CN112395402A - Depth model-based recommended word generation method and device and computer equipment - Google Patents
Depth model-based recommended word generation method and device and computer equipment Download PDFInfo
- Publication number
- CN112395402A CN112395402A CN202011304769.4A CN202011304769A CN112395402A CN 112395402 A CN112395402 A CN 112395402A CN 202011304769 A CN202011304769 A CN 202011304769A CN 112395402 A CN112395402 A CN 112395402A
- Authority
- CN
- China
- Prior art keywords
- recommended
- question
- depth model
- similar
- dialect
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3343—Query execution using phonetics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
- G06F40/35—Discourse or dialogue representation
Abstract
The invention discloses a recommended dialect generation method and device based on a depth model and computer equipment. The method comprises the following steps: the method comprises the steps that a first multi-dimensional feature vector of a first recommended dialect and at least one second multi-dimensional feature vector of a second recommended dialect are used as input of a preset depth model, the probability of each recommended dialect is obtained through the output of the preset depth model, and the preset depth model is obtained by classifying various types of dialects based on a naive Bayes classifier; and generating a plurality of recommended dialogs of the current problem according to the probability of each recommended dialog. Therefore, by adopting the embodiment of the application, enough recommended dialogues associated with the current problems of the user can be generated, the generation process of the recommended dialogues is automatically completed, and the tedious complicated flow of manually configuring the recommended dialogues based on the current problems is avoided.
Description
Technical Field
The invention relates to the technical field of call center communication, in particular to a recommended word generation method and device based on a depth model and computer equipment.
Background
At present, the number of intelligent voice systems on the market is large, and the intelligent voice system is mainly applied to a plurality of fields such as intelligent marketing, intelligent customer service and the like. The intelligent marketing mainly uses a task-type multi-turn dialogue system to guide users and know the requirements of the users by specific dialogues, so that the purpose of accurately marketing the users is achieved. However, such a system requires professional service personnel to perform the speech configuration, and the professional requirements on the service personnel are high. Moreover, the labor of the personnel is limited, a large number of words and skills cannot be configured to deal with the inquiry of the client, and the user cannot be accurately identified due to insufficient corpus when the intention of the user is judged, so that the user cannot respond correctly.
Disclosure of Invention
Based on this, it is necessary to provide a method, an apparatus, a computer device and a storage medium for generating a recommendation language based on a depth model, aiming at the problem of low accuracy of the existing recommendation language.
In a first aspect, an embodiment of the present application provides a depth model-based recommended speech generation method, where the method includes:
acquiring a first recommended dialect of a current problem of a user, wherein the first recommended dialect is at least one recommended dialect which is configured in a system in advance;
acquiring a similar problem set similar to the current problem and a recommendation language of any similar problem in the similar problem set, and taking the recommendation language of any similar problem as a second recommendation language;
taking the first multidimensional feature vector of the first recommended dialect and at least one second multidimensional feature vector of the second recommended dialect as the input of a preset depth model, and obtaining the probability of each recommended dialect through the output of the preset depth model, wherein the preset depth model is obtained by classifying various types of dialects based on a naive Bayes classifier;
and generating a plurality of recommended dialogs of the current problem according to the probability of each recommended dialog.
In one embodiment, before the obtaining a set of similar questions similar to the user question, the method further comprises:
performing semantic analysis on voice data to obtain a plurality of keywords of a current problem, wherein the content of the voice data is the current problem of a user;
obtaining a plurality of similar problems similar to the current problem according to the plurality of keywords;
and forming a similar problem set by a plurality of similar problems.
In one embodiment, prior to said semantically analyzing the speech data, the method further comprises:
performing semantic analysis on the voice data to obtain a problem text of the current problem of the user;
and searching a recommended word operation corresponding to the question text in the system by using the question text as a keyword, and using the recommended word operation as the first recommended word operation.
In one embodiment, prior to said semantically analyzing said speech data, said method further comprises:
and judging whether a pre-configured automatic question-answering program is started or not according to the voice data, wherein the automatic question-answering program is used for generating at least one associated question in the voice data, and the associated question is a question with the association degree with the current question being larger than or equal to a preset association degree threshold value.
