CN114154500A

CN114154500A - Text proofreading method, apparatus, device, medium, and program product

Info

Publication number: CN114154500A
Application number: CN202111511711.1A
Authority: CN
Inventors: 艾巍
Original assignee: CCB Finetech Co Ltd
Current assignee: CCB Finetech Co Ltd
Priority date: 2021-12-06
Filing date: 2021-12-06
Publication date: 2022-03-08

Abstract

The application provides a text proofreading method, a text proofreading device, a text proofreading apparatus, a text proofreading medium and a program product, which relate to the technical field of artificial intelligence, wherein the text proofreading method comprises the following steps: acquiring a voice recognition result of the current inquiry node, acquiring a target pinyin corresponding to the character to be corrected according to the pinyin of each character in the character to be corrected, and correcting the character to be corrected according to a preset pinyin correction set and the target pinyin to obtain the corrected character. In the technical scheme, the pinyin proofreading set is configured for the characters which are easy to generate misrecognition in the current query node, and the pinyin proofreading set is configured for the characters which are easy to generate misrecognition in the business process where the current query node is located, so that the range of the characters involved in proofreading can be accurately controlled, and the text obtained by the voice recognition technology is more accurate.

Description

Text proofreading method, apparatus, device, medium, and program product

Technical Field

The present application relates to the field of artificial intelligence technologies, and in particular, to a text proofreading method, apparatus, device, medium, and program product.

Background

With the development of artificial intelligence technology, the intelligent customer service system is applied, the intelligent customer service system obtains the spoken voice information of the user through voice communication with the user, and analyzes the real intention of the user from the voice information, so that the user can be answered by selecting corresponding dialect according to the real intention of the user, and the labor cost can be reduced.

In the prior art, mainly through an automatic speech recognition technology, speech information spoken by a user is recognized and converted into text information, and then keywords are extracted from the text to obtain the real intention of the user.

However, in the automatic speech recognition technology of the prior art, word confusion is easily generated in the speech recognition process, so that the recognized text information is inaccurate.

Disclosure of Invention

The application provides a text proofreading method, a text proofreading device, text proofreading equipment, a text proofreading medium and a program product, which are used for solving the problem that texts obtained by automatic speech recognition technology are inaccurate.

In a first aspect, an embodiment of the present application provides a text proofreading method, including:

acquiring a voice recognition result of a current query node, wherein the voice recognition result comprises characters to be corrected, which are obtained by recognizing voice input at the current query node;

acquiring target pinyin corresponding to the characters to be corrected according to the pinyin of each character in the characters to be corrected;

according to a preset pinyin proofreading set and the target pinyin, proofreading the characters to be proofread to obtain proofread characters; the pinyin proofreading set comprises a first pinyin proofreading set or a second pinyin proofreading set, the first pinyin proofreading set comprises corrected pinyin and corrected characters for proofreading characters contained in a business process where the current query node is located, the second pinyin proofreading set comprises corrected pinyin and corrected characters for proofreading the characters contained in the current query node, and the corrected characters in the first pinyin proofreading set are different from the corrected characters in the second pinyin proofreading set.

In a possible design of the first aspect, the proofreading the character to be proofread according to the pre-configured pinyin proofreading set and the target pinyin to obtain a proofread character includes:

acquiring a business process where the current inquiry node is located, and determining whether the business process is configured with the first pinyin proofreading set;

if the business process is configured with the first pinyin proofreading set, proofreading the characters to be proofread according to the first pinyin proofreading set and the target pinyin;

if the first pinyin proofreading set is not configured in the business process, acquiring a second pinyin proofreading set configured in the current inquiry node;

and correcting the characters to be corrected according to the second pinyin correction set and the target pinyin to obtain corrected characters.

In another possible design of the first aspect, the proofreading the text to be proofread according to the first pinyin proofreading set and the target pinyin includes:

according to the target pinyin and the corrected pinyins in the first pinyin proofreading sets, determining a target first pinyin proofreading set matched with the target pinyin in each first pinyin proofreading set;

and correcting the characters to be corrected into the corrected characters in the target first pinyin correction set.

In yet another possible design of the first aspect, the determining, in each first pinyin proofreading set, a target first pinyin proofreading set that matches the target pinyin based on the target pinyin and corrected pinyins in each first pinyin proofreading set includes:

obtaining confusion pinyin corresponding to the target pinyin, wherein the similarity values of the confusion pinyin and the target pinyin are higher than a preset threshold value;

and comparing the target pinyin and the confusing pinyin with the corrected pinyins in the first pinyin proofreading sets to determine a target first pinyin proofreading set matched with the target pinyin.

