoneme subsequence of the Chinese character after the character “小” below until a completely matching phoneme subsequence is found or not found at all, such as the character “的” above; and then according to the counting result of each Chinese character in the target text and the length value of the target text, the third similarity between the target text and the candidate word is obtained.

[0082] Figure 8 is a schematic diagram according to the eighth embodiment of the present disclosure. Figure 8 As shown, the text error correction device 800 of this embodiment includes:

[0083] An acquisition unit 801 is used to acquire a text to be processed and determine a target text in the text to be processed;

[0084] The processing unit 802 is configured to obtain at least one candidate phoneme sequence according to the phoneme dictionary and the original phoneme sequence of the target text;

[0085] A selection unit 803 is configured to select a target word from the plurality of candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the plurality of candidate words;

[0086] The error correction unit 804 is used to obtain an error correction result of the text to be processed according to the target word.

[0087] When acquiring the text to be processed, the acquisition unit 801 can use the text input by the input end as the text to be processed, or use the text selected by the input end from the network as the text to be processed, or recognize the speech input by the input end and use the recognized text as the text to be processed.

[0088] After the obtaining unit 801 obtains the text to be processed, it performs the step of determining the target text in the obtained text to be processed. The target text determined by the obtaining unit 801 may be the text to be processed itself, or may be a part of the text to be processed, such as keywords or entity words in the text to be processed.

[0089] When the obtaining unit 801 determines the target text in the text to be processed, an optional implementation method that can be adopted is: input the text to be processed into an entity extraction model, and use the output result of the entity extraction model as the target text in the text to be processed. That is, the obtaining unit 801 uses the entity extraction model and takes the entity words extracted by the entity extraction model from the text to be processed as the target text.

[0090] In this embodiment, after the obtaining unit 801 determines the target text in the text to be processed, the processing unit 802 obtains at least one candidate phoneme sequence corresponding to the target text according to the phoneme dictionary and the original phoneme sequence of the determined target text.

[0091] Among them, the phoneme dictionary used by the processing unit 802 is pre-set. The phoneme dictionary contains multiple original phonemes and the similar phonemes corresponding to each original phoneme; the number of similar phonemes corresponding to an original phoneme can be one or multiple.

[0092] The original phoneme sequence or candidate phoneme sequence in this embodiment is composed of at least one phoneme subsequence, and each phoneme subsequence corresponds to a Chinese character; the phoneme subsequence is composed of at least one phoneme, and the phoneme represents the smallest unit of pronunciation; in order to simplify the processing process, the phonemes in this embodiment can be phonemes without tones.

[0093] The processing unit 802 can obtain the phoneme subsequence corresponding to each character in the target text, and then obtain the original phoneme sequence of the target text according to the obtained phoneme subsequences of each character; the original phoneme sequence obtained by the processing unit 802 can be one or multiple.

[0094] When the processing unit 802 obtains at least one candidate phoneme sequence according to the phoneme dictionary and the original phoneme sequence of the target text, an optional implementation method that can be adopted is: for each phoneme in the original phoneme sequence, obtain the similar phoneme corresponding to the phoneme in the phoneme dictionary; after replacing the phoneme in the original phoneme sequence with its corresponding similar phoneme, use the phoneme replacement result and the original phoneme sequence as at least one candidate phoneme sequence.

[0095] That is to say, since the provided phoneme dictionary contains different phonemes and their corresponding similar phonemes, the processing unit 802 can achieve the purpose of obtaining all possible phoneme sequences of the target text based on the original phoneme sequence of the target text, thereby ensuring that the obtained candidate phoneme sequences correspond to different pronunciations of the target text and improving the comprehensiveness of the obtained candidate phoneme sequences.

[0096] It can be understood that if the processing unit 802 fails to obtain similar phonemes, the original phoneme sequence can be directly used as the candidate phoneme sequence.

[0097] After the processing unit 802 obtains at least one candidate phoneme sequence in this embodiment, the selection unit 803 selects a target word from multiple candidate words according to the obtained at least one candidate phoneme sequence and the phoneme sequences of multiple candidate words.

[0098] Among them, the multiple candidate words used by the selection unit 803 can be preset at the input end, that is, the input end can customize the candidate words according to actual needs, so as to realize the personalization of text error correction.

