CN113676527A

CN113676527A - Information pushing method, device, equipment and storage medium

Info

Publication number: CN113676527A
Application number: CN202110915633.5A
Authority: CN
Inventors: 王鹏
Original assignee: Weikun Shanghai Technology Service Co Ltd
Current assignee: Weikun Shanghai Technology Service Co Ltd
Priority date: 2021-08-10
Filing date: 2021-08-10
Publication date: 2021-11-19

Abstract

The application provides an information pushing method, an information pushing device, information pushing equipment and a storage medium, wherein the method comprises the following steps: receiving first audio information of a client communication terminal, wherein the client communication terminal and a customer service communication terminal are in a communication connection state currently; identifying first voiceprint characteristic information corresponding to the first audio information, and identifying emotion reply information corresponding to the first voiceprint characteristic information in a preset information base; translating the first audio information into first character information, and extracting a plurality of text entity information from the first character information; identifying reference statement information corresponding to the text entity information in a preset information base; filling the emotion reply information and the reference sentence information into a preset push template to generate first push information; and sending the first push information to the customer service communication terminal. The method and the device effectively solve the problems that only passive query can be performed and the query accuracy is low.

Description

Information pushing method, device, equipment and storage medium

Technical Field

The present application relates to the field of information processing technologies, and in particular, to an information pushing method, an information pushing apparatus, an information pushing device, and a storage medium.

Background

At present, a sales specialist or a customer service is often communicated with a customer according to the past communication experience of the sales specialist or the customer service preferentially in the process of telephone communication with the customer, if a question raised by the customer exceeds the past communication experience of the sales specialist or the customer service, the customer needs to temporarily inquire relevant information such as a speech template or product data and then communicate with the customer, and finally the question of the customer is answered or product transaction is facilitated.

However, in the prior art, only the keywords temporarily input by sales personnel or customer service can be passively received for query, the query accuracy is low, and the obtained query result is difficult to meet the actual needs of customers.

Disclosure of Invention

The application provides an information pushing method, an information pushing device, information pushing equipment and a storage medium, and aims to solve the problems that in the prior art, only passive query can be performed, and the query accuracy is low.

In order to achieve the above object of the invention, the present application provides an information pushing method, including:

receiving first audio information of a client communication terminal, wherein the client communication terminal and a customer service communication terminal are in a communication connection state currently;

identifying first voiceprint characteristic information corresponding to the first audio information, and identifying emotion reply information corresponding to the first voiceprint characteristic information in a preset information base;

translating the first audio information into first character information, and extracting a plurality of text entity information from the first character information;

identifying reference statement information corresponding to the text entity information in a preset information base;

filling the emotion reply information and the reference sentence information into a preset push template to generate first push information;

and sending the first push information to the customer service communication terminal.

Further, the identifying first voiceprint feature information corresponding to the first audio information and identifying emotion reply information corresponding to the first voiceprint feature information in a preset information base includes:

performing pulse sampling on the first audio information, and performing feature extraction on sampled pulse data to obtain first voiceprint feature information;

identifying standard emotion voiceprint characteristics matched with the first voiceprint characteristic information in the information base, wherein a plurality of standard emotion voiceprint characteristics are stored in the information base, and each standard emotion voiceprint characteristic corresponds to at least one emotion state;

and acquiring corresponding emotion reply information from the information base according to the emotion state corresponding to the standard emotion voiceprint feature.

Further, the push template comprises an emotion replacement flag bit used for filling the emotion reply information, a reference replacement flag bit used for filling the reference sentence information and a plurality of bearing sentences; the step of filling the emotional response information and the reference sentence information into a preset push template to generate first push information includes:

identifying the number of the text entity information contained in the reference sentence information;

if the number of the text entity information is equal to 1, the reference statement information is used as independent reference information, the independent reference information containing different text entity information is randomly combined, and the text entity information is arranged in the first character information according to the sequence of the text entity information to obtain a plurality of first information lists;

sequentially filling each independent reference information in each first information list into each reference replacement identification position according to an arrangement sequence, and filling the emotion reply information into the emotion replacement identification positions to obtain the first push information;

if the number of the text entity information is greater than 1, the reference sentence information is used as joint reference information, each piece of the joint reference information is respectively filled into the reference replacement identification positions in each pushing template, the emotion reply information is respectively filled into the emotion replacement identification positions in each pushing template, and each filled pushing template is respectively used as one piece of the first pushing information.

Further, the extracting of the text entity information from the first text information includes:

filtering the assist words in the first character information to obtain first filtering information;

performing word segmentation on the first filtering information by adopting a maximum probability word segmentation method to obtain a plurality of first matching entities;

and matching the first matching entity with an entity dictionary by adopting a multi-pattern matching algorithm, and taking the first matching entity successfully matched as the text entity information.

