CN115188374A

CN115188374A - Method and device for updating dialect

Info

Publication number: CN115188374A
Application number: CN202210713080.XA
Authority: CN
Inventors: 冯鑫
Original assignee: Bairong Ruicheng Information Technology Co ltd
Current assignee: Bairong Ruicheng Information Technology Co ltd
Priority date: 2022-06-22
Filing date: 2022-06-22
Publication date: 2022-10-14

Abstract

The application discloses a conversation updating method and device, and relates to the technical field of communication. The method of the present application comprises: determining a jargon sentence with the highest unit forming rate from historical data, and recording the jargon sentence as a supplementary sentence, wherein the historical data comprises order information corresponding to each order in a voice dialing process and voice information corresponding to each order; selecting a first target node from a conversational flow tree, wherein the conversational flow tree comprises at least two nodes, each node corresponds to a conversational term sentence, and the first target node is a node except a root node in the conversational flow tree; fusing the conversational terminology sentence of the first target node with the supplementary sentence to obtain a first fused sentence; and replacing the first fusion statement with the conversational statement of the first target node.

Description

Method and device for updating dialect

Technical Field

The present application relates to the field of communications technologies, and in particular, to a method and an apparatus for speech technology update.

Background

In the operation process of service industries such as banks, insurance and the like, such enterprises generally need a large number of online customer service personnel. With the advent of intelligent products, most enterprises generally choose to introduce intelligent voice products in order to save labor costs, for example, the intelligent voice products use preset speech data to perform automatic voice communication with users, so that the users can make orders based on the voice communication. In the process, the accuracy of the preset tactical data is directly applied to the ordering condition of the user, so that the enterprises can update the tactical data frequently so as to ensure the ordering amount.

At present, in the process of updating the conversational data, a worker with maintenance experience is often needed to adjust and update the conversational terminology sentence in the conversational data in a manual mode, but in practical application, the current manual mode of conversational updating needs to have certain requirements on the working experience of an operator, which leads to excessive dependence on the manual experience in the updating process, and once the operator fails or has insufficient experience, the accuracy of conversational updating is low, and then the single quantity is affected.

Disclosure of Invention

The embodiment of the application provides a method and a device for updating a conversational operation, and mainly aims to solve the problem that the accuracy of the conversational operation updating is low in the current conversational operation updating process.

In order to solve the above technical problem, an embodiment of the present application provides the following technical solutions:

in a first aspect, the present application provides a conversation update method, including:

determining a jargon sentence with the highest unit forming rate from historical data, and recording the jargon sentence as a supplementary sentence, wherein the historical data comprises order information corresponding to each order in a voice dialing process and voice information corresponding to each order;

selecting a first target node from a conversational flow tree, wherein the conversational flow tree comprises at least two nodes, each node corresponds to a conversational term sentence, and the first target node is a node except a root node in the conversational flow tree;

fusing the speech clause of the first target node with the supplementary sentence to obtain a first fused sentence;

and replacing the first fusion statement with the conversational statement of the first target node.

Optionally, the jargon sentence determined to be the highest single rate from the historical data includes:

in the historical data, determining the jargon sentence according to the voice information corresponding to the order with the order result being successful in the order information;

determining the total number of orders corresponding to each speech term sentence;

determining the order result corresponding to each speech term sentence as a successful order number;

carrying out quotient calculation according to the amount of orders and the total number of orders to obtain the order forming rate corresponding to the speech term sentence;

determining the conversational phrase with the highest singleton rate as the supplementary phrase in all the conversational phrases.

Optionally, the determining the jargon sentence according to the voice information corresponding to the order in which the order result is successful in the order information includes:

judging whether the order result is successful or not according to the order information;

if the order information is successful, the voice information corresponding to the order information is obtained;

and processing the voice information by using a preset speech model according to the voice information to obtain the speech term sentence, wherein the preset speech model is used for generating a speech corresponding to the semantics based on the semantics of the voice information, the voice information is obtained from knowledge bases, and each knowledge base comprises a user question and a corresponding response content.

Optionally, the selecting a first target node in the conversational flow tree includes:

respectively determining first semantics of the supplementary sentences through a semantic analysis model;

determining the semantics of the dialect statement corresponding to each node in the dialect flow tree through the semantic analysis model, and recording the semantics as second semantics;

determining the second semantic meaning with the same category as the first semantic meaning from a plurality of second semantic meanings, and marking the second semantic meaning as a target semantic meaning;

and determining the corresponding node as the first target node according to the conversational term sentence corresponding to the target semantic.