In another possible design of the first aspect, the proofreading the text to be proofread according to the second pinyin proofreading set and the target pinyin includes:

according to the target pinyin and the corrected pinyins in the second pinyin proofreading sets, determining a target second pinyin proofreading set matched with the target pinyin in each second pinyin proofreading set;

and correcting the characters to be corrected into the corrected characters in the target second pinyin correction set.

In yet another possible design of the first aspect, the determining, in each second pinyin proofreading set, a target second pinyin proofreading set that matches the target pinyin based on the target pinyin and corrected pinyins in each second pinyin proofreading set includes:

obtaining confusion pinyin corresponding to the target pinyin, wherein the similarity value of the confusion pinyin and the target pinyin is higher than a preset threshold value;

and comparing the target pinyin and the confusing pinyin with the corrected pinyins in each second pinyin proofreading set to determine a target second pinyin proofreading set matched with the target pinyin.

In yet another possible design of the first aspect, the obtaining the business process in which the current query node is located includes:

determining a previous node and a next node of the current query node according to a preset flow chart;

and determining the business process of the current query node according to the previous node and the next node of the current query node.

In another possible design of the first aspect, the obtaining a target pinyin corresponding to the text to be corrected according to the pinyin for each text in the text to be corrected includes:

deleting phonetic symbols of the pinyin of each character to obtain the phonetic symbols-free pinyin corresponding to each character;

and separating the phonetic alphabets without phonetic symbols of all the characters according to preset separators to obtain the target phonetic alphabets corresponding to the characters to be corrected.

In yet another possible design of the first aspect, the method further includes:

acquiring a first user intention matched with the characters to be corrected and a matching value corresponding to the first user intention;

acquiring a second user intention matched with the corrected characters and a matching value corresponding to the second user intention;

determining the first user intention or the second user intention as a real user intention according to the matching value corresponding to the first user intention and the matching value corresponding to the second user intention;

and obtaining the reply information corresponding to the voice input at the current query node according to the real user intention.

and writing the characters to be corrected and the corrected characters into a preset interaction record.

and generating a proofreading report according to the characters to be proofread, the pre-configured pinyin proofreading set and the proofread characters.

In a second aspect, an embodiment of the present application provides a text proofreading apparatus, including:

the result acquisition module is used for acquiring a voice recognition result of the current query node, wherein the voice recognition result comprises characters to be corrected, which are obtained by recognizing the voice input at the current query node;

the pinyin conversion module is used for acquiring the target pinyin corresponding to the characters to be corrected according to the pinyin of each character in the characters to be corrected;

the character proofreading module is used for proofreading the characters to be proofread according to a preset pinyin proofreading set and the target pinyin to obtain the proofread characters; the pinyin proofreading set comprises a first pinyin proofreading set or a second pinyin proofreading set, the first pinyin proofreading set comprises corrected pinyin and corrected characters for proofreading characters contained in a business process where the current query node is located, the second pinyin proofreading set comprises corrected pinyin and corrected characters for proofreading the characters contained in the current query node, and the corrected characters in the first pinyin proofreading set are different from the corrected characters in the second pinyin proofreading set.

In a third aspect, an embodiment of the present application provides a computer device, including: a processor, and a memory communicatively coupled to the processor;

the memory stores computer-executable instructions;

the processor executes the computer-executable instructions stored by the memory to implement the methods described above.

In a fourth aspect, the present application provides a readable storage medium, in which computer instructions are stored, and when executed by a processor, the computer instructions are used to implement the method described above.

In a fifth aspect, the present application provides a program product including computer instructions, which when executed by a processor implement the method described above.

According to the text proofreading method, the text proofreading device, the text proofreading equipment, the text proofreading medium and the text program product, the pinyin proofreading set is configured for the characters which are easy to generate the misrecognition in the current query node, and the pinyin proofreading set is configured for the characters which are easy to generate the misrecognition in the business process where the current query node is located, so that the range of the characters involved in the proofreading can be accurately controlled, and the text obtained through the voice recognition technology is more accurate.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and together with the description, serve to explain the principles of the application;

fig. 1 is a scene schematic diagram of a text proofreading method provided in an embodiment of the present application;

fig. 2 is a schematic flowchart of a first embodiment of a text proofing method according to an embodiment of the present application;

fig. 3 is a schematic structural diagram of a business process provided in an embodiment of the present application;

fig. 4 is a schematic flowchart of a second text proofreading method according to an embodiment of the present application;

FIG. 5 is a schematic diagram of data processing interaction of an intelligent customer service system according to an embodiment of the present application;

fig. 6 is a schematic structural diagram of a text proofreading apparatus according to an embodiment of the present application;

fig. 7 is a schematic structural diagram of a computer device according to an embodiment of the present application.