In one embodiment, the determining whether to start a preconfigured automatic question answering program according to the voice data includes:
under the condition that the voice data does not comprise associated questions, starting the automatic question answering program until the voice data comprises at least one associated question;
and under the condition that the voice data comprises the associated question, neglecting to start the automatic question answering program.
In one embodiment, the method further comprises:
randomly pushing any one of the generated plurality of recommended dialogs to the user.
In one embodiment, the method further comprises:
acquiring historical behavior data of the user;
analyzing the historical behavior data to obtain an analysis result;
generating at least one preferred conversation recommendation operation with the matching degree of the analysis result being larger than the preset matching degree according to the analysis result and the plurality of conversation recommendation operations;
randomly pushing at least one preferred recommended utterance to the user.
In a second aspect, an embodiment of the present application provides a depth model-based recommended speech generation apparatus, including:
the system comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring a first recommended dialect of a current problem of a user, and the first recommended dialect is at least one recommended dialect which is configured in advance in the system; and
acquiring a similar problem set similar to the current problem and a recommendation language of any similar problem in the similar problem set, and taking the recommendation language of any similar problem as a second recommendation language;
the processing unit is used for taking the first multidimensional feature vector of the first recommended speech and at least one second multidimensional feature vector of the second recommended speech, which are acquired by the acquiring unit, as the input of a preset depth model, and obtaining the probability of each recommended speech through the output of the preset depth model, wherein the preset depth model is obtained by classifying various types of speech based on a naive Bayes classifier;
and the recommended dialect generating unit is used for generating a plurality of recommended dialects of the current problem according to the probability of each recommended dialect output by the processing unit.
In a third aspect, embodiments of the present application provide a computer device, including a memory and a processor, where the memory stores computer-readable instructions, and the computer-readable instructions, when executed by the processor, cause the processor to perform the above-mentioned method steps.
In a fourth aspect, embodiments of the present application provide a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the above-mentioned method steps.
The technical scheme provided by the embodiment of the application can have the following beneficial effects:
in the embodiment of the application, a first multidimensional feature vector of a first recommended dialect and at least one second multidimensional feature vector of a second recommended dialect are used as the input of a preset depth model, the probability of each recommended dialect is obtained through the output of the preset depth model, and the preset depth model is obtained by classifying various types of dialects based on a naive Bayes classifier; and generating a plurality of recommended dialogs of the current problem according to the probability of each recommended dialog. Therefore, by adopting the embodiment of the application, enough recommended dialogues associated with the current problems of the user can be generated, the generation process of the recommended dialogues is automatically completed, and the tedious complicated flow of manually configuring the recommended dialogues based on the current problems is avoided.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention.
FIG. 1 is a diagram of an implementation environment for a depth model-based conversational recommendation generation method provided in one embodiment;
FIG. 2 is a block diagram showing an internal configuration of a computer device according to an embodiment;
FIG. 3 is a flowchart illustrating a method for generating recommended dialogs based on a depth model according to an embodiment of the present disclosure;
fig. 4 is a schematic structural diagram of a recommended speech generation apparatus based on a depth model according to an embodiment of the present disclosure.
Detailed Description
The following description and the drawings sufficiently illustrate specific embodiments of the invention to enable those skilled in the art to practice them.
It should be understood that the described embodiments are only some embodiments of the invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Alternative embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.
Fig. 1 is a diagram of an implementation environment of a depth model-based recommended dialog generation method provided in an embodiment, as shown in fig. 1, in which a computer device 110 and a terminal 120 are included.
The computer device 110 is a depth model-based recommended speech generation device, and the depth model-based recommended speech generation tool is installed on the computer device 110.
It should be noted that the terminal 120 and the computer device 110 may be, but are not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, and the like. The computer device 110 and the terminal 110 may be connected through bluetooth, USB (Universal Serial Bus), or other communication connection methods, which is not limited herein.