Optionally, after the replacing the first fused sentence with the jargon sentence of the first target node, the method further includes:

determining a jargon sentence with the lowest single rate from the historical data, and recording the jargon sentence as an abort sentence;

selecting a second target node in the conversational flow tree, wherein the second target node is a node in the conversational flow tree except the root node and the first target node;

fusing the conversational terminology sentence of the second target node with the suspension sentence to obtain a second fused sentence;

and replacing the conversational term sentence of the second target node with the second fusion sentence, setting a child node for the second target node, and recording the child node as an ending child node, wherein the ending child node is used for stopping voice interaction when a reply of a user based on the second homonymy sentence meets the judgment condition of the ending child node.

In a second aspect, the present application further provides a tactical update apparatus, comprising:

the first determining unit is used for determining a jargon sentence with the highest unit rate from historical data and recording the jargon sentence as a supplementary sentence, wherein the historical data comprises order information corresponding to each order in a voice dialing process and voice information corresponding to each order;

a first selecting unit, configured to select a first target node from a conversational flow tree, where the conversational flow tree includes at least two nodes, each node corresponds to a conversational term sentence, and the first target node is a node except a root node in the conversational flow tree;

the first fusion unit is used for fusing the jargon sentence of the first target node with the supplementary sentence to obtain a first fusion sentence;

a replacing unit, configured to replace the first fusion statement with the conversational statement of the first target node.

Optionally, the first determining unit includes:

a first determining module, configured to determine, in the historical data, the jargon sentence according to the voice information corresponding to the order in which the order result in the order information is successful;

the second determining module is used for determining the total number of orders corresponding to each conversational term sentence;

a third determining module, configured to determine that the order result corresponding to each conversational term sentence is a successful amount of orders;

the calculating module is used for carrying out quotient calculation according to the amount of orders and the total number of orders to obtain the order forming rate corresponding to the speech term sentence;

a fourth determining module, configured to determine, as the supplementary sentence, the conversational terminology sentence with the largest singleton rate in all the conversational sentences.

Optionally, the first determining module includes:

the judging submodule is used for judging whether the order result is successful or not according to the order information;

the obtaining submodule is used for obtaining the voice information corresponding to the order information if the order result is judged to be successful according to the order information;

and the processing submodule is used for processing by using a preset speech model according to the voice information to obtain the speech term sentence, the preset speech model is used for generating a speech corresponding to the semantics based on the semantics of the voice information, the voice information is obtained from knowledge bases, and each knowledge base comprises a user question and a corresponding response content.

Optionally, the first selecting unit includes:

the first determining module is used for respectively determining the first semantics of the supplementary sentences through a semantic analysis model;

a second determining module, configured to determine, through the semantic analysis model, a semantic of a conversational sentence corresponding to each node in the conversational flow tree, and write the semantic as a second semantic;

a third determining module, configured to determine, from the plurality of second semantics, the second semantics having the same category as the first semantics and serve as a target semantics;

a fourth determining module, configured to determine, according to the jargon sentence corresponding to the target semantic, the corresponding node as the first target node.

Optionally, the apparatus further comprises:

a second determining unit configured to determine a jargon sentence with the lowest single rate from the history data and to record the jargon sentence as an abort sentence;

a second selecting unit, configured to select a second target node from the conversational flow tree, where the second target node is a node in the conversational flow tree except for the root node and the first target node;

the second fusion unit is used for fusing the conversational terminology sentence of the second target node with the pause sentence to obtain a second fusion sentence;

and the operation unit is used for replacing the conversational term sentence of the second target node with the second fusion sentence, setting a child node for the second target node and recording the child node as an ending child node, wherein the ending child node is used for stopping voice interaction when a reply of a user based on the second homonymy sentence meets the judgment condition of the ending child node.

In a third aspect, an embodiment of the present application provides a storage medium, where the storage medium includes a stored program, and when the program runs, a device in which the storage medium is located is controlled to execute the tactical update method according to any one of the first aspect.

In a fourth aspect, embodiments of the present application provide a tactical update apparatus, the apparatus comprising a storage medium; and one or more processors, the storage medium coupled with the processors, the processors configured to execute program instructions stored in the storage medium; the program instructions when executed perform the method of updating a dialog according to any one of the first aspect.