[0099] In some technical fields, such as finance, technology, medicine and other technical fields, there are a large number of professional words, and it is difficult to correct errors in professional words. Therefore, the selection unit 803 may further include the following: determining the technical field to which the target text belongs; obtaining multiple words corresponding to the determined technical field as multiple candidate words, and the selection unit 803 can obtain multiple candidate words through the word libraries corresponding to different technical fields.

[0100] In order to further improve the error correction efficiency, the phoneme sequences of the candidate words used by the selection unit 803 can be preset; when the selection unit 803 presets the phoneme sequences of the candidate words, only the case of polyphonic characters needs to be considered.

[0101] The selection unit 803 can also use the above method of obtaining the candidate phoneme sequences of the target text, and use the obtained at least one candidate phoneme sequence as the phoneme sequence of the candidate word according to the original phoneme sequence of the candidate word. The specific process will not be elaborated here.

[0102] Specifically, when the selection unit 803 selects a target word from multiple candidate words according to the obtained at least one candidate phoneme sequence and the phoneme sequences of multiple candidate words, the optional implementation method that can be adopted is: obtaining the similarity between the target text and multiple candidate words according to the obtained at least one candidate phoneme sequence and the phoneme sequences of multiple candidate words; using the candidate word corresponding to the maximum value of the obtained similarity as the target word.

[0103] When obtaining the similarity between the target text and multiple candidate words based on at least one candidate phoneme sequence and the phoneme sequences of multiple candidate words, the selection unit 803 may adopt the following method: for each candidate word, select a candidate phoneme sequence and a phoneme sequence of the candidate word; when it is determined that the target text and the candidate word have the same length value, count the number of occurrences of the phonemes included in the phoneme subsequence of the Chinese character at each position in the candidate phoneme sequence in the phoneme subsequence of the Chinese character at the corresponding position in the phoneme sequence of the candidate word; according to the counted number of occurrences and the target length value, obtain the first similarity between the target text and the candidate word.

[0104] That is to say, based on the characteristics of Chinese pinyin, when the target text and the candidate word are sequences of equal length, the selection unit 803 calculates the phoneme similarity at the Chinese character granularity according to the candidate phoneme sequence of the target text and the phoneme sequence of the candidate word, and obtains the first similarity between the target text and the candidate word, thereby improving the accuracy when calculating the similarity of sequences of equal length.

[0105] It can be understood that if there are multiple phoneme sequences corresponding to the current candidate word, when calculating the similarity between the target text and the current candidate word, the selection unit 803 can calculate according to the candidate phoneme sequence and each phoneme sequence, and then take the maximum value of the calculated similarities as the first similarity between the target text and the current candidate word.

[0106] Among them, the target length value used by the selection unit 803 is specifically the length value of the shorter sequence among the candidate phoneme sequence and the phoneme sequence.

[0107] When obtaining the similarity between the target text and multiple candidate words based on at least one candidate phoneme sequence and the phoneme sequences of multiple candidate words, the selection unit 803 may also adopt the following method: for each candidate word, select a candidate phoneme sequence and a phoneme sequence of the candidate word; when it is determined that the target text and the candidate word have different length values, take the one with the smaller length value as the first entity and the other as the second entity; slide the first entity on the second entity, and obtain the corresponding similarity at each slide according to the phoneme sequences of the two entities at each slide; according to the corresponding similarity at each slide, obtain the second similarity between the target text and the candidate word.

[0108] That is to say, due to various reasons in the actual scenario that cause the problem of unequal lengths between the target text and the candidate word, when the selection unit 803 has a non-equal-length sequence of the target text and the candidate word, it distinguishes the first entity and the second entity according to the length size, and then obtains the second similarity between the target text and the candidate word by sliding the first entity on the second entity, thereby improving the accuracy when calculating the similarity of non-equal-length sequences.

[0109] It can be understood that when the selection unit 803 obtains the similarity corresponding to each sliding according to the phoneme sequences corresponding to the two entities at each sliding, since the two entities have the same length at each sliding, this embodiment can use the method for calculating the similarity of equal-length sequences to obtain the similarity corresponding to each sliding.