Further, after the translating the first audio information into the first text information, the method includes:

identifying whether the first text information contains an identity mark;

if yes, searching corresponding user history information in the information base according to the identity, wherein the information base stores a plurality of pieces of user history information, and each identity corresponds to at least one piece of user history information;

and sending the user history information to the customer service communication terminal.

Further, after the sending the first push information to the customer service communication terminal, the method includes:

receiving second audio information sent to the client communication terminal by the customer service communication terminal in a first time period after the first push information is sent to the customer service communication terminal;

identifying second text information corresponding to the second audio information;

calculating a first bag-of-word vector of each piece of the first push information and a second bag-of-word vector of the second text information;

respectively calculating the text similarity of each first bag-of-word vector and each second bag-of-word vector, and judging whether each text similarity meets a preset similarity threshold value;

and matching and storing the second text information and each piece of first push information of which the text similarity does not meet the similarity threshold, and recording each matching result as an information group to be checked.

Further, after recording each pairing result as an information group to be checked, the method includes:

responding to an information checking instruction, and identifying a checking range corresponding to the information checking instruction;

reading a plurality of information groups to be checked in the checking range, and sending the information groups to be checked to the customer service communication terminal so that the customer service communication terminal displays the information groups to be checked one by one;

and responding to a selection result fed back by the customer service communication terminal, and storing the information group to be checked corresponding to the selection result into the information base.

The present application further provides an information pushing apparatus, which includes:

the system comprises an audio receiving unit, a first audio processing unit and a second audio processing unit, wherein the audio receiving unit is used for receiving first audio information of a client communication end, and the client communication end and a customer service communication end are in a communication connection state at present;

the emotion recognition unit is used for recognizing first voiceprint feature information corresponding to the first audio information and recognizing emotion reply information corresponding to the first voiceprint feature information in a preset information base;

the entity extraction unit is used for translating the first audio information into first character information and extracting preset text entity information from the first character information;

the sentence recognition unit is used for recognizing reference sentence information corresponding to the text entity information in a preset information base;

the push information generating unit is used for filling the emotion reply information and the reference sentence information into a preset push template to generate first push information;

and the information pushing unit is used for sending the first pushing information to the customer service communication terminal.

The application also provides a computer device, which includes a memory and a processor, where the memory stores a computer program, and the processor implements the steps of the information push method when executing the computer program.

The present application also proposes a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the steps of the above-mentioned information push method.

The technical scheme can have the following advantages or beneficial effects: the application provides an information pushing method, an information pushing system, a computer and a storage medium, wherein when a client communication end is in communication connection with a customer service communication end, real-time audio of a call of the client communication end is acquired as first audio information, so that information pushing is achieved in a communication process, and timeliness of the information pushing is improved; by identifying the emotional characteristics in the first audio information, the current emotional state of the client can be accurately identified, so that the response can be conveniently carried out according to the emotional response information in the first push information under the condition that the experience of the salesperson or customer service is insufficient, and the effectiveness of communication is improved; by identifying the first audio information as the character information, the problem of entity identification error caused by the same or similar pronunciations under the condition of directly carrying out entity identification on the audio information is avoided; text entity recognition is carried out on the first character information through a preset semantic model, so that the influence of other meaningless words or characters on message pushing is eliminated; the method comprises the steps of identifying standard word information corresponding to text entity information in a preset information base to obtain reference sentence information possibly required for replying the first audio information, and pushing first push information formed by combining the reference sentence information to a customer service communication end, so that under the condition that salesmen or the customer service and other staff at the customer service communication end do not actively inquire the data, related data required for replying questions put forward by a user at a client communication end can be actively inquired according to the first push information, the data inquiry time in the real-time communication process is reduced, the communication timeliness is improved, and the inquiry accuracy is improved.

Drawings

Fig. 1 is a flowchart of an information pushing method according to an embodiment of the present application;

fig. 2 is a detailed flowchart of an information pushing method according to an embodiment of the present application;

fig. 3 is a detailed flowchart of an information pushing method according to an embodiment of the present application;

fig. 4 is a detailed flowchart of an information pushing method according to an embodiment of the present application;

fig. 5 is a detailed flowchart of an information pushing method according to an embodiment of the present application;

fig. 6 is a detailed flowchart of an information pushing method according to an embodiment of the present application;

fig. 7 is a detailed flowchart of an information pushing method according to an embodiment of the present application;

FIG. 8 is a block diagram illustrating a structure of an information pushing apparatus according to an embodiment of the present application;

FIG. 9 is a block diagram illustrating a computer device according to an embodiment of the present application;

the implementation, functional features and advantages of the objectives of the present application will be further explained with reference to the accompanying drawings.

Detailed Description

The technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the 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.