By means of the technical scheme, the technical scheme provided by the application at least has the following advantages:

the application provides a speech updating method and device, which can determine a speech term sentence with the highest single rate from historical data, record the speech term sentence as a supplementary sentence, select a first target node from a speech process tree, fuse the speech term sentence of the first target node with the supplementary sentence to obtain a first fused sentence, and finally replace the speech term sentence of the first target node with the first fused sentence, so that a speech updating function is realized. Compared with the prior art, the implementation process of the method can be automatically implemented, namely, the processes of determining the supplementary sentences, selecting the first target node, fusing the dialect sentences of the first target node and the supplementary sentences and replacing the dialect sentences of the first target node with the fused first fused sentences can be automatically implemented, so that the dialect updating method provided by the application does not need to rely on labor, and the labor cost is saved. Meanwhile, because the historical data in the updating process contains the order information corresponding to each order in the voice dialing process and the voice information corresponding to each order, the dialect sentence which is most needed to be added in the updating process of the dialect flow tree, namely the supplement sentence, can be determined from the historical data based on the single-forming rate in the updating process of the dialect sentence, so that the supplement sentence can be used as the basis for updating the dialect, the problem of low accuracy of the dialect updating caused by misoperation or insufficient experience in the manual dialect updating process can be solved, and the accuracy of the dialect updating is improved. In addition, the speech process tree in the method includes at least two nodes, each node corresponds to a speech term sentence, and the first target node is a node except a root node in the speech process tree, that is, in the execution process of the speech updating method of the present application, a required first target node is selected from the speech process tree, and the speech term sentence in the first target node is updated, that is, no intermediate node is additionally added, so that the approximate structure of the speech process tree is not changed in the updating process, the structural characteristics of the speech process tree are ensured, the problem that the whole speech process tree fails due to the structural change of the speech process tree possibly caused by updating is avoided, and the accuracy of the speech updating is further improved.

The foregoing description is only an overview of the technical solutions of the present application, and the present application can be implemented according to the content of the description in order to make the technical means of the present application more clearly understood, and the following detailed description of the present application is given in order to make the above and other objects, features, and advantages of the present application more clearly understandable.

Drawings

The above and other objects, features and advantages of exemplary embodiments of the present application will become readily apparent from the following detailed description read in conjunction with the accompanying drawings. Several embodiments of the present application are illustrated by way of example and not by way of limitation in the figures of the accompanying drawings and in which like reference numerals refer to similar or corresponding parts and in which:

fig. 1 is a flowchart illustrating a method for updating dialog provided by an embodiment of the present application;

FIG. 1-A is a schematic diagram illustrating an implementation of a dialog updating method according to an embodiment of the present application;

FIG. 1-B is a schematic diagram illustrating an implementation of a dialog updating method according to an embodiment of the present application;

FIG. 2 is a flow chart of another speech updating method provided by the embodiment of the present application;

FIG. 2-A is a schematic diagram illustrating an implementation of a dialog updating method according to an embodiment of the present application;

FIG. 2-B is a schematic diagram illustrating an implementation of a dialog updating method according to an embodiment of the present application;

FIG. 3 is a block diagram illustrating components of a speech updating apparatus provided by an embodiment of the present application;

fig. 4 shows a block diagram of another speech updating apparatus provided in the embodiment of the present application.

Detailed Description

Exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

It is to be noted that, unless otherwise specified, technical or scientific terms used herein shall have the ordinary meaning as understood by those skilled in the art to which this application belongs.

An embodiment of the present application provides a conversation updating method, specifically as shown in fig. 1, the method includes:

101. the jargon sentence with the highest single rate is determined from the historical data and is recorded as a supplementary sentence.

The historical data comprises order information corresponding to each order in the voice dialing process and voice information corresponding to each order.

In this embodiment, a specific application scenario may be understood that, in the intelligent voice dialing process, the intelligent voice dialing system invokes a spoken sentence in the conversational flow tree to communicate with the user, and after the user makes an inquiry each time, the intelligent voice dialing system queries a corresponding answer, that is, a conversational term sentence, from the conversational flow tree, and then replies with the conversational term sentence. However, in practical applications, the number of the conversational sentences contained in the conversational flow tree is relatively limited, and the questions of the user are various, so that all the conversational terminology sentences cannot be included in one conversational flow tree, and at this time, different sentences are generally stored by using knowledge bases to be used as alternatives, and each knowledge base contains a set of question and answer sentences.

In view of this, the historical data in this step may be understood as a voice message record including all knowledge bases, and since the voices of different knowledge bases are different in probability of being triggered in practical application, after analysis, it is found that some voices are obviously increased to a single rate after being triggered, that is, after the intelligent voice dialing system generates a dialog statement based on some voices in the historical data, the number of orders placed by the user is obviously increased, that is, the single rate is higher. Therefore, in the implementation process of the speech updating method of this embodiment, it is first performed to determine which speech term sentence has the highest unit rate and mark it as a supplementary sentence, that is, the speech term sentence that needs to be used as an updating basis in the speech updating process.

102. A first target node is selected in the conversational flow tree.

The conversational flow tree comprises at least two nodes, each node corresponds to a conversational term sentence, and the first target node is a node except a root node in the conversational flow tree.