With the above figures, there are shown specific embodiments of the present application, which will be described in more detail below. These drawings and written description are not intended to limit the scope of the inventive concepts in any manner, but rather to illustrate the inventive concepts to those skilled in the art by reference to specific embodiments.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The terms referred to in this application are explained first:

media resource server:

the media resource server provides media resource functions required for implementing various services on the IP network under the control of a control device (a soft switch device, an application server), including service voice provision, conferencing, interactive response (IVR), notification, high-level language services, and the like.

Media resource control protocol:

media Resource Control Protocol (MRCP) is a communication Protocol of computer network application layer, and is used for providing various voice services (such as voice recognition, voice synthesis, and voiceprint recognition) to a client by a voice server.

Automatic speech recognition technology:

automatic Speech Recognition (ASR) is a technology for converting human Speech into text.

And (3) natural language processing:

natural Language Processing (NLP) is a science integrating linguistics, mathematics and computer science. The core goal of the system is to convert human natural language into computer readable instructions, which simply means to make a machine read human language.

Fig. 1 is a scene schematic diagram of a text proofreading method provided in an embodiment of the present application. As shown in fig. 1, the smart client system 10 may actively dial a user telephone number, and after the user connects a telephone call on the mobile terminal 11, the smart client system 10 may query the user, for example, to determine whether the user needs to transact an installment service, and the user may reply to the query question. The intelligent customer service system 10 needs to feed back corresponding information to the user according to the response of the user, so as to finally realize intelligent question answering and reduce manual input. In the actual life application, in order to improve the question-answer interaction effect, the intelligent customer service system 10 needs to feed back an accurate word to the user according to the response of the user. The process mainly relates to the recognition of the voice of the user, the characters are extracted, the real intention of the user is recognized according to the characters, and finally the corresponding dialect is found and fed back to the user according to the real intention of the user.

The intelligent customer service system 10 mainly includes four major modules, namely a media resource server, an intelligent voice product, an NLP and an application system. The data processing involved in a basic interaction flow of the intelligent customer service system 10 mainly includes the following aspects: (1) the voice spoken by the user is transmitted to the intelligent voice product through the media resource server, the ASR model in the intelligent voice product converts the voice of the user into characters, and the characters are returned to the media resource server; (2) and the media resource server sends the identification result to an application system, and the application system calls the NLP model to understand the user intention and then selects a reply dialect according to the user intention. Then driving the media resource server to send a speech-to-speech request to the intelligent speech product; (3) the intelligent voice product converts the dialogs into voice, transmits the voice to the media resource server and plays the voice to the user. In the data processing process, due to the reasons of homophones of Chinese characters, more confusing voices, limited ASR models and the like, the results of characters converted by the ASR models are easy to be wrong, so that the whole intelligent customer service system is possibly influenced, and the question-answer interaction effect is not ideal.

In the prior art, in the process of converting user speech into characters through an ASR model, two modes are mainly involved, one mode is to divide the ASR model into a basic model and a specific model, the basic model is suitable for the field with weak specialty, and the recognition accuracy is slightly low due to wide range of consideration. The specific model is a specially trained model aiming at the fields with strong speciality, such as the financial field, the insurance field and the like, and has the advantage of high recognition rate in the specific field. However, a large amount of tuning training work needs to be performed on a specific model, such as corpus training set collection, algorithm adjustment training parameter adjustment and the like, the whole process is time-consuming and labor-consuming, and the cost is high. And the application range of a specific model is narrow, and the specific model cannot be adjusted and adapted quickly when new services appear. And the other method is to utilize a third-party self-learning training platform to add hot keywords, a customized language model, an acoustic model and the like, and the user performs self-training and tuning to solve some simple recognition problems. However, in this way, the self-learning training platform is expensive to use; moreover, since such tuning (e.g. hot keywords) is likely to cause overkill, the correct recognition result of the old model is recognized as the configured hot keywords with similar pronunciation; meanwhile, the tuning is coarse-grained, cannot control the application range (such as a process and a process node), and has a large influence range.

In view of the above problems, the text proofreading method, apparatus, device, medium, and program product provided in the embodiments of the present application further correct the ASR recognition result by adding the pinyin proofreading set based on the pinyin recognition error correction based on the specific business process and the specific process node, and finally can improve the accuracy of converting speech into text, perform preprocessing on subsequent NLPs, and improve the interaction effect of the entire intelligent customer service system.

The technical solution of the present application will be described in detail below with reference to specific examples. It should be noted that the following specific embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

Fig. 2 is a schematic flow chart of a text proofreading method according to an embodiment of the present application. The method can be applied to an intelligent customer service system, and words in dialogs are further corrected after the intelligent voice products convert voices into dialogs. As shown in fig. 2, the method may specifically include the following steps:

s201, obtaining a voice recognition result of the current inquiry node.