It should be noted that all directional indicators (such as up, down, left, right, front, back, etc.) in the embodiments of the present application are only used to explain the relative position relationship between the components, the motion situation, etc. in a specific posture (as shown in the drawings), and if the specific posture is changed, the directional indicator is changed accordingly, and the connection may be a direct connection or an indirect connection.

The term "and/or" herein is merely an association describing an associated object, meaning that three relationships may exist, e.g., a and B, may mean: a exists alone, A and B exist simultaneously, and B exists alone.

In addition, descriptions in this application as to "first", "second", etc. are for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicit to the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In addition, technical solutions between various embodiments may be combined with each other, but must be realized by a person skilled in the art, and when the technical solutions are contradictory or cannot be realized, such a combination should not be considered to exist, and is not within the protection scope of the present application.

Referring to fig. 1, an information push method according to an embodiment of the present application includes:

s1: receiving first audio information of a client communication terminal, wherein the client communication terminal and a customer service communication terminal are in a communication connection state currently;

s2: identifying first voiceprint characteristic information corresponding to the first audio information, and identifying emotion reply information corresponding to the first voiceprint characteristic information in a preset information base;

s3: translating the first audio information into first character information, and extracting preset text entity information from the first character information;

s4: identifying reference statement information corresponding to the text entity information in a preset information base;

s5: filling the emotion reply information and the reference sentence information into a preset push template to generate first push information;

s6: and sending the first push information to the customer service communication terminal.

According to the embodiment, when the client communication end is in communication connection with the customer service communication end, the real-time audio of the communication of the client communication end is acquired as the first audio information, so that information push is realized in the communication process, and the timeliness of the information push is improved; by identifying the emotional characteristics in the first audio information, the current emotional state of the client can be accurately identified, so that the response can be conveniently carried out according to the emotional response information in the first push information under the condition that the experience of the salesperson or customer service is insufficient, and the effectiveness of communication is improved; by identifying the first audio information as the character information, the problem of entity identification error caused by the same or similar pronunciations under the condition of directly carrying out entity identification on the audio information is avoided; text entity recognition is carried out on the first character information through a preset semantic model, so that the influence of other meaningless words or characters on message pushing is eliminated; the method comprises the steps of identifying standard word information corresponding to text entity information in a preset information base to obtain reference sentence information possibly required for replying the first audio information, and pushing first push information formed by combining the reference sentence information to a customer service communication end, so that under the condition that salesmen or the customer service and other staff at the customer service communication end do not actively inquire the data, related data required for replying questions put forward by a user at a client communication end can be actively inquired according to the first push information, the data inquiry time in the real-time communication process is reduced, the communication timeliness is improved, and the inquiry accuracy is improved.

For the step S1, the client communication end is usually a device such as a fixed phone or a mobile phone capable of initiating a real-time communication application, and the client communication end is usually a device such as a fixed phone or a mobile phone capable of receiving a real-time communication application; the client communication end and the customer service communication end are connected through a telephone repeater generally. The telephone repeater binds the telephone lines at two ends, and in the embodiment, the telephone repeater can acquire the first audio information; the customer service communication terminal is generally a terminal device with a display function, or is connected with a terminal device with a display function.

With respect to the above step S2, since the customer may have an emotion such as anger, worry or happiness due to personal status during the communication process, if the customer has a negative emotion such as anger, the sales specialist or customer service may not be able to calm down in time, which may further aggravate the emotion of the customer; therefore, voiceprint acquisition needs to be performed on the first audio information through a preset acquisition model to obtain first voiceprint characteristic information, so that dialogues corresponding to different emotional states are inquired in a preset information base.

For the above step S3, the input first audio information may be encoded, decoded and output by a Hidden Markov Model (HMM), a deep neural network based "end-to-end" method; the first audio information can be converted into corresponding first text information through a speech recognition engine; in a specific embodiment, the speech recognition engine may be a tench cloud intelligent speech recognition engine, a news-flight speech recognition engine, a Baidu real-time speech recognition engine, or the like, that is, it may be capable of converting audio information into text information.

Specifically, text entity information may be extracted from the first text information through a preset NLP (Natural Language Processing) model, specifically, the first text information is input into the NLP model, and is semantically understood through an NLU (Natural Language Understanding) unit in the NLP model, for example, if a client at a client communication end wants to query an account balance, "i want to check the account balance of the user a," i want to check how much money is left in my account a "may also be said," i want to check how much money is left in my account a "may be said if the client wants to query the account balance, at this time, semantic feature extraction is performed through an RNN (Recurrent Neural Networks) in the NLU unit, and recognition results" account a "and" balance "are output through an output layer of the RNN, and recognition results" account a "and" balance "are output, that is the text entity information.