When the dialect with the highest singleton rate, that is, the supplementary sentence, is determined, it is next necessary to determine at which position in the dialect flow tree the supplementary sentence needs to be supplemented, so as to achieve the effect of dialect updating.

In this embodiment, the conversational flow tree may be as shown in fig. 1-a, where each conversational flow tree is a tree structure, each node corresponds to a conversational term sentence, and a branch option is a question asked by a user, so that in one tree structure, when the content of one of the nodes needs to be adjusted, it needs to determine which node is specifically selected for adjustment and modification.

In the specific implementation process, the selection of the first target node must exclude the root node, namely the "feed, hello" node in fig. 1-a, so that the root node is prevented from being modified as the initial sentence of voice dialing, and the situation that the intelligent voice dialing process is too abrupt is ensured.

In addition, the selection of the first target node is specifically selected, and nodes with similar semantics can be searched in the conversational flow tree according to the meaning of the supplementary statement, so that the convenience of subsequent fusion can be ensured. Of course, the method for selecting the first target node in the conversational flow tree in the specific application process includes, but is not limited to, the above method, and may also be selected by the user according to the user's needs.

103. And fusing the conversational terminology sentence of the first target node with the supplementary sentence to obtain a first fused sentence.

After the node needing replacement is determined, the first target node is obtained. At this time, the jargon sentence of the first target node can be extracted and fused with the supplementary sentence to obtain a first fused sentence. The specific fusion process can fuse the two sentences by using a conventional natural language processing model, so as to obtain a new sentence containing the meanings of the two sentences, namely the first fusion sentence.

For example, when it is determined that the supplementary sentence is "the product is nationally combined with insurance and is under national regulation", and the first target node's dialect sentence is "i went out a xxx product and ask for an interest to know" then the first target node and the second target node are fused based on the method of the present step to obtain "i went out a xxx product newly, and the product is nationally combined with insurance and is under national regulation and ask for an interest to know.

104. And replacing the first fusion statement with the conversational statement of the first target node.

After the first fusion statement is obtained, the first fusion statement can replace the preceding jargon statement of the first target node, so that the effect of tactical updating is realized. For example, as shown in fig. 1-B, in connection with the foregoing step example, "i went newly released xxx product, this product is a national joint guarantee, and is regulated by the country, and ask for interest to know" the phrase "i went newly released xxx product, ask for interest to know" in place of the previous first target node.

The embodiment provides a speech updating method, which can determine a speech clause with the highest single rate from historical data, record the speech clause as a supplementary sentence, select a first target node from a speech flow tree, fuse the speech clause of the first target node with the supplementary sentence to obtain a first fused sentence, and replace the speech clause of the first target node with the first fused sentence, so that a speech updating function is realized. Compared with the prior art, the implementation process of the method can be automatically implemented, namely, the processes of determining the supplementary sentences, selecting the first target node, fusing the dialect sentences of the first target node and the supplementary sentences and replacing the dialect sentences of the first target node with the fused first fused sentences can be automatically implemented, so that the dialect updating method provided by the application does not need to rely on labor, and the labor cost is saved. Meanwhile, because the historical data in the updating process contains the order information corresponding to each order in the voice dialing process and the voice information corresponding to each order, the dialect sentence which is most needed to be added in the updating process of the dialect flow tree, namely the supplement sentence, can be determined from the historical data based on the single-forming rate in the updating process of the dialect sentence, so that the supplement sentence can be used as the basis for updating the dialect, the problem of low accuracy of the dialect updating caused by misoperation or insufficient experience in the manual dialect updating process can be solved, and the accuracy of the dialect updating is improved. In addition, the speech process tree in the method includes at least two nodes, each node corresponds to a speech term sentence, and the first target node is a node except a root node in the speech process tree, that is, in the execution process of the speech updating method of the present application, a required first target node is selected from the speech process tree, and the speech term sentence in the first target node is updated, that is, no intermediate node is additionally added, so that the approximate structure of the speech process tree is not changed in the updating process, the structural characteristics of the speech process tree are ensured, the problem that the whole speech process tree fails due to the structural change of the speech process tree possibly caused by updating is avoided, and the accuracy of the speech updating is further improved.

To explain in more detail below, an embodiment of the present application provides another access control method, specifically as shown in fig. 2, the method includes:

201. the jargon sentence with the highest single rate is determined from the historical data and is recorded as a supplementary sentence.

Since the dialect update is aimed at achieving the accuracy of the update result in practical application, and the dialect statement is used for ordering the dialed user, the statement rate in the process of updating the dialect is an important basis in the process of updating, that is, which dialect statement has a high statement rate, the dialect update should be performed according to the dialect statement. Therefore, the process of determining the dialect with the highest single rate is very important.