The voice recognition result comprises characters to be corrected, which are obtained by recognizing the voice input at the current inquiry node. The speech may be speech spoken by the user, such as speech generated by the intelligent customer service system by dialing a telephone call from the user, the user placing the telephone call and answering a question from the intelligent customer service system.

For example, the speech recognition result may be obtained by recognizing the user speech through an ASR model, wherein the speech includes text. For example, the intelligent customer service system asks the user whether the user needs to stage, and the user says "needs to stage", then the words recognized by the ASR model for the user's speech may be "needs to stage" or "needs to diverge". The latter is the character recognition result generated by the ASR for the user speech misrecognition because the divergence and the stage are homophones, and the speech recognition result of the "needing divergence" needs to be corrected to be the former by using the correction method of the speech recognition of the present application.

In this embodiment, according to different fields related to the intelligent customer service system, the business process is different, and the corresponding query node is different, for example, if the intelligent customer service system relates to a financial field, the business process may include a loan service, a repayment service, and the like. Illustratively, the loan service may include query nodes that query whether an installment node, a query end node, and so on.

Exemplarily, fig. 3 is a schematic structural diagram of a service flow provided in the embodiment of the present application, as shown in fig. 3, where a plurality of nodes are included, such as whether an installment node, an end node, and the like. Each node has a corresponding grammar and a set of intentions that the user may express, for example, whether to stage the node, the grammar may be to ask the user "ask whether to handle the stage", and the set of intentions includes "positive intentions" and "negative intentions". The business process is a structure in which nodes are used as units and are connected in series through arrow lines.

In some embodiments, the flow structure of the business process in fig. 3 may be used as a preset flow table, so as to determine the previous node and the next node of the current query node, and thus determine the business process in which the current query node is located.

Illustratively, the business processes may include a staging business process (i.e., the business process in FIG. 3) and a customer complaint process business process. Different business processes have different first pinyin proofreading sets.

S202, obtaining target pinyin corresponding to the characters to be corrected according to the pinyin of each character in the characters to be corrected.

In this embodiment, the characters to be corrected at least include two Chinese characters, such as "need to diverge", where each Chinese character has a corresponding pinyin, and the target pinyin, such as "xuyaofenqi", is obtained by concatenating the pinyins of each Chinese character.

S203, according to the preset pinyin proofreading set and the target pinyin, proofreading characters to be proofread to obtain the proofread characters.

The pinyin proofreading set comprises a first pinyin proofreading set or a second pinyin proofreading set, the first pinyin proofreading set comprises corrected pinyin and corrected characters for proofreading characters contained in a business process where the current query node is located, the second pinyin proofreading set comprises corrected pinyin and corrected characters for proofreading characters contained in the current query node, and the corrected characters in the first pinyin proofreading set are different from the corrected characters in the second pinyin proofreading set.

In this embodiment, the pinyin proof set includes the pinyin and the characters corresponding to the pinyin, and the pinyin may be "fenqi", and the corresponding characters are "installments". If the character to be corrected is 'need to be diverged' and the target pinyin is 'xuyaofenqi', the character to be corrected is corrected into stages and used as the corrected character.

In this embodiment, the number of the first pinyin proof set and the number of the second pinyin proof set may be multiple, for example, there is a node where the ASR model has a high recognition error rate, and many characters obtained by the user speech recognition in the node are all wrong, for example, the ASR model recognizes "three-phase" spoken by the user as "notoginseng" and recognizes "nine-phase" as "wine set", and at this time, the node needs two second pinyin proof sets to respectively proof "notoginseng" and "wine set".

Besides the high recognition error rate of the ASR model at a certain node, the recognition error rate of the ASR model to some speech in the whole business process may also be high, and at this time, a plurality of first pinyin proofreading sets are required to respectively proofread the speech which is easy to generate misrecognition in the business process.

In this embodiment, the first pinyin proofreading set is used for proofreading voices which are easy to generate misrecognition in the whole business process, and the second pinyin proofreading set is used for proofreading voices which are easy to generate misrecognition in a single node.

Illustratively, for the business process and the node, the pinyin proofreading sets configured respectively are as follows table 1:

TABLE 1

According to the method and the device, the corresponding pinyin proofreading set is configured in the current query node or the service process where the current query node is located, so that characters which are easy to be misrecognized in the current query node or the service process can be accurately proofread, and the text obtained after ASR recognition can more accurately describe the voice input by the user.