[0110] When the selection unit 803 obtains the second similarity between the target text and the candidate word according to the similarity corresponding to each sliding, it can use the average value of the similarities corresponding to each sliding as the second similarity between the target text and the candidate word, or use the maximum value of the similarities corresponding to each sliding as the second similarity between the target text and the candidate word.

[0111] When the selection unit 803 obtains the similarity between the target text and multiple candidate words according to at least one candidate phoneme sequence and the phoneme sequences of multiple candidate words, the following method can also be adopted: for each candidate word, select a candidate phoneme sequence and a phoneme sequence of the candidate word; sequentially match the phoneme subsequences corresponding to each Chinese character in the candidate phoneme sequence with the phoneme subsequences corresponding to each Chinese character in the phoneme sequence of the candidate word; in the case where it is determined that there is a phoneme subsequence in the phoneme sequence of the candidate word that completely matches the phoneme subsequence of the current Chinese character in the candidate phoneme sequence, use the number of phonemes in the phoneme subsequence as the counting result of the current Chinese character; obtain the third similarity between the target text and the candidate word according to the counting result of each Chinese character in the target text and the length value of the target text.

[0112] That is to say, the selection unit 803 calculates the similarity by sliding at the Chinese character granularity, so as to obtain the third similarity between the target text and the candidate word whether the target text and the candidate word are of equal length, further improving the calculation efficiency and calculation accuracy of the similarity.

[0113] It can be understood that after the selection unit 803 completes the matching of the current Chinese character in the target text, for the Chinese character following the current Chinese character in the target text, the starting point when matching with the Chinese characters in the candidate word is specifically the Chinese character following the Chinese character that matches the current Chinese character in the candidate word; when the selection unit 803 fails to complete the matching of the current Chinese character in the target text, for the Chinese character following the current Chinese character in the target text, the starting point when matching with the Chinese characters in the candidate word is specifically the first Chinese character in the candidate word.

[0114] When the selection unit 803 obtains the third similarity between the target text and the candidate word according to the counting result of each Chinese character in the target text and the length value of the target text, the division result between the sum of the counting results of each Chinese character and the length value of the target text can be used as the third similarity between the two.

[0115] It can be understood that since the method for calculating the similarity provided by the selection unit 803 can be applied to the scenario where the target text and the candidate word are non-equal-length sequences, the selection unit 803 can also adopt the above method for calculating the similarity of non-equal-length sequences, and obtain the second similarity by means of sequence sliding, and then take the maximum value of the second similarity obtained by sequence sliding and the third similarity obtained by character sliding in this embodiment as the similarity between the target text and the candidate word.

[0116] Since a candidate word may also have multiple phoneme sequences, the selection unit 803 can respectively select one from at least one candidate phoneme sequence and multiple phoneme sequences of a candidate word, and then calculate the similarity multiple times according to the candidate phoneme sequence and the phoneme sequence selected each time to obtain the similarity between the target text and the candidate word.

[0117] In this embodiment, after the selection unit 803 selects the target word from multiple candidate words, the error correction unit 804 obtains the error correction result of the text to be processed according to the selected target word.

[0118] The error correction unit 804 can replace the target text in the text to be processed with the selected target word, and then take the replacement result as the error correction result of the text to be processed; the error correction unit 804 can also directly output the selected target word as the error correction result of the text to be processed.

[0119] That is to say, the error correction unit 804 replaces the target text in the text to be processed with the selected target word, realizes the purpose of determining the target word among multiple candidate words through the similarity of sounds, and can improve the efficiency and accuracy of text error correction.

[0120] In the technical solution of the present disclosure, the acquisition, storage, and application of the user's personal information involved all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0121] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0122] As Figure 9 shown, it is a block diagram of an electronic device for a text error correction method according to an embodiment of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as, for example, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, for example, a personal digital processor, a cellular phone, a smart phone, a wearable device, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely exemplary and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0123] As Figure 9 shown, the device 900 includes a computing unit 901, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 902 or a computer program loaded from a storage unit 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the device 900 can also be stored. The computing unit 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0124] A plurality of components in the device 900 are connected to the I / O interface 905, including: an input unit 906, such as a keyboard, a mouse, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a magnetic disk, an optical disc, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the device 900 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0125] The computing unit 901 can be various general-purpose and / or special-purpose processing components having processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 901 executes the various methods and processes described above, such as the text error correction method. For example, in some embodiments, the text error correction method can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 908.