Based on this, the present step can be performed as follows:

firstly, in the historical data, determining the conversational phrase sentence according to the voice information corresponding to the order with the order result being successful in the order information;

secondly, determining the total number of orders corresponding to each conversational term sentence;

then, determining the order result corresponding to each said conversational term sentence as a successful order number;

then, carrying out quotient calculation according to the amount of orders and the total number of orders to obtain the order rate corresponding to the speech term sentence;

finally, among all the conversational sentences, the conversational phrase sentence with the largest singleton rate is determined as the supplementary sentence.

In the above step, since the history data includes each order, and the corresponding voice information and order information, the order information may be divided into success and failure, that is, it can be determined whether the user has performed an order placing action based on a dialect used by the intelligent voice dialing system after each order is dialed based on the intelligent voice dialing system, if the order is placed, the order information of the order is successful, and if the order is not placed, the order information of the order is failed. Therefore, when determining the dialect sentence with the highest single rate, it is first necessary to determine which users in the order corresponding to the dialect sentences are placed. Then, after the voice information corresponding to the order with the order result being successful is obtained, the corresponding conversational terminology sentence can be determined based on the voice information, and then the order forming rate of each conversational sentence is determined, wherein the specific calculation mode of the order forming rate is to carry out quotient calculation through the successful order number of the order and all the order numbers related to the conversational sentence, namely the total order number, so as to obtain the corresponding percentage, namely the order forming rate. These jargon sentences are then ranked based on singleton rate to determine the highest singleton rate jargon sentence.

Further, in a specific implementation process, since only the speech information is contained in the history data, not the linguistic sentence, the speech information actually needs to be converted into the corresponding linguistic sentence in the process of determining the linguistic sentence with the highest single rate. Based on this, the determining the jargon sentence according to the voice information corresponding to the order with the order result being successful in the order information in the aforementioned step may be divided into the following steps when executing:

and processing by using a preset speech model according to the voice information to obtain the speech term sentence. The preset speech technology model is used for generating speech technology corresponding to the semantics based on the semantics of the voice information, the voice information is obtained from knowledge bases, and each knowledge base comprises a user question and corresponding response content.

In this embodiment, since not all the triggered voice messages can enable the user to place an order, in order to reduce unnecessary processing and save system resources in the process of processing the voice messages into conversational phrases, the order information of the voice messages may be first determined based on the above steps, and then the voice messages whose order results are successful are screened out, and then the voice messages whose order information is successful are processed by the preset conversational model, so that each obtained conversational phrase is a sentence that can prompt the user to place an order.

202. A first target node is selected in the conversational flow tree.

Specifically, the step may include:

firstly, respectively determining first semantics of the supplementary sentences through a semantic analysis model;

then, determining the semantics of the conversational sentence corresponding to each node in the conversational flow tree through the semantic analysis model, and recording the semantics as second semantics;

then, determining the second semantic meaning with the same category as the first semantic meaning from the plurality of second semantic meanings, and marking as a target semantic meaning;

and finally, determining the corresponding node as the first target node according to the speech term sentence corresponding to the target semantics.

Because the meaning of the conversational sentence of each node in the conversational flow tree is different, there are different categories, for example, as shown in fig. 1-a, the category of the root node "feed, hello" is a greeting; "I have newly introduced xxx products asking for interest in understanding" category is product introduction. Therefore, in determining the first target node, the selection may be based on the category of the sentence. That is, the semantic analysis model determines which node in the conversational flow tree has the same semantic category as the complementary sentence, and then determines which node is the corresponding node in the subsequent fusion of the conversational flow and replacement, that is, the first target node.

203. And fusing the conversational terminology sentence of the first target node with the supplementary sentence to obtain a first fused sentence.

In this embodiment, the manner of fusing the linguistic sentence and the supplementary sentence of the first target node may be the same as that in step 103 in the foregoing embodiment, and will not be described herein again.

204. And replacing the first fusion statement with the conversational statement of the first target node.

Specifically, in the replacement process, when the first fusion statement is used to replace the linguistic statement of the first target node, the linguistic statement of the first target node can be directly deleted, and then the first fusion statement is added to the first target node, so that the replacement effect is achieved.

205. The jargon sentence with the lowest single rate is determined from the historical data and is recorded as an abort sentence.

In practical application, it can be found through analysis that after some voices are triggered, the ordering situation of the user is obviously reduced, so in the embodiment, in order to ensure the execution efficiency of the intelligent voice dialing system, the utterance updating can be performed through the utterance term sentence with the lowest ordering rate, so that invalid telephone dialing is avoided, and the dialing of the next user is performed after the telephone is terminated as soon as possible. Therefore, in the implementation process of the present embodiment, the linguistic sentence with the least success rate can be found based on the historical data, and of course, the success rate can be 0, which is subject to the practical situation, and is not limited herein.