In some embodiments, the step S203 may be specifically implemented by the following steps:

acquiring a business process where a current inquiry node is located, and determining whether the business process is configured with a first pinyin proofreading set or not;

if the business process is configured with a first pinyin proofreading set, proofreading characters to be proofread according to the first pinyin proofreading set and the target pinyin;

and checking the characters to be checked according to the second pinyin checking set and the target pinyin.

In this embodiment, the targets corrected by the first pinyin correction set and the second pinyin correction set are different, and the pinyin and characters contained in the first pinyin correction set mainly face the chinese characters that may be erroneously identified in the entire business process. The pinyin and characters contained in the second pinyin proofreading set mainly face Chinese characters which may be mistakenly identified at a specific node.

In the present query node, a plurality of first pinyin proofreading sets or a plurality of second pinyin proofreading sets may be configured, for example, as shown in table 1 above, three second pinyin proofreading sets < sanqi, third generation >, < xinlong, newlong >, < xinlong, and newlong > may be set in the present query node. If the target pinyin is sanqi, the characters to be corrected can be corrected to obtain corrected characters in the third stage.

In this embodiment, the first pinyin proof set or the second pinyin proof set is configured to avoid overusing, for example, the ASR model recognizes that the text obtained by the user speech is "divergent", and if the current query node is in the user complaint processing business process, the current query node should not be corrected to "instanced", and the ASR model recognition result should be considered to be correct.

According to the embodiment of the application, the first pinyin proofreading set or the second pinyin proofreading set is configured on the current query node, words or characters in different ranges can be proofread, the proofreading granularity is accurately controlled, and the targeted pinyin proofreading set is configured according to limited answer contents which may appear in a limited range of a specific service and a user, so that the situation that the words or characters are overruled is avoided.

On the basis of the above embodiments, in some embodiments, the "proofreading the character to be proofread according to the first pinyin proofreading set and the target pinyin" may be specifically implemented by the following steps:

In this embodiment, the corrected pinyin and the corrected text in the first pinyin correction set may be in the form of key-value pairs, i.e., < key, value >, where key represents the corrected pinyin and value represents the corrected text. For example, the number of the first pinyin proofreading set may be one, for example, < fenqi, installment >, and if the target pinyin includes fenqi or the target pinyin is fenqi, the text to be proofread may be proofread as "installment".

By setting the proofreading pinyin and the proofreading characters, the embodiment of the application can proofread the misrecognized characters output by the ASR model, obtain more accurate texts, provide powerful support for the subsequent work of the intelligent customer service system, and improve the interaction effect of the intelligent customer service system and a user.

Further, in some embodiments, the step "determining a target first pinyin proofreading set matching the target pinyin in each first pinyin proofreading set according to the target pinyin and the corrected pinyins in each first pinyin proofreading set" may be specifically implemented by the following steps:

obtaining confusion pinyin corresponding to the target pinyin;

And the similarity value of the confusing pinyin and the target pinyin is higher than a preset threshold value. For example, taking the target pinyin as fenqi as an example, the confusing pinyin may be a rear nasal sound that is easily confused with the target pinyin, for example, the confusing pinyin may be fengqi.

Illustratively, the first pinyin proofreading set may include two, such as < fengqi, installments > and < fenqi, installments >, i.e. the target first pinyin proofreading set matching the target pinyin is < fenqi, installments >, and the target first pinyin proofreading set matching the confusing pinyin is < fengqi, installments >, and the final corrected text is "installments" through the two target first pinyin proofreading sets.

In the embodiment, by acquiring the confusion tone of the target pinyin and configuring the pinyin proofreading set matched with the confusion tone in the first pinyin proofreading set, the wrongly recognized characters of the ASR model caused by inaccurate pronunciation can be proofread into correct characters aiming at the problem of inaccurate pronunciation of the user, so that the recognition accuracy is improved.

In some embodiments, the step of "proofreading the character to be proofread according to the second pinyin proofreading set and the target pinyin" may be specifically implemented by the following steps:

Further, in some embodiments, the "determining a target second pinyin proofreading set matching the target pinyin in each second pinyin proofreading set according to the target pinyin and the corrected pinyins in each second pinyin proofreading set" may specifically be implemented by the following steps:

obtaining confusion pinyin corresponding to the target pinyin;

And the similarity value of the confusing pinyin and the target pinyin is higher than a preset threshold value.

Illustratively, the second pinyin proof set may include three, such as < sanqi >, third generation >, < xinlong, new dragon >. When the target pinyin is xinlon, the corresponding confusing pinyin may be xinlong or xinglong. At this time, the target second pinyin proofreading set corresponding to the confusing pinyin is < xinlong, newlong >, and the finally obtained proofread character is a newlong.