[0126] In some embodiments, part or all of the computer program may be loaded and / or installed onto the device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into the RAM 903 and executed by the computing unit 901, one or more steps of the text correction method described above may be performed. Alternatively, in other embodiments, the computing unit 901 may be configured to perform the text correction method by any other suitable means (e.g., by means of firmware).

[0127] The various implementations of the systems and techniques described herein may be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), systems on a chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implemented in one or more computer programs that may be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a special-purpose or general-purpose programmable processor, that receives data and instructions from, and transmits data and instructions to, a storage system, at least one input device, and at least one output device.

[0128] The program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that the program codes, when executed by the processor or controller, cause the functions / operations specified in the flowchart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine, or entirely on the remote machine or server.

[0129] In the context of this disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0130] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, speech input, or tactile input).

[0131] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0132] A computer system may include a client and a server. The client and the server are generally far away from each other and usually interact via a communication network. The relationship between the client and the server is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services (“Virtual Private Server”, or simply “VPS”). The server may also be a server of a distributed system or a server combined with a blockchain.

[0133] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and no limitations are imposed herein.

[0134] The above specific embodiments do not constitute a limitation on the protection scope of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the protection scope of this disclosure.

Claims

1. A text error correction method, comprising: Obtaining a text to be processed and determining a target text in the text to be processed; Obtaining at least one candidate phoneme sequence according to a phoneme dictionary and an original phoneme sequence of the target text; Selecting a target word from the plurality of candidate words according to the at least one candidate phoneme sequence and phoneme sequences of the plurality of candidate words; Obtaining an error correction result of the text to be processed according to the target word; Wherein, the selecting a target word from the plurality of candidate words according to the at least one candidate phoneme sequence and phoneme sequences of the plurality of candidate words includes: Obtaining a similarity between the target text and the plurality of candidate words according to the at least one candidate phoneme sequence and phoneme sequences of the plurality of candidate words; Taking the candidate word corresponding to the maximum value of the similarity as the target word; Wherein, the obtaining a similarity between the target text and the plurality of candidate words according to the at least one candidate phoneme sequence and phoneme sequences of the plurality of candidate words includes: For each candidate word, selecting a candidate phoneme sequence and a phoneme sequence of the candidate word; Sequentially matching the phoneme subsequences corresponding to each Chinese character in the candidate phoneme sequence with the phoneme subsequences corresponding to each Chinese character in the phoneme sequence of the candidate word; When it is determined that there is a phoneme subsequence in the phoneme sequence of the candidate word that completely matches the phoneme subsequence of the current Chinese character in the candidate phoneme sequence, taking the number of phonemes in the phoneme subsequence as the counting result of the current Chinese character; Obtaining a third similarity between the target text and the candidate word according to the counting result of each Chinese character in the target text and the length value of the target text.

2. The method according to claim 1, wherein, The obtaining at least one candidate phoneme sequence according to a phoneme dictionary and an original phoneme sequence of the target text includes: For each phoneme in the original phoneme sequence, obtaining a similar phoneme corresponding to the phoneme in the phoneme dictionary; After replacing the phonemes in the original phoneme sequence with their corresponding similar phonemes, taking the phoneme replacement result and the original phoneme sequence as at least one candidate phoneme sequence.

3. The method according to any one of claims 1-2, further comprising Determining the technical field to which the target text belongs; Obtaining a plurality of words corresponding to the technical field as the plurality of candidate words.

4. The method according to claim 1, wherein, The obtaining a similarity between the target text and the plurality of candidate words according to the at least one candidate phoneme sequence and phoneme sequences of the plurality of candidate words includes: For each candidate word, selecting a candidate phoneme sequence and a phoneme sequence of the candidate word; When it is determined that the target text and the candidate word have the same length value, counting the number of occurrences of the phonemes included in the phoneme subsequence of the Chinese character at each position in the candidate phoneme sequence in the phoneme subsequence of the Chinese character at the corresponding position in the phoneme sequence of the candidate word; Obtaining a first similarity between the target text and the candidate word according to the counted number of occurrences and the target length value.