In addition, in this step, the determination manner of the term sentence determined to have the lowest success rate based on the historical data may be the same as that in step 201, and the difference between the two may be only to adjust the highest success rate to the lowest success rate.

206. A second target node is selected in the conversational flow tree.

Wherein the second target node is a node in the conversational flow tree other than the root node and the first target node.

In this step, the manner of selecting the second target node from the conversational flow tree may be the same as the manner of determining the first target node in step 202, that is, the second target node may also be determined by searching all nodes of the conversational flow tree for a node with the same semantic type based on the semantic type of the abort statement. In addition, it should be noted that, since the first target node is already a node replacing the first fusion statement, the node needs to be excluded and selected from the remaining nodes, in addition to excluding the root node.

207. And fusing the language term sentence of the second target node with the suspension sentence to obtain a second fused sentence.

Specifically, when the dialect statement and the abort statement of the second target node are merged, the merging manner may be the same as that in step 103 of the foregoing embodiment, and may be selected based on actual needs without further description.

208. And replacing the conversational phrase of the second target node with the second fusion statement, setting a child node for the second target node, and recording as an ending child node.

Wherein the ending sub-node is used for stopping voice interaction when the answer of the user based on the second homonymy sentence meets the judgment condition of the ending sub-node.

In this step, since the effect of the stop statement is to end the voice as soon as possible, after the second fused statement replaces the jargon statement of the second target node, a node capable of jumping out of the whole speech flow tree, that is, an end child node, is also needed to be added on the basis of the second target node, that is, when the answer of the user meets the jumping-out condition, the user directly goes to the end child node, so that the current dialing behavior is ended as soon as possible, the next telephone dialing is conducted as soon as possible, and the dialing efficiency is improved.

Specifically, as shown in fig. 2-a and 2-B, by analyzing the singleton rate of the historical data, when the fact that voice information is involved is found, that the student can apply for the application, which is not good, the singleton rate is lower after the student only applies for the application by the person with fixed income. At this time, it can be determined that the singleton rate corresponding to the voice message is the lowest based on the method of the foregoing steps, then the voice message can be processed into a corresponding termination statement "can be applied with fixed income", then the jargon sentence "i went new and released xxx product" corresponding to the second target node in the jargon flow tree is selected based on the type of the jargon meaning to be fused, and a new jargon "i went new and released xxx product" is obtained, and this product can be applied with fixed income, and then it is requested to be applied with interest to know xxx ", that is, the second fused statement. Then, the second fusion statement replaces the original dialect statement, and meanwhile, a jump-out node is added to the second target node, namely, the end child node and the jump-out condition are ' student ', ' so that the effect of dialect updating is realized.

In order to achieve the above object, according to another aspect of the present application, an embodiment of the present application further provides a storage medium, where the storage medium includes a stored program, where the program, when executed, controls a device on which the storage medium is located to perform the above-mentioned session update method.

In order to achieve the above object, according to another aspect of the present application, an embodiment of the present application further provides a dialog updating apparatus, which includes a storage medium; and one or more processors, the storage medium coupled with the processors, the processors configured to execute program instructions stored in the storage medium; the program instructions when executed perform the above-described dialog updating method.

Further, as an implementation of the method shown in fig. 1 and fig. 2, another embodiment of the present application further provides a speech updating apparatus. The embodiment of the speech updating apparatus corresponds to the foregoing method embodiment, and for convenience of reading, details of the foregoing method embodiment are not repeated in this embodiment of the speech updating apparatus again, but it should be clear that the apparatus in this embodiment can correspondingly implement all the contents of the foregoing method embodiment. The technical update device mainly aims at solving the problem that the accuracy of the current technical update is low, and specifically as shown in fig. 3, the technical update device comprises:

a first determining unit 31, configured to determine a jargon sentence with the highest unit rate from historical data, as a supplementary sentence, where the historical data includes order information corresponding to each order in a voice-based dialing process and voice information corresponding to each order;

a first selecting unit 32, configured to select a first target node from a conversational flow tree, where the conversational flow tree includes at least two nodes, each node corresponds to a conversational term sentence, and the first target node is a node except a root node in the conversational flow tree;

a first fusing unit 33, configured to fuse the conversational phrase of the first target node with the supplementary sentence to obtain a first fused sentence;

a replacing unit 34, configured to replace the first fusion statement with the conversational statement of the first target node.