In the embodiment, by acquiring the confusion tone of the target pinyin and configuring the pinyin proofreading set matched with the confusion tone in the second pinyin proofreading set, the wrongly recognized characters of the ASR model caused by inaccurate pronunciation can be proofread into correct characters aiming at the problem of inaccurate pronunciation of the user, so that the recognition accuracy is improved.

In some embodiments, the step S202 may be specifically implemented by the following steps:

and separating the phonetic alphabets without phonetic symbols of all the characters according to the preset separators to obtain the target phonetic alphabets corresponding to the characters to be corrected.

In this embodiment, the characters recognized by the ASR model are tonal and are represented by phonetic symbols, for example, the phonetic symbols of "term" and "difference" are different, but when some users speak, the pronunciation of some characters may be confused, for example, the user reads the word that should read the fourth sound as the second sound, which causes misrecognition of the ASR model. For this purpose, the pinyin for each character may be removed, and for example, the "period" and "divergence" may be converted to the non-phonetic pinyin qi.

Illustratively, the preset separator may be a slash, a crossbar, a punctuation mark, or the like. For example, the target pinyin for the "installments" is fen-qi.

The embodiment of the application can be compatible with the problem that different pronunciation tones of different users are different by converting each character into phonetic transcription without phonetic symbols, and meanwhile, the separator is arranged to divide, so that the unclear Pinyin boundary of a plurality of characters is avoided, and the recognition effect and compatibility can be effectively improved.

In some embodiments, the method may further include the steps of:

determining the first user intention or the second user intention as the real user intention according to the matching value corresponding to the first user intention and the matching value corresponding to the second user intention;

For example, in the case of "pseudo-ginseng" as the character to be checked, and "third-stage" as the checked character, a mapping relationship between a voice recognition result and an intention is stored in the intelligent customer service system, a first user intention mapped and matched by "pseudo-ginseng" may be a traditional Chinese medicine in the intelligent customer service system, a matching value corresponding to the first user intention may be 0.1, and a second user intention mapped and matched by "third-stage" needs to stage the loan into third stage for the user in the intelligent customer service system. The second user intent corresponds to a match value of 0.8. And if the matching value corresponding to the second user intention is larger, selecting the second user intention as the user real intention, matching the corresponding dialect according to the user real intention, and replying the response information to the user.

According to the embodiment of the application, the corresponding user intentions are determined by respectively sending the characters before the proofreading and the characters after the proofreading, the user intention with the largest matching value is taken as the real user intention, the voice proofreading can be carried out, the phenomenon of overusing during the voice proofreading process is avoided, and the interaction effect of the intelligent customer service system is improved.

Exemplarily, fig. 4 is a schematic flow diagram of a second embodiment of a text proofing method provided in an embodiment of the present application, and as shown in fig. 4, the method may specifically include the following steps:

s401, inquiring a user;

s402, obtaining customer response voice;

s403, obtaining a voice recognition result through the recognition of an intelligent voice product ASR model;

s404, converting the voice recognition result into a target pinyin;

s405, a service process pinyin proofreading set is provided;

s4051, a pinyin proofreading set of the next business process;

s4052, acquiring corrected pinyin and corrected characters of a pinyin proofreading set of a business process;

s4053, matching the corrected pinyin with the target pinyin;

s4054, replacing the voice recognition result with a correction character;

s4055, traversing the pinyin proofreading set of the next service process;

s406, configuring a pinyin proofreading set for the current inquiry node;

s4061, a next node pinyin proofreading set is provided;

s4062, acquiring corrected pinyin and corrected characters of a node pinyin proofreading set;

s4063, matching the corrected pinyin with the target pinyin;

s4064, replacing the voice recognition result with a correction character;

s4065, traversing the pinyin proofreading set of the next node process;

s407, acquiring a user intention corresponding to the voice recognition result and a user intention corresponding to the corrected characters;

and S408, determining the real user intention.

In some embodiments, the above method further comprises the steps of:

According to the embodiment of the application, through establishing a post tracking mechanism, the characters to be corrected and the corrected characters are written into the interactive records, so that managers of the intelligent customer service system can conveniently compare the records afterwards, and whether the correction effect is in accordance with expectations is verified.

In some embodiments, the above method further comprises the steps of:

In this embodiment, the checking report can be provided for the administrator of the intelligent customer service system to perform post analysis, and whether the checking effect meets the expectation is determined.