FIG. 2 is a diagram showing an internal configuration of a computer device according to an embodiment. As shown in fig. 2, the computer device includes a processor, a non-volatile storage medium, a memory, and a network interface connected through a system bus. The non-volatile storage medium of the computer device stores an operating system, a database and computer readable instructions, the database can store control information sequences, and the computer readable instructions can enable the processor to realize a recommended dialogs generation method based on the depth model when being executed by the processor. The processor of the computer device is used for providing calculation and control capability and supporting the operation of the whole computer device. The memory of the computer device may have stored therein computer readable instructions that, when executed by the processor, may cause the processor to perform a method of recommended speech generation based on a depth model. The network interface of the computer device is used for connecting and communicating with the terminal. Those skilled in the art will appreciate that the architecture shown in fig. 2 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
As shown in fig. 3, an embodiment of the present disclosure provides a recommended word technique generating method based on a depth model, which is applied to a server side, and specifically includes the following method steps:
s302: acquiring a first recommended dialect of a current problem of a user;
in an embodiment of the present application, the first recommendation dialog is at least one recommendation dialog configured in the system in advance.
S304: acquiring a similar problem set similar to the current problem and a recommendation language of any similar problem in the similar problem set, and taking the recommendation language of any similar problem as a second recommendation language;
in one possible implementation, before obtaining a set of similar questions similar to the user question, the method further comprises the steps of:
performing semantic analysis on the voice data to obtain a plurality of keywords of the current problem, wherein the content of the voice data is the current problem of the user;
obtaining a plurality of similar problems similar to the current problem according to the plurality of keywords;
and forming a similar problem set by a plurality of similar problems.
In the embodiment of the application, a similar problem set associated with the current problem can be generated, so that a similar recommendation language can be searched according to any one similar problem in the similar problem set, and the similar recommendation language is used as at least one candidate recommendation language of the current problem; therefore, the precision of the speech recommending technology can be effectively improved, and the user experience is improved.
In one possible implementation, before performing semantic analysis on the speech data, the method further comprises the steps of:
performing semantic analysis on the voice data to obtain a problem text of the current problem of the user;
and searching a recommended word operation corresponding to the question text in the system by taking the question text as a keyword, and taking the recommended word operation as a first recommended word operation.
In the embodiment of the application, the first recommendation language is determined according to the question text as the key word, so that the relevance between the searched recommendation language and the user question is high, and the matching degree between the recommendation language and the user question and the precision of the recommendation language are improved.
In one possible implementation, before performing semantic analysis on the speech data, the method further comprises the steps of:
and judging whether a pre-configured automatic question-answering program is started or not according to the voice data, wherein the automatic question-answering program is used for generating at least one associated question in the voice data, and the associated question is a question with the association degree with the current question being larger than or equal to a preset association degree threshold value.
In the embodiment of the application, the automatic question answering program is introduced, so that the dialogue with the user can be continuously performed until at least one associated question is generated in the voice data, and the intelligence of the dialogue recommendation method can be effectively improved.
In different application scenarios, a preset relevance threshold between the relevant question and the current question may be configured in advance.
Specifically, if the search range of the associated problem corresponding to the current problem is to be expanded, the preset association threshold between the two may be set to be smaller, as long as there is some association; on the contrary, if the association problem closely related to the current problem is desired to be obtained, the preset association threshold between the two may be set to be larger, and the preset association threshold in different application scenarios is not specifically limited herein, and may be adjusted according to the requirements of different application scenarios, which is not described herein again.
In the embodiment of the application, judging whether to start a pre-configured automatic question answering program or not according to the voice data comprises the following steps:
under the condition that the voice data does not comprise the associated questions, starting an automatic question answering program until the voice data comprises at least one associated question;
in the case where the associated question is included in the voice data, the start of the automatic question-answering program is ignored.
S306: the method comprises the steps that a first multi-dimensional feature vector of a first recommended dialect and at least one second multi-dimensional feature vector of a second recommended dialect are used as input of a preset depth model, the probability of each recommended dialect is obtained through the output of the preset depth model, and the preset depth model is obtained by classifying various types of dialects based on a naive Bayes classifier;
in the embodiment of the application, in order to ensure the accuracy of the preset depth model, a naive Bayes classifier is introduced to classify various types of dialects.