Further, as shown in fig. 4, the first determining unit 31 includes:

the first determining module 311 may be configured to determine, in the historical data, the conversational phrase sentence according to the voice information corresponding to the order in which the order result is successful in the order information;

a second determining module 312, configured to determine a total number of orders corresponding to each of the conversational terminology sentences;

a third determining module 313, configured to determine that the order result corresponding to each conversational term sentence is a successful amount of orders;

a calculating module 314, configured to perform quotient calculation according to the amount of orders and the total number of orders, so as to obtain the order forming rate corresponding to the conversational terminology sentence;

a fourth determining module 315, configured to determine, as the supplementary sentence, the conversational phrase with the largest singleton rate in all the conversational sentences.

Further, as shown in fig. 4, the first determining module 311 includes:

the determining sub-module 3111 may be configured to determine whether the order result is successful according to the order information;

the obtaining sub-module 3112 is configured to, if the order result is determined to be successful according to the order information, obtain the voice information corresponding to the order information;

the processing sub-module 3113 may be configured to perform processing according to the voice information by using a preset speech model to obtain the speech term sentence, where the preset speech model may be configured to generate a speech corresponding to the semantics based on the semantics of the voice information, and the voice information is obtained from knowledge bases, where each knowledge base includes a user question and a corresponding response content.

Further, as shown in fig. 4, the first selecting unit 32 includes:

a first determining module 321, configured to determine first semantics of the supplementary sentences through a semantic analysis model respectively;

a second determining module 322, configured to determine, through the semantic analysis model, a semantic of a conversational sentence corresponding to each node in the conversational flow tree, and record the semantic as a second semantic;

a third determining module 323, configured to determine, from the plurality of second semantics, the second semantics having the same category as the first semantics, and to label the second semantics as a target semantics;

a fourth determining module 324, configured to determine the corresponding node as the first target node according to the jargon sentence corresponding to the target semantic.

Further, as shown in fig. 4, the apparatus further includes:

a second determining unit 35, configured to determine a jargon sentence with the lowest single rate from the historical data, and to record the jargon sentence as an abort sentence;

a second selecting unit 36, configured to select a second target node from the conversational flow tree, where the second target node is a node in the conversational flow tree except for the root node and the first target node;

a second fusion unit 37, configured to fuse the jargon sentence of the second target node with the suspension statement to obtain a second fusion statement;

an operation unit 38, configured to replace the conversational phrase of the second target node with the second fused phrase, and set a child node for the second target node, which is denoted as an end child node, where the end child node may be configured to terminate voice interaction when a user answers to the end child node based on the second synonym phrase.

The embodiment of the application provides a speech updating method and device, which can determine a speech clause with the highest single rate from historical data, record the speech clause as a supplementary sentence, select a first target node from a speech flow tree, fuse the speech clause of the first target node with the supplementary sentence to obtain a first fused sentence, and finally replace the speech clause of the first target node with the first fused sentence, so that a speech updating function is realized. Compared with the prior art, the implementation process of the method can be automatically implemented, namely, the processes of determining the supplementary sentences, selecting the first target node, fusing the dialect sentences of the first target node and the supplementary sentences and replacing the dialect sentences of the first target node with the fused first fused sentences can be automatically implemented, so that the dialect updating method provided by the application does not need to rely on labor, and the labor cost is saved. Meanwhile, because the historical data in the updating process contains the order information corresponding to each order in the voice dialing process and the voice information corresponding to each order, the dialect sentence which is most needed to be added in the updating process of the dialect flow tree, namely the supplement sentence, can be determined from the historical data based on the single-forming rate in the updating process of the dialect sentence, so that the supplement sentence can be used as the basis for updating the dialect, the problem of low accuracy of the dialect updating caused by misoperation or insufficient experience in the manual dialect updating process can be solved, and the accuracy of the dialect updating is improved. In addition, the speech process tree in the method includes at least two nodes, each node corresponds to a speech term sentence, and the first target node is a node except a root node in the speech process tree, that is, in the execution process of the speech updating method of the present application, a required first target node is selected from the speech process tree, and the speech term sentence in the first target node is updated, that is, no intermediate node is additionally added, so that the approximate structure of the speech process tree is not changed in the updating process, the structural characteristics of the speech process tree are ensured, the problem that the whole speech process tree fails due to the structural change of the speech process tree possibly caused by updating is avoided, and the accuracy of the speech updating is further improved.

The embodiment of the application provides a storage medium, which comprises a stored program, wherein when the program runs, the device where the storage medium is located is controlled to execute the above-mentioned dialect updating method.

The storage medium may include volatile memory in a computer readable medium, random Access Memory (RAM) and/or nonvolatile memory such as Read Only Memory (ROM) or flash memory (flash RAM), and the memory includes at least one memory chip.