Fig. 5 is a schematic diagram of data processing interaction of the intelligent customer service system according to an embodiment of the present application. The intelligent customer service system mainly relates to four big modules of a resource media server, an intelligent voice product, an NLP (non line segment) and an application system, and the intelligent customer service system specifically relates to the following processes:

s501, starting an outbound task;

s502, dialing a user telephone;

s503, the user connects the call;

s504, requesting a conversation;

s505, returning;

s506, requesting the text to be converted into voice;

s507, returning the synthesized voice;

s508, playing the synthesized voice;

s509, answering questions by a user;

s510, sending a voice recognition request;

s511, returning a voice recognition result;

s512, sending a voice recognition result;

s513, performing pinyin proofreading and error correction on the recognition result;

s514, sending a proofreading result;

s515, returning the user intention;

s516, mapping the reply dialect according to the user intention;

and S517, returning to the reply dialog.

The main content of the steps S501-S503 is that the application system starts an outbound task, triggers the media resource server to dial a client call, and the client connects the call; the main content of the steps S504-S505 is that the media resource server informs the application system that the telephone is connected, the application system starts the process and returns the open talk to the media resource server; S506-S508 are that the media resource server sends the open talk to the intelligent voice server, the intelligent voice server returns the voice synthesized by the TTS module, and the media resource server plays the voice to the client for listening; the main content of the steps S509-S512 is that the client answers the question, the media resource server sends the client voice to the intelligent voice product, and an ASR module of the intelligent voice product converts the client voice into characters and returns the characters to the media resource server; the media resource server sends the request to the application system to request a reply dialog; the main content of steps S513-S515 is that the application system receives the reply dialect, and checks whether the service flow and the flow node are configured with a pinyin proof set (the set is a key value pair of < K, V >, K is pinyin, and V is correct characters). If the configuration exists, traversing the pinyin of the configuration set, comparing the pinyin with the ASR recognition result, and if the comparison is successful, replacing the corresponding Chinese character in the ASR recognition result with the Chinese character corresponding to the pinyin in the configuration set; then sending the corrected ASR recognition result to the NLP; the NLP returns the client intention; the main content of steps S516-S517 is that the application system maps the corresponding reply dialogs according to the client intention and returns to the media resource server. This completes a closed loop of customer interaction.

The following are embodiments of the apparatus of the present application that may be used to perform embodiments of the method of the present application. For details which are not disclosed in the embodiments of the apparatus of the present application, reference is made to the embodiments of the method of the present application.

Fig. 6 is a schematic structural diagram of a text proofing device according to an embodiment of the present application, where the device may be integrated in an intelligent customer service system, or may be independent of the intelligent customer service system and cooperate with the intelligent customer service system to implement the technical solution. As shown in fig. 6, the text proofing apparatus 60 includes: a result obtaining module 61, a pinyin conversion module 62 and a character proofreading module 63.

The result obtaining module 61 is configured to obtain a voice recognition result of the current query node. The pinyin conversion module 62 is configured to obtain a target pinyin corresponding to the character to be corrected according to the pinyin of each character in the character to be corrected. The character proofreading module 63 is configured to proofread the characters to be proofread according to the pre-configured pinyin proofreading set and the target pinyin, so as to obtain the proofread characters.

The voice recognition result comprises characters to be corrected, which are obtained by recognizing voice input at the current query node, the pinyin correction set comprises a first pinyin correction set or a second pinyin correction set, the first pinyin correction set comprises corrected pinyin and corrected characters for correcting characters contained in a business process where the current query node is located, the second pinyin correction set comprises corrected pinyin and corrected characters for correcting characters contained in the current query node, and the corrected characters in the first pinyin correction set are different from the corrected characters in the second pinyin correction set.

In some embodiments, the text proofing module may be specifically configured to:

and according to the second pinyin proofreading set and the target pinyin, proofreading the characters to be proofread to obtain the proofread characters.

obtaining confusion pinyin corresponding to the target pinyin;

determining a previous node and a next node of a current query node according to a preset flow chart;

In some embodiments, an intent acquisition module is further included for:

In some embodiments, the device further includes a writing module, configured to write the text to be collated and the collated text into a preset interaction record.

In some embodiments, the apparatus further includes a generating module, configured to generate a collation report according to the characters to be collated, the pre-configured pinyin collation set, and the collated characters.

The apparatus provided in the embodiment of the present application may be used to execute the method in the above embodiments, and the implementation principle and the technical effect are similar, which are not described herein again.

It should be noted that the division of the modules of the above apparatus is only a logical division, and the actual implementation may be wholly or partially integrated into one physical entity, or may be physically separated. And these modules can be realized in the form of software called by processing element; or may be implemented entirely in hardware; and part of the modules can be realized in the form of calling software by the processing element, and part of the modules can be realized in the form of hardware. For example, the result obtaining module may be a processing element separately set up, or may be implemented by being integrated in a chip of the apparatus, or may be stored in a memory of the apparatus in the form of program code, and a processing element of the apparatus calls and executes the function of the result obtaining module. Other modules are implemented similarly. In addition, all or part of the modules can be integrated together or can be independently realized. The processing element here may be an integrated circuit with signal processing capabilities. In implementation, each step of the above method or each module above may be implemented by an integrated logic circuit of hardware in a processor element or an instruction in the form of software.