As described in step S4, a standard word information in the information base corresponds to at least one reference sentence information, and the corresponding reference sentence information can be obtained by respectively identifying the standard word information corresponding to each text entity information in the information base, where the standard word information is a word having the same or similar semantic meaning as the text entity information; specifically, the information base includes, but is not limited to, a preset search base, a knowledge map, a manufacturer product database, an emotion feature database, and other databases with relevant reference information; after the text entity information is generated, the standard word information with the same or similar semantics as the text entity information is identified in the information base, for example, if the text entity information is "Qinghai", "Play" and "ticket", the standard word information "Qinghai", "Tourism" and "train ticket" corresponding to "Qinghai", "Play" and "ticket" are respectively searched, and the reference sentence information corresponding to the standard word information "Qinghai", "Tourism" and "train ticket" is respectively obtained in the information base, for example, the reference sentence information corresponding to "Qinghai" includes the introduction to Qinghai province, the reference sentence information packet tourism notice and tourism strategy and the like corresponding to "Tourism", and the reference sentence information corresponding to "train ticket" is the train timetable after the current time and the purchase amount of the train ticket and the like.

As described in the step S5, after obtaining the pieces of reference sentence information, the pieces of reference sentence information may be combined to obtain one or more pieces of first push information, and after obtaining the pieces of reference sentence information corresponding to "Qinghai", "Tourism", and "train ticket", for example, a piece of reference sentence information corresponding to "Qinghai", a piece of reference sentence information corresponding to "Tourism", and a piece of reference sentence information corresponding to "train ticket" may be extracted and combined to obtain a piece of combined sentence related to "Qinghai", "Tourism", and "train ticket", and after extracting and combining for many times, a plurality of combined sentences related to "Qinghai", "Tourism", and "train ticket" are obtained; and combining one combined sentence with one emotion reference sentence, filling the combined sentence into a preset push template to form first push information with high readability for sales specialists or customer service, and obtaining a plurality of pieces of first push information after performing the operation on different combined sentences for a plurality of times.

In a specific embodiment, the push template may include the emotion replacement flag, a reference replacement flag, and a plurality of supporting sentences; the adapting sentences are arranged between the emotion replacing identification positions and the reference replacing identification positions or between different reference replacing identification positions, and the adapting sentences can comprise respectful terms, turning sentences and the like; illustratively, the take-up statement may be "respected customer, your good, the data you need is as follows: "," honored customer, hello, the information you query is as follows: "," and "," in addition ", etc. to be convenient for the sales specialist or customer service to read the first push information conveniently, reduced the information understanding time, and then improved communication efficiency.

As described in step S6, the first push messages are sent to the customer service communication end in a wireless or wired transmission manner, so that the display unit connected to the customer service communication end displays the one or more pieces of first push messages in turn in real time, and the display unit is placed in front of the visual field of the sales specialist or customer service staff, so as to prompt the sales specialist or customer service staff in real time.

In an embodiment, referring to fig. 2, the identifying the first voiceprint feature information corresponding to the first audio information and the identifying the emotional response information S2 corresponding to the first voiceprint feature information in the preset information base includes:

s21: performing pulse sampling on the first audio information, and performing feature extraction on sampled pulse data to obtain first voiceprint feature information;

s22: identifying standard emotion voiceprint characteristics matched with the first voiceprint characteristic information in the information base, wherein a plurality of standard emotion voiceprint characteristics are stored in the information base, and each standard emotion voiceprint characteristic corresponds to at least one emotion state;

s23: and acquiring corresponding emotion reply information from the information base according to the emotion state corresponding to the standard emotion voiceprint feature.

For step S21, PCM (Pulse Code Modulation) data sampling may be performed on the first audio information to discretize the first audio information, obtain an amplitude of a Pulse of the first audio information, and perform feature extraction on the PCM data to obtain first voiceprint feature information.

In step S22, the preset information base stores standard emotion voice print characteristics including emotion states of happiness, sadness, anger and the like, and when the first voice print characteristic information corresponding to the first audio information is acquired, the first voice print characteristic information is matched with the standard emotion voice print characteristics, so as to identify the current emotion state of the client.

For step S23, after the current emotional state of the customer is identified, corresponding emotional response information may be obtained, so that the sales specialist or customer service may respond to the emotional state of the customer according to the emotional response information, for example, refer to a suitable dialect to pacify the customer when the emotional state of the customer is anger or worries, thereby avoiding a situation that the sales specialist or customer service should further irritate the customer' S emotion due to insufficient experience, and improving communication efficiency and communication effectiveness.