5. The method according to claim 1, wherein, The obtaining a similarity between the target text and the plurality of candidate words according to the at least one candidate phoneme sequence and phoneme sequences of the plurality of candidate words includes: For each candidate word, select a candidate phoneme sequence and a phoneme sequence of the candidate word; When it is determined that the target text and the candidate word have different length values, take the one with the smaller length value as the first entity and the other as the second entity; Slide the first entity on the second entity, and obtain the corresponding similarity for each slide according to the phoneme sequences of the two entities at each slide; Obtain the second similarity between the target text and the candidate word according to the similarity corresponding to each slide.

6. A text error correction device, comprising: An acquisition unit, configured to acquire a text to be processed and determine a target text in the text to be processed; A processing unit, configured to obtain at least one candidate phoneme sequence according to a phoneme dictionary and an original phoneme sequence of the target text; A selection unit, configured to select a target word from the multiple candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the multiple candidate words; An error correction unit, configured to obtain an error correction result of the text to be processed according to the target word; Wherein, when the selection unit selects a target word from the multiple candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the multiple candidate words, it specifically performs: Obtain the similarity between the target text and the multiple candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the multiple candidate words; Take the candidate word corresponding to the maximum value of the similarity as the target word; Wherein, when the selection unit obtains the similarity between the target text and the multiple candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the multiple candidate words, it specifically performs: For each candidate word, select a candidate phoneme sequence and a phoneme sequence of the candidate word; Sequentially match the phoneme subsequences corresponding to each Chinese character in the candidate phoneme sequence with the phoneme subsequences corresponding to each Chinese character in the phoneme sequence of the candidate word; When it is determined that there is a phoneme subsequence in the phoneme sequence of the candidate word that completely matches the phoneme subsequence of the current Chinese character in the candidate phoneme sequence, take the number of phonemes in the phoneme subsequence as the counting result of the current Chinese character; Obtain the third similarity between the target text and the candidate word according to the counting result of each Chinese character in the target text and the length value of the target text.

7. The apparatus according to claim 6, wherein, When the processing unit obtains at least one candidate phoneme sequence according to the phoneme dictionary and the original phoneme sequence of the target text, it specifically performs: For each phoneme in the original phoneme sequence, obtain the similar phoneme corresponding to the phoneme in the phoneme dictionary; After replacing the phonemes in the original phoneme sequence with their corresponding similar phonemes, take the phoneme replacement result and the original phoneme sequence as at least one candidate phoneme sequence.

8. The device according to any one of claims 6-7, wherein the selection unit is further configured to perform: Determine the technical field to which the target text belongs; Obtain multiple words corresponding to the technical field as the multiple candidate words.

9. The apparatus according to claim 6, wherein, When obtaining the similarity between the target text and the multiple candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the multiple candidate words, the selection unit specifically performs the following: For each candidate word, select a candidate phoneme sequence and a phoneme sequence of this candidate word; When it is determined that the target text and this candidate word have the same length value, count the number of occurrences of the phonemes included in the phoneme subsequence of the Chinese character at each position in the candidate phoneme sequence in the phoneme subsequence of the Chinese character at the corresponding position in the phoneme sequence of this candidate word; According to the counted number of occurrences and the target length value, obtain the first similarity between the target text and this candidate word.

10. The apparatus according to claim 6, wherein, When obtaining the similarity between the target text and the multiple candidate words according to the at least one candidate phoneme sequence and the phoneme sequences of the multiple candidate words, the selection unit specifically performs the following: For each candidate word, select a candidate phoneme sequence and a phoneme sequence of this candidate word; When it is determined that the target text and this candidate word have different length values, take the one with the smaller length value as the first entity and the other as the second entity; Slide the first entity on the second entity, and according to the phoneme sequences of the two entities during each slide, obtain the corresponding similarity during each slide; According to the corresponding similarity during each slide, obtain the second similarity between the target text and this candidate word.

11. An electronic device, comprising: At least one processor; And A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor is enabled to execute the method according to any one of claims 1-5.

12. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to execute the method according to any one of claims 1-5.

13. A computer program product, comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-5.

Images