For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as: one or more Application Specific Integrated Circuits (ASICs), or one or more microprocessors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs), among others. For another example, when some of the above modules are implemented in the form of a processing element scheduler code, the processing element may be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor that can call program code. As another example, these modules may be integrated together, implemented in the form of a system-on-a-chip (SOC).

In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. The procedures or functions according to the embodiments of the present application are all or partially generated when the computer program instructions are loaded and executed on a computer. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wirelessly (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.

Fig. 7 is a schematic structural diagram of a computer device according to an embodiment of the present application. As shown in fig. 7, the computer device 70 includes: at least one processor 71, a memory 72, a bus 73, and a communication interface 74.

Wherein: the processor 71, the communication interface 74 and the memory 72 communicate with each other via a bus 73.

The communication interface 74 is used for communication with other devices. The communication interface 74 includes a communication interface for data transmission, a display interface or an operation interface for human-computer interaction, and the like.

The processor 71 is configured to execute computer-executable instructions stored in the memory 72, and may specifically execute the relevant steps in the method described in the above embodiments.

The processor may be a central processing unit, or an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits configured to implement an embodiment of the present invention. The computer device includes one or more processors, which may be the same type of processor, such as one or more CPUs; or may be different types of processors such as one or more CPUs and one or more ASICs.

A memory 72 for storing computer-executable instructions. The memory 72 may comprise high speed RAM memory and may also include non-volatile memory, such as at least one disk memory.

The present embodiment also provides a readable storage medium, in which computer instructions are stored, and when at least one processor of the computer device executes the computer instructions, the computer device executes the text proofreading method provided in the above various embodiments.

The present embodiments also provide a program product comprising computer instructions stored in a readable storage medium. The computer instructions can be read from a readable storage medium by at least one processor of a computer device, and the execution of the computer instructions by the at least one processor causes the computer device to implement the text proofing method provided by the various embodiments described above.

In the present application, "at least one" means one or more, "a plurality" means two or more. "and/or" describes the association relationship of the associated objects, meaning that there may be three relationships, e.g., a and/or B, which may mean: a exists alone, A and B exist simultaneously, and B exists alone, wherein A and B can be singular or plural. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship; in the formula, the character "/" indicates that the preceding and following related objects are in a relationship of "division". "at least one of the following" or similar expressions refer to any combination of these items, including any combination of the singular or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, a-b, a-c, b-c, or a-b-c, wherein a, b, c may be single or multiple.

It is to be understood that the various numerical references referred to in the embodiments of the present application are merely for convenience of description and distinction and are not intended to limit the scope of the embodiments of the present application. In the embodiment of the present application, the sequence numbers of the above-mentioned processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiment of the present application.

Finally, it should be noted that: the above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A text proofing method, comprising:

2. The method of claim 1, wherein the proofreading the text to be proofread according to the pre-configured pinyin proofreading set and the target pinyin to obtain a proofread text, comprises:

3. The method of claim 2, wherein the proofreading the text to be proofread based on the first pinyin proofreading set and the target pinyin comprises:

4. The method of claim 3, wherein determining a target first pinyin proof set in each first pinyin proof set that matches the target pinyin based on the target pinyin and the corrected pinyins in each first pinyin proof set comprises:

5. The method of claim 2, wherein the proofreading the text to be proofread according to the second pinyin proofreading set and the target pinyin comprises:

6. The method of claim 5, wherein determining a target second pinyin proof set in each second pinyin proof set that matches the target pinyin based on the target pinyin and the corrected pinyins in each second pinyin proof set comprises:

7. The method of claim 2, wherein the obtaining the business process in which the current query node is located comprises:

8. The method according to claim 1, wherein the obtaining the target pinyin corresponding to the text to be corrected according to the pinyin for each text in the text to be corrected comprises:

9. The method of claim 1, further comprising:

10. The method of claim 1, further comprising:

11. The method of claim 10, further comprising:

12. A text proofing apparatus, comprising:

13. A computer device, comprising: a processor, and a memory communicatively coupled to the processor;

the memory stores computer-executable instructions;

the processor executes computer-executable instructions stored by the memory to implement the method of any of claims 1-11.

14. A readable storage medium having stored therein computer instructions, which when executed by a processor, are adapted to implement the method of any one of claims 1-11.

15. A program product comprising computer instructions, characterized in that the computer instructions, when executed by a processor, implement the method of any of claims 1-11.