In an embodiment, referring to fig. 3, the push template includes an emotion replacement flag for filling in the emotional response information, a reference replacement flag for filling in the reference sentence information, and a plurality of supporting sentences; the above-mentioned filling the emotional response information and the reference sentence information into a preset push template to generate the first push information S5 includes:

s51: identifying the number of text entity information contained in the reference sentence information;

s52: if the number of the text entity information is equal to 1, the reference statement information is used as independent reference information, the independent reference information containing different text entity information is randomly combined, and the text entity information is arranged in the first character information according to the sequence of the text entity information to obtain a plurality of first information lists;

s53: sequentially filling each independent reference information in each first information list into each reference replacement identification position according to an arrangement sequence, and filling the emotion reply information into the emotion replacement identification positions to obtain the first push information;

s54: if the number of the text entity information is greater than 1, the reference sentence information is used as joint reference information, each piece of the joint reference information is respectively filled into the reference replacement identification positions in each pushing template, the emotion reply information is respectively filled into the emotion replacement identification positions in each pushing template, and each filled pushing template is respectively used as one piece of the first pushing information.

For step S51, a reference sentence information includes at least one text entity information, for example, for the text entity information of "qinghai", "tourism", and "train ticket", a reference sentence information Y1 may be qinghai tourism strategy, and includes two text entity information of "qinghai" and "tourism"; a reference sentence information Y2 may be a Qinghai food ranking list, which includes a text entity information of "Qinghai"; a reference sentence information Y3 can be a travel necessary inventory, and contains a text entity information of "travel".

For steps S52-S53, for the reference sentence information Y2 and Y3, the two may be combined in the order of Y2-Y3 to obtain a first information list, so that the first information list can simultaneously contain relevant data of "Qinghai" and "travel", at this time, the first information list is filled in the reference replacement flag of the push template, and the emotional response information is filled in the emotional replacement flag of the push template to obtain the first push information suitable for reading, and the first push information can contain enough text entity information, thereby avoiding the problem that a sales specialist or a customer service can obtain comprehensive data through multiple times of searching.

For step S54, since the reference sentence information Y1 includes the related data of "qinghai" and "travel", that is, includes a plurality of text entity information, at this time, the reference sentence information can be directly filled into the reference replacement flag of the push template, and the emotional response information is filled into the emotional replacement flag of the push template, so as to obtain the first push information suitable for reading, and the first push information can also include enough text entity information, thereby avoiding the problem that the sales specialist or customer service can obtain comprehensive data by turning over for many times. In a specific embodiment, each piece of reference sentence information can be sorted from large to small according to the number of text entity information contained in the piece of reference sentence information, and it can be understood that in real life, the larger the number of text entity information contained in the piece of reference sentence information is, the larger the association degree between the piece of reference sentence information and the first text information is; therefore, the reference statement information is sequenced according to the sequence from large to small in number, the reference statement information with the maximum relevance can be arranged at the forefront, the first push information formed by the reference statement information is sent to the customer service communication end according to the sequence, a sales specialist or customer service can conveniently and preferentially see the first push information with the maximum relevance after receiving a plurality of pieces of first push information, the time for the sales specialist or customer service to obtain the reference statement information with the maximum relevance is shortened, and the response efficiency is further improved.

In an embodiment, referring to fig. 4, the extracting the text entity information S3 from the first text information includes:

s31: filtering the assist words in the first character information to obtain first filtering information;

s32: performing word segmentation on the first filtering information by adopting a maximum probability word segmentation method to obtain a plurality of first matching entities;

s33: and matching the first matching entity with an entity dictionary by adopting a multi-pattern matching algorithm, and taking the first matching entity successfully matched as the text entity information.

For step S31, since the vocabularies such as "how", "bar", "cala" and "o" will appear in the spoken language communication process, in order to improve the efficiency of subsequent text entity recognition, the vocabularies are filtered before the text entity recognition, thereby simplifying the length of the text to be recognized. Specifically, a speech-assisted word library may be set, and words that are located at the end of the first text information and are the same as words in the speech-assisted word library may be used as speech-assisted words and filtered.

For step S32, when some sentences are subjected to entity segmentation, the maximum length of the words may not be the optimal segmentation, but if the maximum length of the words is not the optimal segmentation, multiple segmentation results may occur in the same sentence. Therefore, at this time, the probability of each segmentation result needs to be calculated by a maximum probability segmentation method, and the segmentation result with the highest probability is selected as the optimal segmentation.

Specifically, the optimal segmentation is argmaxwP (W | S ═ S), where W is all possible segmentation combinations, and S is the sentence to be segmented; the segmentation probability p (w | s) ═ p (w) p (s | w) p(s), and since the sentence to be segmented is unchanged, i.e. p(s) is unchanged, p (s | w) is always 1, only p (w) needs to be calculated. The word frequency of each word needs to be recorded in a word list, and the joint probability of each independent word is calculated by using unary grammar, which comprises the following steps:

in order to avoid the problem that word probability is too small and overflow can be caused after multiple successive multiplications, in practice, the product is generally replaced by logarithmic summation: logp (w) ═ Σ n1logp (wn), the joint probability of a word is obtained.