The naive bayes classifier assumes that the components of the feature vector are independent of each other, and given the feature vector x of a sample, the probability that the sample belongs to a class is:
since the components of the feature vectors are assumed to be independent of each other, there are:
wherein Z is a normalization factor. In an actual application scenario, the probabilities corresponding to different classes may be configured according to the number of each class of samples in the training samples, for example, the first class of samples accounts for forty percent and the second class accounts for sixty percent in the training samples, so that the probability of the first class may be set to be 0.4, and the probability of the second class may be set to be 0.6.
S308: and generating a plurality of recommended dialogs of the current problem according to the probability of each recommended dialog.
In one possible implementation manner, in obtaining a plurality of recommended dialogues, any one of the generated plurality of recommended dialogues may be randomly pushed to the user; therefore, the generation process of the dialect recommendation is simplified, the process of further screening in a plurality of dialects is reduced, the recommendation algorithm is optimized, and the dialect recommendation process is simplified.
In a possible implementation manner, the recommendation dialog may also be pushed to the user by the following method, which is specifically described as follows:
acquiring historical behavior data of a user;
analyzing the historical behavior data to obtain an analysis result;
generating at least one preferred conversation recommendation operation with the matching degree of the analysis result being greater than the preset matching degree according to the analysis result and the plurality of conversation recommendation operations;
at least one preferred recommended dialog is randomly pushed to the user.
Compared with a random push-based word recommendation method, the multiple generated recommended words are further screened, at least one optimized recommended word is selected preferably, and at least one optimized recommended word is pushed to the user randomly, so that the word recommendation process can be simplified, unnecessary complicated recommendation operation is reduced, the recommendation efficiency is improved, the word recommended to the user is ensured to be at least the optimized recommended word, the matching degree with the user problem is improved, the precision of the recommended word is improved, and the user experience is improved.
In a possible implementation manner, the probabilities of the multiple recommended dialogues obtained in S308 are further calculated to obtain a recommended dialogue with the highest probability, and the recommended dialogues with the highest probability are pushed to the terminal of the user; because the optimal recommended dialogs are pushed to the user, the experience degree of the user is improved.
In the embodiment of the present application, the calculation method for calculating the probabilities of the multiple recommended dialogues is a conventional method, and is not described herein again.
In the recommended speech generation method provided by the embodiment of the present disclosure, the depth model may be a BERT model. It should be noted that BERT is a Bidirectional Encoder characterization from transducers (Bidirectional Encoder), which is a novel language model; BERT is a fine-tuning-based multi-layer bidirectional transformer encoder. The BERT-based model is a conventional model and is not described in detail herein.
In the embodiment of the disclosure, a first multidimensional feature vector of a first recommended dialect and at least one second multidimensional feature vector of a second recommended dialect are used as input of a preset depth model, the probability of each recommended dialect is obtained through the output of the preset depth model, and the preset depth model is a model obtained by classifying various types of dialects based on a naive Bayes classifier; and generating a plurality of recommended dialogs of the current problem according to the probability of each recommended dialog. Therefore, by adopting the embodiment of the application, enough recommended dialogues associated with the current problems of the user can be generated, the generation process of the recommended dialogues is automatically completed, and the tedious complicated flow of manually configuring the recommended dialogues based on the current problems is avoided.
The following is an embodiment of a recommended speech generation apparatus based on a depth model according to the present invention, which can be used to execute an embodiment of a recommended speech generation method based on a depth model according to the present invention. For details not disclosed in the embodiments of the recommended speech generation device based on depth model of the present invention, please refer to the embodiments of the recommended speech generation method based on depth model of the present invention.
Referring to fig. 4, a schematic structural diagram of a depth model-based recommended dialogs generation apparatus according to an exemplary embodiment of the present invention is shown. The recommendation speech generation means may be implemented as all or part of the terminal in software, hardware or a combination of both. The recommended speech generation apparatus includes an acquisition unit 402, a processing unit 404, and a recommended speech generation unit 406.