The embodiment of the application also provides a dialoging updating device, which comprises a storage medium; and one or more processors, the storage medium coupled with the processors, the processors configured to execute program instructions stored in the storage medium; the program instructions when executed perform the above-described dialog updating method.

The embodiment of the application provides equipment, the equipment comprises a processor, a memory and a program which is stored on the memory and can run on the processor, and the following steps are realized when the processor executes the program: determining a conversational phrase with the highest unit rate from historical data, and recording the conversational phrase as a supplementary phrase, wherein the historical data comprises order information corresponding to each order in a voice dialing process and voice information corresponding to each order; selecting a first target node from a conversational flow tree, wherein the conversational flow tree comprises at least two nodes, each node corresponds to a conversational term sentence, and the first target node is a node except a root node in the conversational flow tree; fusing the speech clause of the first target node with the supplementary sentence to obtain a first fused sentence; and replacing the first fusion statement with the conversational statement of the first target node.

Further, the jargon sentence determined to be the highest single rate from the historical data includes:

determining the order result corresponding to each said conversational term sentence to be a successful order number;

carrying out quotient calculation according to the amount of orders and the total number of orders to obtain the order rate corresponding to the speech term sentence;

Further, the determining the jargon sentence according to the voice information corresponding to the order in which the order result is successful in the order information includes:

Further, the selecting a first target node in the conversational flow tree includes:

Further, after the replacing the first fused sentence with the jargon sentence of the first target node, the method further comprises:

and replacing the conversational terminology sentence of the second target node with the second fusion sentence, setting a child node for the second target node, and recording the child node as an ending child node, wherein the ending child node is used for stopping voice interaction when a reply of the user based on the second harmony sentence meets the judgment condition of the ending child node.

The present application further provides a computer program product adapted to perform program code for initializing the following method steps when executed on a data processing device: determining a jargon sentence with the highest unit forming rate from historical data, and recording the jargon sentence as a supplementary sentence, wherein the historical data comprises order information corresponding to each order in a voice dialing process and voice information corresponding to each order; selecting a first target node from a conversational flow tree, wherein the conversational flow tree comprises at least two nodes, each node corresponds to a conversational term sentence, and the first target node is a node except a root node in the conversational flow tree; fusing the conversational terminology sentence of the first target node with the supplementary sentence to obtain a first fused sentence; and replacing the first fusion statement with the conversational statement of the first target node.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.

Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus 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, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another identical element in a process, method, article, or apparatus that comprises the element.

The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A conversational updating method, the method comprising:

fusing the conversational terminology sentence of the first target node with the supplementary sentence to obtain a first fused sentence;

2. The method of claim 1, wherein the determining from the historical data the highest-rate jargon comprises:

and determining the conversational terminology sentence with the highest singleton rate as the supplementary sentence in all the conversational sentences.

3. The method according to claim 2, wherein the determining the jargon sentence according to the speech information corresponding to the order in the order information whose order result is successful comprises:

and processing the voice information by using a preset talk operation model according to the voice information to obtain the talk term sentence, wherein the preset talk operation model is used for generating a talk operation corresponding to the semantics based on the semantics of the voice information, the voice information is obtained from knowledge bases, and each knowledge base comprises a user question and a corresponding response content.

4. The method of claim 1, wherein selecting the first target node in the conversational flow tree comprises:

and determining the corresponding node as the first target node according to the speech term sentence corresponding to the target semantics.

5. The method of any of claims 1-4, wherein after the replacing the first fused statement with the jargon statement of the first target node, the method further comprises:

6. A speech update apparatus, the apparatus comprising:

7. The apparatus according to claim 6, wherein the first determining unit comprises:

8. The apparatus of claim 7, wherein the first determining module comprises:

9. The apparatus of claim 6, wherein the first selecting unit comprises:

the first determining module is used for respectively determining first semantics of the supplementary sentences through a semantic analysis model;

a third determining module, configured to determine, from the plurality of second semantics, the second semantics having the same category as the first semantics and take the second semantics as a target semantics;

10. The apparatus according to any one of claims 6-9, further comprising:

a second selecting unit, configured to select a second target node from the conversational process tree, where the second target node is a node in the conversational process tree except for the root node and the first target node;

the second fusion unit is used for fusing the jargon sentence of the second target node with the suspension sentence to obtain a second fusion sentence;

11. A storage medium comprising a stored program, wherein the program, when executed, controls an apparatus in which the storage medium is located to perform the method of updating dialogues of any of claims 1 to 5.

12. A tactical update apparatus, comprising a storage medium; and one or more processors, the storage medium coupled with the processors, the processors configured to execute program instructions stored in the storage medium; the program instructions when executed perform the method of any of claims 1 to 5.