And (3) performing a Dynamic Programming (DP) (dynamic programming) algorithm according to a Direct Acyclic Graph (DAG), and performing backward calculation from the back to the front of the DAG Graph to accumulate the maximum probability result every time. And calculating the product of the different segmentation word probability and the subsequent maximum segmentation probability of each word as the prefix, taking the maximum result as the optimal segmentation, and recurrently forwarding the sentence in sequence. And finally, obtaining the optimal segmentation word position of the first word of the sentence, sequentially querying the segmentation words backwards from the optimal segmentation word position of the first word, and finally obtaining the optimal segmentation result, namely the first matching entity, so that the accuracy of text entity identification is improved.

For step S33, a DAT (multi-pattern matching) algorithm is used to match the first matching entity with an entity dictionary in the semantic recognition model, and the first matching entity that is successfully matched is used as text entity information, where the entity dictionary may be a chinese entity dictionary such as a dog search entity dictionary or an english entity dictionary.

In one embodiment, referring to fig. 5, the above-mentioned translating the first audio information into the first text information S3 includes:

s34: identifying whether the first text information contains an identity mark;

s35: if yes, searching corresponding user history information in the information base according to the identity, wherein the information base stores a plurality of pieces of user history information, and each identity corresponds to at least one piece of user history information;

s36: and sending the user history information to the customer service communication terminal.

For step S34, the id includes a user name, an id number, a user account, a mobile phone number of the user, and the like.

For steps S35-S36, if the first text information includes the id, it indicates that the customer has recited the information in the call, and at this time, if the sales specialist or the customer service manually inputs the information and then searches for the data corresponding to the id, the waiting time of the customer is longer, and even the problem that the data of the customer cannot be searched at one time due to the input error of the id occurs, so that the communication time efficiency and accuracy further decrease.

In addition, when the client does not report the own identity in the communication process, the client data corresponding to the telephone number information can be searched in the user database by acquiring the telephone number information of the communication end of the client; specifically, before a sales specialist or a customer service connects with a telephone of a customer, a telephone number of the customer is acquired, or an incoming call number or an actively dialed telephone number of a telephone terminal of the customer service is acquired, customer information is searched for from a database according to the telephone number, and the customer information is sent to a customer service communication terminal, for example: product information transacted by the client through the telephone number, related products bound by the telephone number and the like; the related past data of the client is presented to a sales specialist or a customer service in advance, so that the customer service can know the related information of the client in advance before communication, and the next communication is facilitated. The client data may be main data information related to the telephone number, for example, when history data related to the telephone number is multiple services handled, the client data may be names of the multiple services and time, duration and brief introduction of each service. After a sales specialist or a customer service is connected with a telephone of a client, if the client wants to consult when a service transacted before is due, but the name of the client is not clear, and the client may ask 'when the service transacted before is due', at the moment, the sales specialist or the customer service also needs to inquire personal information of the client to know what the service transacted before is, and by acquiring telephone number information of a communication end of the client, the 'service transacted before me' can be searched according to the telephone number information to obtain corresponding service information, and according to the 'due' searching corresponding time for transacting the service and the service limit of the service, the corresponding due time is obtained by adding the service time to the service limit, and reference statement information of 'the service A transacted before will be due at the date B' is obtained. For example, the client may ask "how many days are left for business transacted before me to expire" in this way, at this time, the "business transacted before me" may be searched according to the telephone number information, that is, the information of the business a corresponding to the telephone number information, and the time for transacting the business a, the term of the business a, and the current date may be searched according to the "expiration" and the "how many days", the "due date" and the "remaining days" are calculated by the three, and the reference statement information that the business a will expire on the date B and the remaining days are N days is obtained, so that the expiration time and the remaining days can be quickly and accurately obtained without calculation by a sales specialist or a customer service, and the communication work efficiency is improved.

In an embodiment, referring to fig. 6, after the sending the first push message S5 to the customer service communication end, the method includes:

s61: receiving second audio information sent to the client communication terminal by the customer service communication terminal in a first time period after the first push information is sent to the customer service communication terminal;

s62: identifying second text information corresponding to the second audio information;

s63: calculating a first bag-of-word vector of each piece of the first push information and a second bag-of-word vector of the second text information;

s64: respectively calculating the text similarity of each first bag-of-word vector and each second bag-of-word vector, and judging whether each text similarity meets a preset similarity threshold value;

s65: and matching and storing the second text information and each piece of first push information of which the text similarity does not meet the similarity threshold, and recording each matching result as an information group to be checked.

For steps S61-S62, after the one or more pieces of first push information are sent to the customer service communication end, the reply audio of the sales specialist or customer service to the customer, that is, the second audio information, is obtained, so as to obtain the corresponding second text information.

For step S63, constructing a word-text matrix through a preset word bag model, and performing weight assignment on each word in the matrix according to importance by adopting a TF-IDF (term frequency-inverse document frequency) method; deleting words with assignment results lower than a preset threshold value by adopting singular value decomposition of an SVD matrix so as to perform dimension reduction processing; and obtaining a final word-text matrix so as to obtain a corresponding bag-of-words vector.