Specifically, the obtaining unit 402 is configured to obtain a first recommended language of a current problem of the user, where the first recommended language is at least one recommended language preconfigured in the system; and
acquiring a similar problem set similar to the current problem and a recommendation language of any similar problem in the similar problem set, and taking the recommendation language of any similar problem as a second recommendation language;
a processing unit 404, configured to take the first multidimensional feature vector of the first recommended speech and the at least one second multidimensional feature vector of the second recommended speech, which are acquired by the acquisition unit 402, as inputs of a preset depth model, and obtain probabilities of the respective recommended speech through an output of the preset depth model, where the preset depth model is a model obtained by classifying various types of speech based on a naive bayes classifier;
a recommended utterance generation unit 406, configured to generate a plurality of recommended utterances for the current question according to the probabilities of the respective recommended utterances output by the processing unit 404.
Optionally, before the obtaining unit 402 obtains a similar question set similar to the user question, the processing unit 404 is further configured to:
performing semantic analysis on the voice data to obtain a plurality of keywords of the current problem, wherein the content of the voice data is the current problem of the user;
obtaining a plurality of similar problems similar to the current problem according to the plurality of keywords;
and forming a similar problem set by a plurality of similar problems.
Optionally, before performing semantic analysis on the voice data, the processing unit 404 is further configured to:
performing semantic analysis on the voice data to obtain a problem text of the current problem of the user;
and searching a recommended word operation corresponding to the question text in the system by taking the question text as a keyword, and taking the recommended word operation as a first recommended word operation.
Optionally, the apparatus further comprises:
a judging unit (not shown in fig. 4) configured to judge whether to start a pre-configured automatic question-answering program according to the voice data before the processing unit 404 performs semantic analysis on the voice data, wherein the automatic question-answering program is configured to generate at least one associated question in the voice data, and the associated question is a question with a relevance degree greater than or equal to a preset relevance degree threshold value with respect to the current question.
Optionally, the determining unit is specifically configured to:
under the condition that the voice data does not comprise the associated questions, starting an automatic question answering program until the voice data comprises at least one associated question;
in the case where the associated question is included in the voice data, the start of the automatic question-answering program is ignored.
Optionally, the apparatus further comprises:
a pushing unit (not shown in fig. 4) configured to randomly push any one of the plurality of recommended dialogs generated by the recommended dialogs generating unit 406 to the user.
Optionally, the obtaining unit 402 is further configured to obtain historical behavior data of the user;
the processing unit 404 is further configured to analyze the historical behavior data to obtain an analysis result;
the recommended dialogs generating unit 406 is further configured to: generating at least one preferred word recommendation technique with a matching degree with the analysis result larger than a preset matching degree according to the analysis result of the processing unit 404 and the plurality of word recommendation techniques;
the pushing unit is further used for randomly pushing the at least one preferred recommended dialog to the user.
It should be noted that, when the recommendation language generation device based on the depth model provided in the foregoing embodiment executes the recommendation language generation method based on the depth model, only the division of the functional modules is illustrated, and in practical applications, the function distribution may be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules, so as to complete all or part of the functions described above. In addition, the device for generating a recommended speech based on a depth model and the method for generating a recommended speech based on a depth model provided in the above embodiments belong to the same concept, and the details of the implementation process are described in the embodiments of the method for generating a recommended speech based on a depth model, and are not described herein again.
In the embodiment of the present disclosure, the processing unit is configured to use the first multidimensional feature vector of the first recommended speech and the second multidimensional feature vector of the at least one second recommended speech, which are acquired by the acquisition unit, as inputs of a preset depth model, and obtain probabilities of the respective recommended speech through an output of the preset depth model, where the preset depth model is a model obtained by classifying various types of speech based on a naive bayes classifier; and the recommended speech generation unit is used for generating a plurality of recommended speech of the current problem according to the probability of each recommended speech output by the processing unit. Therefore, by adopting the embodiment of the application, enough recommended dialogues associated with the current problems of the user can be generated, the generation process of the recommended dialogues is automatically completed, and the tedious complicated flow of manually configuring the recommended dialogues based on the current problems is avoided.
In one embodiment, a computer device is proposed, the computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the computer program: acquiring a first recommended dialect of a current problem of a user, wherein the first recommended dialect is at least one recommended dialect which is configured in advance in a system; acquiring a similar problem set similar to the current problem and a recommendation language of any similar problem in the similar problem set, and taking the recommendation language of any similar problem as a second recommendation language; the method comprises the steps that a first multi-dimensional feature vector of a first recommended dialect and at least one second multi-dimensional feature vector of a second recommended dialect are used as input of a preset depth model, the probability of each recommended dialect is obtained through the output of the preset depth model, and the preset depth model is obtained by classifying various types of dialects based on a naive Bayes classifier; and generating a plurality of recommended dialogs of the current problem according to the probability of each recommended dialog.
In one embodiment, a storage medium is provided that stores computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of: acquiring a first recommended dialect of a current problem of a user, wherein the first recommended dialect is at least one recommended dialect which is configured in advance in a system; acquiring a similar problem set similar to the current problem and a recommendation language of any similar problem in the similar problem set, and taking the recommendation language of any similar problem as a second recommendation language; the method comprises the steps that a first multi-dimensional feature vector of a first recommended dialect and at least one second multi-dimensional feature vector of a second recommended dialect are used as input of a preset depth model, the probability of each recommended dialect is obtained through the output of the preset depth model, and the preset depth model is obtained by classifying various types of dialects based on a naive Bayes classifier; and generating a plurality of recommended dialogs of the current problem according to the probability of each recommended dialog.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the computer program is executed. The storage medium may be a non-volatile storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a Random Access Memory (RAM).
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method for recommended word generation based on a depth model, the method comprising:
acquiring a first recommended dialect of a current problem of a user, wherein the first recommended dialect is at least one recommended dialect which is configured in a system in advance;
acquiring a similar problem set similar to the current problem and a recommendation language of any similar problem in the similar problem set, and taking the recommendation language of any similar problem as a second recommendation language;
taking the first multidimensional feature vector of the first recommended dialect and at least one second multidimensional feature vector of the second recommended dialect as the input of a preset depth model, and obtaining the probability of each recommended dialect through the output of the preset depth model, wherein the preset depth model is obtained by classifying various types of dialects based on a naive Bayes classifier;
and generating a plurality of recommended dialogs of the current problem according to the probability of each recommended dialog.
2. The method of claim 1, wherein prior to said obtaining a set of similar questions similar to said user question, said method further comprises:
performing semantic analysis on voice data to obtain a plurality of keywords of a current problem, wherein the content of the voice data is the current problem of a user;
obtaining a plurality of similar problems similar to the current problem according to the plurality of keywords;
and forming a similar problem set by a plurality of similar problems.
3. The method of claim 2, wherein prior to said semantically analyzing the speech data, the method further comprises:
performing semantic analysis on the voice data to obtain a problem text of the current problem of the user;
and searching a recommended word operation corresponding to the question text in the system by using the question text as a keyword, and using the recommended word operation as the first recommended word operation.
4. The method of claim 3, wherein prior to the semantically analyzing the speech data, the method further comprises:
and judging whether a pre-configured automatic question-answering program is started or not according to the voice data, wherein the automatic question-answering program is used for generating at least one associated question in the voice data, and the associated question is a question with the association degree with the current question being larger than or equal to a preset association degree threshold value.
5. The method of claim 4, wherein determining whether to initiate a pre-configured automated question-and-answer procedure based on the voice data comprises:
under the condition that the voice data does not comprise associated questions, starting the automatic question answering program until the voice data comprises at least one associated question;
and under the condition that the voice data comprises the associated question, neglecting to start the automatic question answering program.
6. The method of claim 1, further comprising:
randomly pushing any one of the generated plurality of recommended dialogs to the user.
7. The method of claim 1, further comprising:
acquiring historical behavior data of the user;
analyzing the historical behavior data to obtain an analysis result;
generating at least one preferred conversation recommendation operation with the matching degree of the analysis result being larger than the preset matching degree according to the analysis result and the plurality of conversation recommendation operations;
randomly pushing at least one preferred recommended utterance to the user.