For steps S64-S65, if the text similarity between the first bag-of-word vector and the second bag-of-word vector of a first piece of pushed information does not satisfy the preset similarity threshold, the first piece of pushed information and the second piece of text information are stored, and a piece of second text information corresponding to the first piece of pushed information is recorded as an information group to be checked, so as to facilitate the reason that the semantic difference between the reply of a subsequent sales specialist or customer service and the first piece of pushed information is too large, thereby improving the accuracy of information pushing.

In an embodiment, referring to fig. 7, after recording each pairing result as an information group to be checked S65, the method includes:

s66: responding to an information checking instruction, and identifying a checking range corresponding to the information checking instruction;

s67: reading a plurality of information groups to be checked in the checking range, and sending the information groups to be checked to the customer service communication terminal so that the customer service communication terminal displays the information groups to be checked one by one;

s68: and responding to a selection result fed back by the customer service communication terminal, and storing the information group to be checked corresponding to the selection result into the information base.

For step S66, the checking range includes a global range, a time range, a business range, and the like, for example, all information groups to be checked in a week are checked, or information groups to be checked corresponding to the business system X1 in three days are checked.

For step S67, after the call is over, the information group to be checked is used for service summarization by the sales specialist or customer service, wherein the information group to be checked is displayed, a selection button is provided corresponding to each information group to be checked for the sales specialist or customer service to look up, select and check through the display interface of the customer service communication terminal, and when the sales specialist or customer service performs comparison and review of the information group to be checked, it can be determined whether the answer at that time is correct and reasonable; and if the second audio information replied at the moment is considered to be superior to the first push information, the corresponding information group to be checked is selected, and the information group to be checked is stored in the information base, so that the answer can be directly quoted when the same question is encountered later. If the sales specialist or the customer service finds that the second audio information replied by the sales specialist or the customer service is wrong, the sales specialist or the customer service can know and correct the mistake, and a learning example is provided for the sales specialist or the customer service, so that the communication efficiency and the accuracy of next communication are improved.

Referring to fig. 8, an embodiment of the present application further provides an information pushing apparatus, including:

the audio receiving unit 100 is configured to receive first audio information of a client communication end, where the client communication end and a customer service communication end are currently in a communication connection state;

the emotion recognition unit 200 is configured to recognize first voiceprint feature information corresponding to the first audio information, and recognize emotion reply information corresponding to the first voiceprint feature information in a preset information base;

an entity extracting unit 300, configured to translate the first audio information into first text information, and extract preset text entity information from the first text information;

a sentence recognition unit 400, configured to recognize reference sentence information corresponding to the text entity information in a preset information base;

a pushed information generating unit 500, configured to fill the emotional response information and the reference sentence information into a preset pushed template, and generate first pushed information;

an information pushing unit 600, configured to send the first push information to the customer service communication end.

According to the embodiment, when the client communication end is in communication connection with the customer service communication end, the real-time audio of the communication of the client communication end is acquired as the first audio information, so that information push is realized in the communication process, and the timeliness of the information push is improved; by identifying the emotional characteristics in the first audio information, the current emotional state of the client can be accurately identified, so that the response can be conveniently carried out according to the emotional response information in the first push information under the condition that the experience of the salesperson or customer service is insufficient, and the effectiveness of communication is improved; by identifying the first audio information as the character information, the problem of entity identification error caused by the same or similar pronunciations under the condition of directly carrying out entity identification on the audio information is avoided; text entity recognition is carried out on the first character information through a preset semantic model, so that the influence of other meaningless words or characters on message pushing is eliminated; the method comprises the steps of identifying standard word information corresponding to text entity information in a preset information base to obtain reference statement information possibly required for replying the first audio information, and pushing first push information formed by combining the reference statement information to a customer service communication end, so that under the condition that salesmen or workers such as customer service at the customer service communication end do not actively inquire the data, relevant data required for replying questions put forward by a user at the client communication end can be obtained according to the first push information, the data inquiry time in the real-time communication process is shortened, and the communication timeliness is improved.

In an embodiment, the emotion recognition unit 200 is further configured to:

In one embodiment, the push template comprises an emotion replacement flag bit for filling in the emotion reply information, a reference replacement flag bit for filling in the reference sentence information, and a plurality of supporting sentences; the push information generating unit 500 is further configured to:

identifying the number of text entity information contained in the reference sentence information;

In an embodiment, the entity extracting unit 300 is further configured to:

In an embodiment, the apparatus further includes a user information obtaining unit 700, configured to:

identifying whether the first text information contains an identity mark;

In an embodiment, the information checking unit 800 is further included to:

In an embodiment, the information checking unit 800 is further configured to:

Referring to fig. 9, a computer device, which may be a server and whose internal structure may be as shown in fig. 9, is also provided in the embodiment of the present application. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the computer designed processor is used to provide computational and control capabilities. The memory of the computer device comprises a storage medium and an internal memory. The storage medium stores an operating system, a computer program, and a database. The memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used for storing various historical data and the like. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program can realize the information pushing method of any one of the above embodiments when executed by a processor, and includes the steps of: receiving first audio information of a client communication terminal, wherein the client communication terminal and a customer service communication terminal are in a communication connection state currently; identifying first voiceprint characteristic information corresponding to the first audio information, and identifying emotion reply information corresponding to the first voiceprint characteristic information in a preset information base; translating the first audio information into first character information, and extracting preset text entity information from the first character information; identifying reference statement information corresponding to the text entity information in a preset information base; filling the emotion reply information and the reference sentence information into a preset push template to generate first push information; and sending the first push information to the customer service communication terminal.

Those skilled in the art will appreciate that the architecture shown in fig. 9 is merely a block diagram of some of the structures associated with the present teachings and is not intended to limit the computing devices to which the present teachings may be applied.

An embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, can implement the information pushing method in any of the above embodiments, and the method includes: receiving first audio information of a client communication terminal, wherein the client communication terminal and a customer service communication terminal are in a communication connection state currently; identifying first voiceprint characteristic information corresponding to the first audio information, and identifying emotion reply information corresponding to the first voiceprint characteristic information in a preset information base; translating the first audio information into first character information, and extracting preset text entity information from the first character information; identifying reference statement information corresponding to the text entity information in a preset information base; filling the emotion reply information and the reference sentence information into a preset push template to generate first push information; and sending the first push information to the customer service communication terminal.

In the executed information pushing method, when the client communication terminal is in communication connection with the client service communication terminal, the real-time audio of the call of the client communication terminal is acquired as the first audio information, so that information pushing is realized in the communication process, and the timeliness of the information pushing is improved; by identifying the emotional characteristics in the first audio information, the current emotional state of the client can be accurately identified, so that the response can be conveniently carried out according to the emotional response information in the first push information under the condition that the experience of the salesperson or customer service is insufficient, and the effectiveness of communication is improved; by identifying the first audio information as the character information, the problem of entity identification error caused by the same or similar pronunciations under the condition of directly carrying out entity identification on the audio information is avoided; text entity recognition is carried out on the first character information through a preset semantic model, so that the influence of other meaningless words or characters on message pushing is eliminated; the method comprises the steps of identifying standard word information corresponding to text entity information in a preset information base to obtain reference sentence information possibly required for replying the first audio information, and pushing first push information formed by combining the reference sentence information to a customer service communication end, so that under the condition that salesmen or the customer service and other staff at the customer service communication end do not actively inquire the data, related data required for replying questions put forward by a user at a client communication end can be actively inquired according to the first push information, the data inquiry time in the real-time communication process is reduced, the communication timeliness is improved, and the inquiry accuracy is improved.

It will be understood by those skilled in the art that all or part of the processes of the methods of the above embodiments may be implemented by hardware associated with instructions of a computer program, which may be stored on a non-volatile computer-readable storage medium, and when executed, may include processes of the above embodiments of the methods. Any reference to memory, storage, database, or other medium provided herein and used in the examples may include non-volatile and/or volatile memory. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), double-rate SDRAM (SSRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and bus dynamic RAM (RDRAM).

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, apparatus, article, or method that includes the element.

The above description is only a part of the embodiments of the present application, and not intended to limit the scope of the present application, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the specification and the drawings of the present application, or which are directly or indirectly applied to other related technical fields, are also included in the scope of the present application.

Claims

1. An information pushing method, comprising:

translating the first audio information into first character information, and extracting preset text entity information from the first character information;

2. The information pushing method according to claim 1, wherein the identifying first voiceprint feature information corresponding to the first audio information and identifying emotional response information corresponding to the first voiceprint feature information in a preset information base includes:

3. The information pushing method according to claim 1, wherein the pushing template includes an emotion replacement flag for filling in the emotional response information, a reference replacement flag for filling in the reference sentence information, and a plurality of supporting sentences; the step of filling the emotional response information and the reference sentence information into a preset push template to generate first push information includes:

4. The information pushing method according to claim 1, wherein the extracting of the text entity information from the first text information includes:

5. The information pushing method according to claim 1, wherein after translating the first audio information into first text information, the method comprises:

identifying whether the first text information contains an identity mark;

6. The information push method according to claim 1, wherein after sending the first push information to the customer service communication terminal, the method comprises:

7. The information pushing method according to claim 6, wherein after recording each pairing result as an information group to be checked, the method comprises:

8. An information pushing apparatus, comprising:

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the detection method of an audio system according to any of claims 1 to 7.

10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the detection method of an audio system according to any one of claims 1 to 7.