8. A depth model-based apparatus for recommended speech generation, the apparatus comprising:
the system comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring a first recommended dialect of a current problem of a user, and the first recommended dialect is at least one recommended dialect which is configured in advance in the system; and
acquiring a similar problem set similar to the current problem and a recommendation language of any similar problem in the similar problem set, and taking the recommendation language of any similar problem as a second recommendation language;
the processing unit is used for taking the first multidimensional feature vector of the first recommended speech and at least one second multidimensional feature vector of the second recommended speech, which are acquired by the acquiring unit, as the input of a preset depth model, and obtaining the probability of each recommended speech through the output of the preset depth model, wherein the preset depth model is obtained by classifying various types of speech based on a naive Bayes classifier;
and the recommended dialect generating unit is used for generating a plurality of recommended dialects of the current problem according to the probability of each recommended dialect output by the processing unit.
9. A computer device comprising a memory and a processor, the memory having stored therein computer readable instructions which, when executed by the processor, cause the processor to perform the steps of the method of recommended speech generation according to any of claims 1 to 7.
10. A storage medium having stored thereon computer-readable instructions which, when executed by one or more processors, cause the one or more processors to perform the steps of the method of recommended speech generation according to any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011304769.4A CN112395402A (en) | 2020-11-19 | 2020-11-19 | Depth model-based recommended word generation method and device and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011304769.4A CN112395402A (en) | 2020-11-19 | 2020-11-19 | Depth model-based recommended word generation method and device and computer equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112395402A true CN112395402A (en) | 2021-02-23 |
Family
ID=74606557
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011304769.4A Pending CN112395402A (en) | 2020-11-19 | 2020-11-19 | Depth model-based recommended word generation method and device and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112395402A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113688221A (en) * | 2021-09-08 | 2021-11-23 | 中国平安人寿保险股份有限公司 | Model-based dialect recommendation method and device, computer equipment and storage medium |
-
2020
- 2020-11-19 CN CN202011304769.4A patent/CN112395402A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113688221A (en) * | 2021-09-08 | 2021-11-23 | 中国平安人寿保险股份有限公司 | Model-based dialect recommendation method and device, computer equipment and storage medium |
| CN113688221B (en) * | 2021-09-08 | 2023-07-25 | 中国平安人寿保险股份有限公司 | Model-based conversation recommendation method, device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108153876B (en) | Intelligent question and answer method and system | |
| CN107038220B (en) | Method, intelligent robot and system for generating memorandum | |
| KR101909807B1 (en) | Method and apparatus for inputting information | |
| CN110110062B (en) | Machine intelligent question and answer method and device and electronic equipment | |
| US9582757B1 (en) | Scalable curation system | |
| CN106874441B (en) | Intelligent question-answering method and device | |
| US9190055B1 (en) | Named entity recognition with personalized models | |
| CN111428010B (en) | Man-machine intelligent question-answering method and device | |
| US8010343B2 (en) | Disambiguation systems and methods for use in generating grammars | |
| CN112035599B (en) | Query method and device based on vertical search, computer equipment and storage medium | |
| US20210026890A1 (en) | Faq consolidation assistance device, faq consolidation assistance method, and program | |
| CN110415679A (en) | Voice error correction method, device, equipment and storage medium | |
| CN111899140A (en) | Customer service training method and system based on dialect level improvement | |
| CN110890088A (en) | Voice information feedback method and device, computer equipment and storage medium | |
| CN112487810A (en) | Intelligent customer service method, device, equipment and storage medium | |
| CN112667076A (en) | Voice interaction data processing method and device | |
| US11526512B1 (en) | Rewriting queries | |
| CN112632248A (en) | Question answering method, device, computer equipment and storage medium | |
| CN113591463B (en) | Intention recognition method, device, electronic equipment and storage medium | |
| CN111159334A (en) | Method and system for house source follow-up information processing | |
| CN111209367A (en) | Information searching method, information searching device, electronic equipment and storage medium | |
| CN112395402A (en) | Depth model-based recommended word generation method and device and computer equipment | |
| CN109726002B (en) | Processing flow adjusting method and device | |
| CN112100339A (en) | User intention recognition method and device for intelligent voice robot and electronic equipment | |
| CN109684357B (en) | Information processing method and device, storage medium and terminal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |