CN109815489A - Collection information generating method, device, computer equipment and storage medium - Google Patents
Collection information generating method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109815489A CN109815489A CN201910001811.6A CN201910001811A CN109815489A CN 109815489 A CN109815489 A CN 109815489A CN 201910001811 A CN201910001811 A CN 201910001811A CN 109815489 A CN109815489 A CN 109815489A
- Authority
- CN
- China
- Prior art keywords
- user
- identified
- information
- refund
- refund wish
- 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
Abstract
This application involves a kind of collection information generating method, device, computer equipment and storage mediums.The described method includes: obtaining the voice information of user to be identified；Corresponding text information is converted by voice information, target information is extracted from text information；In the hierarchy model that target information input is pre-established, the output of hierarchy model is obtained as a result, output result includes the refund wish label of user to be identified；The refund wish grade that user to be identified is determined according to the refund wish label of user to be identified generates corresponding loan collection information according to the refund wish grade of user to be identified.This method is based on speech recognition technology, it can be according to the voice information of user, determine the refund wish grade of user, and only generate loan collection information corresponding with the refund wish grade of user, it avoids generating excessive information and wasting server resource, to improve the resource utilization of server.
Description
Technical field
This application involves technical field of data processing, more particularly to a kind of collection information generating method, device, computer
Equipment and storage medium.
Background technique
After loan application people provides a loan to financial institution (such as bank) application, financial institution collection personnel are generally required
Some collection means are taken to overdue loan applicant, to guarantee that fund can flow back in time.
However, collection personnel carry out timely collection to overdue loan applicant for convenience, traditional technology generally passes through clothes
Be engaged in device, obtain a variety of collection information (such as voice push, phone collection etc.), and a variety of collection information recommendations that will acquire to pair
The collection personnel answered.But existing server resource is limited, if being directed to each overdue loan applicant, server is required
Repeatedly to collection personnel push it is excessive to overdue loan applicant and not applicable collection information, will lead to the resource of server
Waste, to reduce the resource utilization of server.
Summary of the invention
Based on this, it is necessary in view of the above technical problems, provide a kind of urging for resource utilization that can be improved server
Receive information generating method, device, computer equipment and storage medium.
A kind of collection information generating method, which comprises
Obtain the voice information of user to be identified；
Corresponding text information is converted by the voice information, target letter is extracted from the text information
Breath；
In the hierarchy model that target information input is pre-established, the output of the hierarchy model is obtained as a result, institute
State the refund wish label that output result includes the user to be identified；
The refund wish grade that the user to be identified is determined according to the refund wish label of the user to be identified, according to
The refund wish grade of the user to be identified generates corresponding loan collection information.
It is described in one of the embodiments, to extract target information from the text information, comprising:
The text information is divided into multiple sub-informations；
The information type of each sub-information is obtained respectively；
The information type of each sub-information is matched with presupposed information type respectively, obtains information type matching
Successful sub-information, as target information.
In one of the embodiments, in the hierarchy model that target information input is pre-established, institute is obtained
State the output result of hierarchy model, comprising:
In the hierarchy model that target information input is pre-established, the hierarchy model is used to be believed according to the target
Breath inquiry first database, obtains mood classification corresponding with the target information, as the mood classification of user to be identified, root
According to the second database of mood Query of the user to be identified, obtain corresponding with the mood classification of the user to be identified
Refund wish label, as output result；
Obtain the output result of the hierarchy model.
The refund wish label according to the user to be identified determines described to be identified in one of the embodiments,
The refund wish grade of user generates corresponding loan collection information, packet according to the refund wish grade of the user to be identified
It includes:
According to the refund wish tag queries third database of the user to be identified, the third database includes multiple
Refund wish grade corresponding with refund wish label；
The refund wish label in third database with the refund wish tag match of the user to be identified is determined, by institute
The corresponding refund wish grade of refund wish label is stated, the refund wish grade as the user to be identified；
Generate loan collection information corresponding with the refund wish grade of the user to be identified.
In one of the embodiments, after the output result for obtaining the hierarchy model, according to described to be identified
The refund wish label of user determines before the refund wish grade of the user to be identified, further includes:
Obtain the user to be identified that the voice information carries answers duration and tone；
According to the refund wish label for answering the user to be identified that duration and tone export the hierarchy model into
Row verifying.
In one of the embodiments, it is described answered according to duration and tone to the hierarchy model export wait know
The refund wish label of other user is verified, comprising:
The corresponding weight of duration is answered described in duration determination according to described answer, the tone pair is determined according to the tone
The weight answered；
The sum of the corresponding weight of duration weight corresponding with the tone is answered described in acquisition, as target weight；
Weight corresponding with the refund wish label of the user to be identified is obtained, the weight is calculated and the target is weighed
Error between weight；
If the error is less than preset first threshold value, it is determined that the refund meaning of the user to be identified of the hierarchy model output
It is willing to that label Verification passes through；
If the error is greater than or equal to the preset first threshold value, it is determined that the use to be identified of the hierarchy model output
The refund wish label Verification at family does not pass through, will refund wish label corresponding with the target weight, as described to be identified
The refund wish label of user.
The hierarchy model is obtained by following methods in one of the embodiments:
Obtain multiple sample object information and corresponding refund wish label；
Sample object information is identified by hierarchy model to be trained, obtains the output knot of the hierarchy model
Fruit；
The output result is compared with corresponding practical refund wish label, obtains identification error；
When the identification error is greater than or equal to default second threshold, according to the identification error to the hierarchy model
Repetition training is carried out, until according to the output result for training obtained hierarchy model to obtain and corresponding practical refund wish label
Between identification error be less than the default second threshold.
A kind of collection information generation device, described device include:
Data obtaining module, for obtaining the voice information of user to be identified；
Information conversion module is believed for converting corresponding text information for the voice information from the text
Target information is extracted in breath；
As a result module is obtained, in the hierarchy model for pre-establishing target information input, obtains the classification
The output of model is as a result, the output result includes the refund wish label of the user to be identified；
Information generating module, for determining the user's to be identified according to the refund wish label of the user to be identified
Refund wish grade generates corresponding loan collection information according to the refund wish grade of the user to be identified.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the processing
Device performs the steps of when executing the computer program
Obtain the voice information of user to be identified；
Corresponding text information is converted by the voice information, target letter is extracted from the text information
Breath；
In the hierarchy model that target information input is pre-established, the output of the hierarchy model is obtained as a result, institute
State the refund wish label that output result includes the user to be identified；
The refund wish grade that the user to be identified is determined according to the refund wish label of the user to be identified, according to
The refund wish grade of the user to be identified generates corresponding loan collection information.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
It is performed the steps of when row
Obtain the voice information of user to be identified；
Corresponding text information is converted by the voice information, target letter is extracted from the text information
Breath；
In the hierarchy model that target information input is pre-established, the output of the hierarchy model is obtained as a result, institute
State the refund wish label that output result includes the user to be identified；
The refund wish grade that the user to be identified is determined according to the refund wish label of the user to be identified, according to
The refund wish grade of the user to be identified generates corresponding loan collection information.
Above-mentioned collection information generating method, device, computer equipment and storage medium, the use to be identified that server will acquire
The voice information at family is converted into corresponding text information, and target information is extracted from text information；Target information is defeated
Enter the hierarchy model pre-established, obtains the refund wish label of the user to be identified of hierarchy model output；According to use to be identified
The refund wish label at family determines the refund wish grade of user to be identified, is generated according to the refund wish grade of user to be identified
Corresponding loan collection information, realizes the voice information according to user to be identified, and automatic identification user's to be identified goes back
The purpose of money wish；The loan collection information for being applicable in user to be identified can be only generated according to the refund wish of user to be identified
It is referred to for collection personnel, it is excessive to user to be identified and not applicable loan collection information without generating, reduce to collection
The information that personnel provide avoids server resource waste, to improve the resource utilization of server；Loan is improved simultaneously
Money collection efficiency, reduces human cost.
Detailed description of the invention
Fig. 1 is the application scenario diagram of collection information generating method in one embodiment；
Fig. 2 is the flow diagram of collection information generating method in one embodiment；
Fig. 3 is flow diagram the step of extracting target information from text information in one embodiment；
Fig. 4 is the flow diagram of collection information generating method in another embodiment；
Fig. 5 is the structural block diagram of collection information generation device in one embodiment；
Fig. 6 is the internal structure chart of computer equipment in one embodiment.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
Collection information generating method provided by the present application, can be applied in application environment as shown in Figure 1.Wherein, eventually
End 110 is communicated by network with server 120, and terminal 110 is mounted with that phone system, phone system have recorded use to be identified
Ticket call voice messaging can be sent to server 120 by the voice information at family.Server 120 will acquire logical
Words voice messaging is converted into corresponding text information, and target information is extracted from text information；Target information input is pre-
In the hierarchy model first established, the output of hierarchy model is obtained as a result, output result includes the refund wish mark of user to be identified
Label；The refund wish grade that user to be identified is determined according to the refund wish label of user to be identified, according to user's to be identified
Refund wish grade generates corresponding loan collection information, and the loan collection information of generation is sent in terminal 110.Its
In, terminal 110 can be, but not limited to be various personal computers, laptop, smart phone and tablet computer etc., server
120 can be realized with the server cluster of the either multiple server compositions of independent server.
In one embodiment, as shown in Fig. 2, providing a kind of collection information generating method, it is applied to Fig. 1 in this way
In server for be illustrated, comprising the following steps:
Step S201 obtains the voice information of user to be identified.
User to be identified refers to loan collection object.Terminal is mounted with phone system, such as IVR (Interactive
Voice Response, interactive voice answering), phone system can establish communication connection with corresponding user terminal, and remember
The voice information for employing family, the voice information as user to be identified；Terminal is led to the user's to be identified of record
Words voice messaging is sent to server, and subsequent server is facilitated to convert corresponding text for the voice information of user to be identified
Word information.Voice information includes the answer information of user, for example, " good, I is aware of, I can refund as early as possible ".
Voice information is converted corresponding text information by step S202, and target letter is extracted from text information
Breath.
Target information refers to the information that can show that the refund wish of user to be identified, such as " refunding as early as possible ", " at once also
Money " etc..For server by speech recognition technology, the voice information that will acquire is converted into corresponding text information, and is based on
NLP (Natural Language Processing, natural language processing) technology, segments text information, with from being divided into
Word in extract effective information, as target information, redundant information is avoided to interfere, is conducive to subsequent server to extracting
Target information analyzed, with obtain accurately user to be identified refund wish label, thus be conducive to server according to
The refund wish of user to be identified only generates the loan collection information for being applicable in user to be identified for collection personnel reference, further
Improve the resource utilization of server.NLP technology be one can be realized Chinese Automatic Word Segmentation, part-of-speech tagging, syntactic analysis,
The technology of the multiple functions such as information extraction.
Target information is inputted in the hierarchy model pre-established, obtains the output of hierarchy model as a result, defeated by step S203
Result includes the refund wish label of user to be identified out.
Refund wish label is used for the refund wish of identity user.Hierarchy model is obtained by repeatedly training, can
The target information of input is analyzed, determines the mood classification (such as positive mood) of user based on the analysis results, obtain with
The corresponding refund wish label of mood classification, as user to be identified refund wish label and export.
In the hierarchy model that server pre-establishes the target information input of extraction, to pass through hierarchy model to input
Target information is analyzed, and exports the refund wish label of user to be identified based on the analysis results；Server obtains classification mould
The refund wish label of the user to be identified of type output, facilitates the subsequent refund wish label according to user to be identified, determine to
The refund wish grade of user is identified, so that loan collection information corresponding with refund wish grade is generated, to carry out to user
Accurate collection, advantageously reduces human cost；Avoid simultaneously server generate it is excessive to user to be identified and not applicable
Loan collection information, to improve the resource utilization of server.
Step S204 determines the refund wish grade of user to be identified, root according to the refund wish label of user to be identified
Corresponding loan collection information is generated according to the refund wish grade of user to be identified.
Refund wish grade is used to measure the power of the refund wish of user, is one-to-one with refund wish label；
Loan collection information refer to user carry out loan collection mode, such as short message collection, voice push or outgoing call robot with
Into etc..
Server identifies label information entrained by the refund wish label of user to be identified, determines going back for user to be identified
Money wish degree, the refund wish grade of user to be identified is determined according to the refund wish degree of user to be identified, and is only generated
Loan collection information corresponding with the refund wish grade of user to be identified, realizes the refund wish of automatic identification user and life
It is excessive to user to be identified and not applicable loan collection information without generating at the purpose of corresponding loan collection information,
The information provided to collection personnel is provided, server resource waste is avoided, to improve the resource utilization of server；
The loan collection information generated simultaneously according to server effectively can carry out accurate collection to user, avoid blindness collection and anti-
Again with urging, to improve collection efficiency, human cost is reduced.
In above-mentioned collection information generating method, the voice information for the user to be identified that server will acquire is converted into pair
The text information answered, extracts target information from text information；Target information is inputted to the hierarchy model pre-established, is obtained
The refund wish label of the user to be identified of hierarchy model output；It is determined according to the refund wish label of user to be identified to be identified
The refund wish grade of user generates corresponding loan collection information according to the refund wish grade of user to be identified, realizes
According to the voice information of user to be identified, the purpose of the refund wish of automatic identification user to be identified；It can be according to wait know
The refund wish of other user only generates the loan collection information for being applicable in user to be identified for collection personnel reference, without generating
It is more to user to be identified and not applicable loan collection information, the information provided to collection personnel is provided, service is avoided
The device wasting of resources, to improve the resource utilization of server；Simultaneously improve loan collection efficiency, reduce manpower at
This.
In view of in the text information that the voice information by user to be identified is transformed, some information are that do not have
Reference value, it is analyzed so needing to extract target information from text information, redundant information is avoided to interfere.
In one embodiment, as shown in figure 3, the step of extracting target information from text information specifically includes:
Text information is divided into multiple sub-informations by step S301.
Step S302 obtains the information type of each sub-information respectively.
Step S303 respectively matches the information type of each sub-information with presupposed information type, obtains info class
The sub-information of type successful match, as target information.
Presupposed information type refers to the information type that can show that the refund wish of user to be identified, is based on history text
Message sample, which is constantly analyzed, to be obtained, such as temporal information type, action message type.For example, server is based on NLP technology,
Automatically the text information extracted is segmented, text information is divided into multiple sub-informations；Such as it is text information is " good
, I will refund after fortnight " it is divided into " good ", " I ", " will in ", " after fortnight ", " refund " this five sons
Information；By part-of-speech tagging means, the information type of this five sub-informations is determined；Respectively by the information type of this five sub-informations
It is matched with presupposed information type (such as temporal information type, action message type), obtains information type successful match
Sub-information " after fortnight ", " refund ", as target information.
Text information is divided into multiple sub-informations by server, and effective information is filtered out from multiple sub-informations as target
Information can be interfered to avoid other information, be conducive to it is subsequent by the target information extracted input hierarchy model in, to obtain standard
The refund wish label of true user to be identified only generates to be conducive to server according to the refund wish of user to be identified
The loan collection information for being applicable in user to be identified is referred to for collection personnel, reduces the information provided to collection personnel, to mention
The high resource utilization of server.
Further, it is contemplated that only one not single meaning of some words, in order to avoid the target information filtered out exists
Ambiguity, server it needs to be determined that the target information filtered out physical meaning.In one embodiment, above-mentioned steps S303,
The sub-information for obtaining information type successful match, after target information, further includes: determining from text information to believe with target
Cease associated contextual information；According to target information and with the associated contextual information of target information, determine target information
Physical meaning.
For example, server is based on NLP technology, target information is extracted " no from text information " I will not refund "
Can refund ", and the contextual information of " will not refund " is combined, determine that the physical meaning of " will not refund " is " can refund ".Service
Device is based on NLP technology, can determine the physical meaning of the target information filtered out, is conducive to the subsequent target information that will be extracted
Physical meaning be input in hierarchy model, with obtain accurately user to be identified refund wish label, to be conducive to take
Device be engaged according to the refund wish of user to be identified, only generates the loan collection information for being applicable in user to be identified for collection personnel ginseng
It examines, the information provided to collection personnel is further reduced, to improve the resource utilization of server.
It, can be in order to accurately obtain the refund wish of user to be identified by target information after obtaining target information
It is analyzed by target information of the trained hierarchy model to input, to export the refund wish label of user to be identified.
In one embodiment, target information is inputted in the hierarchy model pre-established, obtains hierarchy model by above-mentioned steps S203
Export result, comprising: in the hierarchy model for pre-establishing target information input, hierarchy model according to target information for inquiring
First database obtains mood classification corresponding with target information, as the mood classification of user to be identified, according to use to be identified
Second database of mood Query at family obtains refund wish label corresponding with the mood classification of user to be identified, as
Export result；Obtain the output result of hierarchy model.Mood classification include positive mood, negative emotions and other.First data
Library includes multiple mood classifications corresponding with target information, and the second database includes multiple refund wishes corresponding with mood classification
Label.
By hierarchy model, if recognizing the mood classification of user to be identified as positive mood, expression refund meaning is obtained
Strong refund wish label is willing to, as user's refund wish label to be identified；If the mood classification for recognizing user to be identified is
Negative emotions then obtain the refund wish label for indicating that refund wish is weak, as user's refund wish label to be identified；If identification
Mood classification to user to be identified is other moods, then by manual analysis target information, with the feelings of determination user to be identified
Thread classification, so that it is determined that the refund wish label of user to be identified.
By trained hierarchy model, the target information of input can be analyzed, determine use based on the analysis results
The mood classification at family, and obtain corresponding with mood classification refund wish label, as the refund wish label of user to be identified,
Facilitate the subsequent refund wish according to user to be identified, only generates the loan collection information for being applicable in user to be identified for collection personnel
With reference to, it is excessive to user to be identified and not applicable loan collection information without generating, to improve the resource of server
Utilization rate；Simultaneously according to the loan collection information of offer, accurate collection can be carried out to user, to improve loan collection effect
Rate reduces human cost.
In order to further increase hierarchy model output user to be identified refund wish label accuracy rate, can to point
Grade model is repeatedly trained.In one embodiment, hierarchy model is obtained by following methods: obtaining multiple sample object letters
Breath and corresponding refund wish label；Sample object information is identified by hierarchy model to be trained, obtains classification mould
The output result of type；Output result is compared with corresponding practical refund wish label, obtains identification error；When identification misses
When difference is greater than or equal to default second threshold, repetition training is carried out to the hierarchy model according to identification error, until according to instruction
Identification error between the output result that the hierarchy model got obtains and corresponding practical refund wish label is less than described
Default second threshold.
For example, server adjusts classification mould according to identification error when identification error is greater than or equal to default second threshold
The parameter of type；Sample object information is again identified that according to hierarchy model adjusted, acquisition is obtained according to hierarchy model
Output result and corresponding practical refund wish label between identification error, according to identification error to the parameter of hierarchy model
It is adjusted again, to be trained again to hierarchy model, until the output result obtained according to the hierarchy model after training
Identification error between corresponding practical refund wish label is less than default second threshold, terminates training.Server is according to knowledge
Other error carries out repetition training to hierarchy model, convenient to export more accurate refund wish label by hierarchy model, thus into
One step improves the accuracy rate of the refund wish label of the user to be identified of hierarchy model output.
In one embodiment, above-mentioned steps S204 determines use to be identified according to the refund wish label of user to be identified
The refund wish grade at family generates corresponding loan collection information according to the refund wish grade of user to be identified, comprising: according to
The refund wish tag queries third database of user to be identified, third database include multiple corresponding with refund wish label
Refund wish grade；Determine the refund wish label in third database with the refund wish tag match of user to be identified, it will
The corresponding refund wish grade of refund wish label, the refund wish grade as user to be identified；It generates and user to be identified
The corresponding loan collection information of refund wish grade.
Server is directed to the refund wish label of user to be identified, and user is divided into different refund wish grades, and
Only generate loan collection information corresponding with refund wish grade.For example, the user high for refund wish grade, server is only
Generate the loan collection information for being applicable in the user of the grade, such as short message collection；For the low user of refund wish grade, service
Device only generates the loan collection information for being applicable in the user of the grade, such as manual telephone system follow-up.It realizes according to user to be identified
Refund wish, only generate the purpose for being applicable in the loan collection information of user to be identified, it is excessive to use to be identified without generating
Family and not applicable loan collection information, reduce the information that server is provided to collection personnel, avoid server resource wave
Take, to improve the resource utilization of server；The loan collection information for being applicable in user to be identified is taken to wait knowing to this simultaneously
Other user carries out collection, can be improved loan collection efficiency, to reduce human cost.
In order to accurately generate the loan collection information for being applicable in user to be identified, server hierarchy model can be exported to
The refund wish label of identification user is verified.In one embodiment, after the output result for obtaining hierarchy model,
Before the refund wish grade for determining user to be identified according to the refund wish label of user to be identified, further includes: obtain call
The user's to be identified that voice messaging carries answers duration and tone；According to answer duration and tone to hierarchy model export to
The refund wish label of identification user is verified.
User to be identified answer duration (including whether hanging up halfway) length harmony be turned up it is low also can reflect it is to be identified
The refund wish degree of strength of user；For example, user to be identified to answer duration longer, midway is not hung up, and tone is more gentle,
Illustrate that the refund wish of user to be identified is stronger.What server comprehensively considered user to be identified answers duration and tone, can be right
The refund wish label of the user to be identified of hierarchy model output is verified, and is avoided the occurrence of mistake, is further improved service
Device generates the accuracy for being applicable in the loan collection information of user to be identified, avoids generating and excessive to user to be identified and is not suitable for
Loan collection information, to improve the resource utilization of server.
In one embodiment, according to the refund wish for answering the user to be identified that duration and tone export hierarchy model
Label is verified, comprising: is answered the corresponding weight of duration according to duration determination is answered, is determined the corresponding power of tone according to tone
Weight；The sum of the corresponding weight of duration weight corresponding with tone is answered in acquisition, as target weight；It obtains with user's to be identified
The corresponding weight of refund wish label calculates the error between weight and target weight；If error is less than preset first threshold value,
Determine that the refund wish label Verification of the user to be identified of hierarchy model output passes through；If the error is greater than or equal to default the
One threshold value, it is determined that the refund wish label Verification of the user to be identified of hierarchy model output does not pass through, will be with target weight pair
The refund wish label answered, the refund wish label as user to be identified.For example, user to be identified to answer duration longer,
Tone is more gentle, illustrates that the refund wish of user to be identified is stronger, then corresponding target weight is also higher.Hierarchy model output
Refund wish tag representation refund wish height, can also be measured with weight.
The error that server recognizes between the corresponding weight of refund wish label of user to be identified and target weight is small
In preset first threshold value, illustrate the refund wish label of the user to be identified exported by hierarchy model, it is corresponding with target weight
Refund wish label it is close, it is determined that the refund wish label Verification of user to be identified of hierarchy model output passes through；If knowing
Error is clipped to more than or equal to preset first threshold value, illustrates the refund wish mark of the user to be identified exported by hierarchy model
Label, refund wish label corresponding with target weight be not close, it is determined that the refund meaning of the user to be identified of hierarchy model output
It is willing to that label Verification does not pass through, the final refund wish mark with the corresponding refund wish label of target weight, as user to be identified
Label.
The user's to be identified carried in server combination voice information answers duration and tone, defeated to hierarchy model
The refund wish label of user to be identified out is verified, and mistake is avoided the occurrence of, and can effectively determine going back for user to be identified
Money wish grade facilitates server to accurately generate the loan collection information for being applicable in user to be identified, avoids generating excessive pair
User to be identified and not applicable loan collection information and waste server resource, to improve the utilization of resources of server
Rate.
In one embodiment, as shown in figure 4, providing another collection information generating method, comprising the following steps:
Step S401 obtains the voice information of user to be identified.
Voice information is converted corresponding text information by step S402, and target letter is extracted from text information
Breath.
Target information is inputted in the hierarchy model pre-established, obtains the output of hierarchy model as a result, defeated by step S403
Result includes the refund wish label of user to be identified out.
Step S404, obtain the user to be identified that voice information carries answers duration and tone；When according to answering
Long and tone verifies the refund wish label for the user to be identified that hierarchy model exports.
Step S405, after the refund wish label Verification for determining user to be identified passes through, according to going back for user to be identified
Money wish label determines the refund wish grade of user to be identified, is generated according to the refund wish grade of user to be identified corresponding
Loan collection information.
Above-mentioned collection information generating method, realizes the voice information according to user to be identified, and automatic identification waits knowing
The purpose of the refund wish of other user；The loan for being applicable in user to be identified can be only generated according to the refund wish of user to be identified
Money collection information is referred to for collection personnel, excessive to user to be identified and not applicable loan collection information without generating, and is subtracted
The information provided to collection personnel has been provided, server resource waste has been avoided, to improve the resource utilization of server.Together
When in conjunction with the user to be identified carried in voice information answer duration and tone, to the use to be identified of hierarchy model output
The refund wish label at family is verified, and mistake is avoided the occurrence of, and can effectively determine the refund wish grade of user to be identified, into
And server is facilitated to accurately generate the loan collection information for being applicable in user to be identified, it avoids generating excessive to user to be identified
And not applicable loan collection information and waste server resource, to improve the resource utilization of server.
It should be understood that although each step in the flow chart of Fig. 2-4 is successively shown according to the instruction of arrow,
These steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly stating otherwise herein, these steps
Execution there is no stringent sequences to limit, these steps can execute in other order.Moreover, at least one in Fig. 2-4
Part steps may include that perhaps these sub-steps of multiple stages or stage are not necessarily in synchronization to multiple sub-steps
Completion is executed, but can be executed at different times, the execution sequence in these sub-steps or stage is also not necessarily successively
It carries out, but can be at least part of the sub-step or stage of other steps or other steps in turn or alternately
It executes.
In one embodiment, as shown in figure 5, providing a kind of collection information generation device, comprising: data obtaining module
510, information conversion module 520, result obtain module 530 and information generating module 540, in which:
Data obtaining module 510, for obtaining the voice information of user to be identified.
Information conversion module 520 is mentioned from text information for converting corresponding text information for voice information
Take out target information.
As a result module 530 is obtained, in the hierarchy model for pre-establishing target information input, obtains hierarchy model
Output is as a result, output result includes the refund wish label of user to be identified.
Information generating module 540 determines the refund of user to be identified for the refund wish label according to user to be identified
Wish grade generates corresponding loan collection information according to the refund wish grade of user to be identified.
In one embodiment, information conversion module is also used to for text information to be divided into multiple sub-informations；It obtains respectively
The information type of each sub-information；The information type of each sub-information is matched with presupposed information type respectively, obtains letter
The successful sub-information of type matching is ceased, as target information.
In one embodiment, module is as a result obtained to be also used to input target information in the hierarchy model pre-established,
Hierarchy model is used to inquire first database according to target information, mood classification corresponding with target information is obtained, as wait know
The mood classification of other user obtains the feelings with user to be identified according to the second database of mood Query of user to be identified
The corresponding refund wish label of thread classification, as output result；Obtain the output result of hierarchy model.
In one embodiment, information generating module is also used to the refund wish tag queries third according to user to be identified
Database, third database include multiple refund wish grades corresponding with refund wish label；Determine in third database with
The refund wish label of the refund wish tag match of user to be identified, by the corresponding refund wish grade of refund wish label,
Refund wish grade as user to be identified；Generate loan collection letter corresponding with the refund wish grade of user to be identified
Breath.
In one embodiment, collection information generation device further includes label Verification module, is used for information generating module root
Before the refund wish grade for determining user to be identified according to the refund wish label of user to be identified, obtains voice information and take
The user's to be identified of band answers duration and tone；The user's to be identified that hierarchy model is exported according to duration and tone is answered
Refund wish label is verified.
In one embodiment, label Verification module be also used to according to answer duration determination answer the corresponding weight of duration,
The corresponding weight of tone is determined according to tone；The sum of the corresponding weight of duration weight corresponding with tone is answered in acquisition, as mesh
Mark weight；Weight corresponding with the refund wish label of user to be identified is obtained, the error between weight and target weight is calculated；
If error is less than preset first threshold value, it is determined that the refund wish label Verification of the user to be identified of hierarchy model output passes through；
If error is greater than or equal to preset first threshold value, it is determined that the refund wish label Verification of the user to be identified of hierarchy model output
Do not pass through, will refund wish label corresponding with target weight, the refund wish label as user to be identified.
In one embodiment, collection information generation device further includes model training module, for obtaining multiple sample mesh
Mark information and corresponding refund wish label；Sample object information is identified by hierarchy model to be trained, is divided
The output result of grade model；Output result is compared with corresponding practical refund wish label, obtains identification error；Work as knowledge
When other error is greater than or equal to default second threshold, repetition training is carried out to hierarchy model according to identification error, until according to instruction
Identification error between the output result that the hierarchy model got obtains and corresponding practical refund wish label is less than default
Second threshold.
The voice information of above-mentioned each embodiment, the user to be identified that collection information generation device will acquire is converted into
Corresponding text information, extracts target information from text information；Target information is inputted to the hierarchy model pre-established, is obtained
The refund wish label for the user to be identified for taking hierarchy model to export；It is determined according to the refund wish label of user to be identified wait know
The refund wish grade of other user generates corresponding loan collection information according to the refund wish grade of user to be identified, realizes
According to the voice information of user to be identified, the purpose of the refund wish of automatic identification user to be identified；Can according to
The refund wish for identifying user only generates the loan collection information for being applicable in user to be identified for collection personnel reference, without generating
It is excessive to user to be identified and not applicable loan collection information, the information provided to collection personnel is provided, clothes are avoided
The business device wasting of resources, to improve the resource utilization of server；Simultaneously improve loan collection efficiency, reduce manpower at
This.
Specific about collection information generation device limits the limit that may refer to above for collection information generating method
Fixed, details are not described herein.Modules in above-mentioned collection information generation device can fully or partially through software, hardware and its
Combination is to realize.Above-mentioned each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also be with
It is stored in the memory in computer equipment in a software form, in order to which processor calls the above modules of execution corresponding
Operation.
In one embodiment, a kind of computer equipment is provided, which can be server, internal junction
Composition can be as shown in Figure 6.The computer equipment include by system bus connect processor, memory, network interface and
Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment
Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data
Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
The database of machine equipment is for storing collection information.The network interface of the computer equipment is used to pass through network with external terminal
Connection communication.To realize a kind of collection information generating method when the computer program is executed by processor.
It will be understood by those skilled in the art that structure shown in Fig. 6, only part relevant to application scheme is tied
The block diagram of structure does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific computer equipment
It may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, a kind of computer equipment, including memory and processor are provided, which is stored with
Computer program, the processor perform the steps of when executing computer program
Obtain the voice information of user to be identified；
Corresponding text information is converted by voice information, target information is extracted from text information；
In the hierarchy model that target information input is pre-established, the output of hierarchy model is obtained as a result, output result packet
Include the refund wish label of user to be identified；
The refund wish grade that user to be identified is determined according to the refund wish label of user to be identified, according to use to be identified
The refund wish grade at family generates corresponding loan collection information.
In one embodiment, it is also performed the steps of when processor executes computer program and is divided into text information
Multiple sub-informations；The information type of each sub-information is obtained respectively；Respectively by the information type and presupposed information of each sub-information
Type is matched, and the sub-information of information type successful match is obtained, as target information.
In one embodiment, it is also performed the steps of when processor executes computer program target information input is pre-
In the hierarchy model first established, hierarchy model is used to inquire first database according to target information, obtains corresponding with target information
Mood classification, obtained as the mood classification of user to be identified according to the second database of mood Query of user to be identified
Refund wish label corresponding with the mood classification of user to be identified is taken, as output result；Obtain the output knot of hierarchy model
Fruit.
In one embodiment, it also performs the steps of when processor executes computer program according to user's to be identified
Refund wish tag queries third database, third database include multiple refund wishes corresponding with refund wish label etc.
Grade；The refund wish label in third database with the refund wish tag match of user to be identified is determined, by refund wish mark
Corresponding refund wish grade is signed, the refund wish grade as user to be identified；Generate the refund wish with user to be identified
The corresponding loan collection information of grade.
In one embodiment, it is also performed the steps of when processor executes computer program and is obtaining hierarchy model
After exporting result, before the refund wish label according to user to be identified determines the refund wish grade of user to be identified,
Obtain the user to be identified that voice information carries answers duration and tone；According to answering duration and tone to hierarchy model
The refund wish label of the user to be identified of output is verified.
In one embodiment, it is also performed the steps of when processor executes computer program determining according to duration is answered
The corresponding weight of duration is answered, the corresponding weight of tone is determined according to tone；The corresponding weight of duration and tone pair are answered in acquisition
The sum of weight answered, as target weight；Obtain corresponding with the refund wish label of user to be identified weight, calculate weight and
Error between target weight；If error is less than preset first threshold value, it is determined that the user's to be identified of hierarchy model output goes back
Money wish label Verification passes through；If error is greater than or equal to preset first threshold value, it is determined that the use to be identified of hierarchy model output
The refund wish label Verification at family does not pass through, will refund wish label corresponding with target weight, as user to be identified also
Money wish label.
In one embodiment, it is also performed the steps of when processor executes computer program and obtains multiple sample objects
Information and corresponding refund wish label；Sample object information is identified by hierarchy model to be trained, is classified
The output result of model；Output result is compared with corresponding practical refund wish label, obtains identification error；Work as identification
When error is greater than or equal to default second threshold, repetition training is carried out to hierarchy model according to identification error, until according to training
Identification error between the output result that obtained hierarchy model obtains and corresponding practical refund wish label is less than default the
Two threshold values.
Above-mentioned each embodiment, computer equipment are realized by the computer program run on processor according to wait know
The voice information of other user, the purpose of the refund wish of automatic identification user to be identified；It can be according to user's to be identified
Refund wish only generates the loan collection information for being applicable in user to be identified for collection personnel reference, without generating excessive treat
It identifies user and not applicable loan collection information, the information provided to collection personnel is provided, avoid server resource wave
Take, to improve the resource utilization of server.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program performs the steps of when being executed by processor
Obtain the voice information of user to be identified；
Corresponding text information is converted by voice information, target information is extracted from text information；
In the hierarchy model that target information input is pre-established, the output of hierarchy model is obtained as a result, output result packet
Include the refund wish label of user to be identified；
The refund wish grade that user to be identified is determined according to the refund wish label of user to be identified, according to use to be identified
The refund wish grade at family generates corresponding loan collection information.
In one embodiment, it is also performed the steps of when computer program is executed by processor and divides text information
At multiple sub-informations；The information type of each sub-information is obtained respectively；Respectively by the information type of each sub-information and default letter
Breath type is matched, and the sub-information of information type successful match is obtained, as target information.
In one embodiment, it is also performed the steps of when computer program is executed by processor and inputs target information
In the hierarchy model pre-established, hierarchy model is used to inquire first database according to target information, obtains and target information pair
The mood classification answered, as the mood classification of user to be identified, according to the second database of mood Query of user to be identified,
Refund wish label corresponding with the mood classification of user to be identified is obtained, as output result；Obtain the output of hierarchy model
As a result.
In one embodiment, it also performs the steps of when computer program is executed by processor according to user to be identified
Refund wish tag queries third database, third database includes multiple refund wishes corresponding with refund wish label etc.
Grade；The refund wish label in third database with the refund wish tag match of user to be identified is determined, by refund wish mark
Corresponding refund wish grade is signed, the refund wish grade as user to be identified；Generate the refund wish with user to be identified
The corresponding loan collection information of grade.
In one embodiment, it is also performed the steps of when computer program is executed by processor and is obtaining hierarchy model
Output result after, the refund wish label according to user to be identified determine user to be identified refund wish grade it
Before, obtain the user to be identified that voice information carries answers duration and tone；According to answering duration and tone to classification
The refund wish label of the user to be identified of model output is verified.
In one embodiment, it is also performed the steps of when computer program is executed by processor true according to duration is answered
Surely the corresponding weight of duration is answered, the corresponding weight of tone is determined according to tone；The corresponding weight of duration and tone are answered in acquisition
The sum of corresponding weight, as target weight；Weight corresponding with the refund wish label of user to be identified is obtained, weight is calculated
Error between target weight；If error is less than preset first threshold value, it is determined that the user's to be identified of hierarchy model output
Refund wish label Verification passes through；If error is greater than or equal to preset first threshold value, it is determined that hierarchy model exports to be identified
The refund wish label Verification of user does not pass through, will refund wish label corresponding with target weight, as user's to be identified
Refund wish label.
In one embodiment, it is also performed the steps of when computer program is executed by processor and obtains multiple sample mesh
Mark information and corresponding refund wish label；Sample object information is identified by hierarchy model to be trained, is divided
The output result of grade model；Output result is compared with corresponding practical refund wish label, obtains identification error；Work as knowledge
When other error is greater than or equal to default second threshold, repetition training is carried out to hierarchy model according to identification error, until according to instruction
Identification error between the output result that the hierarchy model got obtains and corresponding practical refund wish label is less than default
Second threshold.
Above-mentioned each embodiment, computer readable storage medium by its store computer program, realize according to
Identify the voice information of user, the purpose of the refund wish of automatic identification user to be identified；It can be according to user to be identified
Refund wish, only generate and be applicable in the loan collection information of user to be identified and referred to for collection personnel, without generating excessive pair
User to be identified and not applicable loan collection information reduce the information provided to collection personnel, avoid server resource
Waste, to improve the resource utilization of server；Loan collection efficiency is improved simultaneously, reduces human cost.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
The several embodiments of the application above described embodiment only expresses, the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the concept of this application, various modifications and improvements can be made, these belong to the protection of the application
Range.Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of collection information generating method, which comprises
Obtain the voice information of user to be identified；
Corresponding text information is converted by the voice information, extracts target information from the text information；
In the hierarchy model that target information input is pre-established, the output of the hierarchy model is obtained as a result, described defeated
Result includes the refund wish label of the user to be identified out；
The refund wish grade that the user to be identified is determined according to the refund wish label of the user to be identified, according to described
The refund wish grade of user to be identified generates corresponding loan collection information.
2. the method according to claim 1, wherein described extract target information from the text information,
Include:
The text information is divided into multiple sub-informations；
The information type of each sub-information is obtained respectively；
The information type of each sub-information is matched with presupposed information type respectively, obtains information type successful match
Sub-information, as target information.
3. the method according to claim 1, wherein described input the target information classification pre-established
In model, the output result of the hierarchy model is obtained, comprising:
In the hierarchy model that target information input is pre-established, the hierarchy model according to the target information for looking into
First database is ask, mood classification corresponding with the target information is obtained, as the mood classification of user to be identified, according to institute
The second database of mood Query of user to be identified is stated, refund corresponding with the mood classification of the user to be identified is obtained
Wish label, as output result；
Obtain the output result of the hierarchy model.
4. the method according to claim 1, wherein the refund wish label according to the user to be identified
The refund wish grade for determining the user to be identified generates corresponding loan according to the refund wish grade of the user to be identified
Money collection information, comprising:
According to the refund wish tag queries third database of the user to be identified, the third database includes multiple and goes back
The corresponding refund wish grade of money wish label；
It determines the refund wish label in third database with the refund wish tag match of the user to be identified, described will go back
The corresponding refund wish grade of money wish label, the refund wish grade as the user to be identified；
Generate loan collection information corresponding with the refund wish grade of the user to be identified.
5. the method according to claim 1, which is characterized in that in the output for obtaining the hierarchy model
As a result after, the refund wish label according to the user to be identified determine the user to be identified refund wish grade it
Before, further includes:
Obtain the user to be identified that the voice information carries answers duration and tone；
It answers duration and tone according to described the refund wish label for the user to be identified that the hierarchy model exports is tested
Card.
6. according to the method described in claim 5, it is characterized in that, duration and the tone of answering according to is to the classification
The refund wish label of the user to be identified of model output is verified, comprising:
The corresponding weight of duration is answered described in duration determination according to described answer, determines that the tone is corresponding according to the tone
Weight；
The sum of the corresponding weight of duration weight corresponding with the tone is answered described in acquisition, as target weight；
Obtain corresponding with the refund wish label of the user to be identified weight, calculate the weight and the target weight it
Between error；
If the error is less than preset first threshold value, it is determined that the refund wish mark of the user to be identified of the hierarchy model output
Label are verified；
If the error is greater than or equal to the preset first threshold value, it is determined that the user's to be identified of the hierarchy model output
Refund wish label Verification does not pass through, will refund wish label corresponding with the target weight, as the user to be identified
Refund wish label.
7. according to the method described in claim 5, it is characterized in that, the hierarchy model is obtained by following methods:
Obtain multiple sample object information and corresponding refund wish label；
Sample object information is identified by hierarchy model to be trained, obtains the output result of the hierarchy model；
The output result is compared with corresponding practical refund wish label, obtains identification error；
When the identification error is greater than or equal to default second threshold, the hierarchy model is carried out according to the identification error
Repetition training, until according between the obtained obtained output result of hierarchy model of training and corresponding practical refund wish label
Identification error be less than the default second threshold.
8. a kind of collection information generation device, which is characterized in that described device includes:
Data obtaining module, for obtaining the voice information of user to be identified；
Information conversion module, for converting corresponding text information for the voice information, from the text information
Extract target information；
As a result module is obtained, in the hierarchy model for pre-establishing target information input, obtains the hierarchy model
Output as a result, it is described output result include the user to be identified refund wish label；
Information generating module determines the refund of the user to be identified for the refund wish label according to the user to be identified
Wish grade generates corresponding loan collection information according to the refund wish grade of the user to be identified.
9. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In the step of processor realizes any one of claims 1 to 7 the method when executing the computer program.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method described in any one of claims 1 to 7 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910001811.6A CN109815489A (en) | 2019-01-02 | 2019-01-02 | Collection information generating method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910001811.6A CN109815489A (en) | 2019-01-02 | 2019-01-02 | Collection information generating method, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109815489A true CN109815489A (en) | 2019-05-28 |
Family
ID=66603730
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910001811.6A Pending CN109815489A (en) | 2019-01-02 | 2019-01-02 | Collection information generating method, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109815489A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110570295A (en) * | 2019-07-25 | 2019-12-13 | 深圳壹账通智能科技有限公司 | Resource collection method and device, computer equipment and storage medium |
| CN110599324A (en) * | 2019-07-25 | 2019-12-20 | 阿里巴巴集团控股有限公司 | Method and device for predicting refund rate |
| CN111061948A (en) * | 2019-11-25 | 2020-04-24 | 成都铂锡金融信息技术有限公司 | User label recommendation method and device, computer equipment and storage medium |
| CN111552827A (en) * | 2020-04-28 | 2020-08-18 | 腾讯科技（深圳）有限公司 | Labeling method and device, and behavior willingness prediction model training method and device |
| CN111739537A (en) * | 2020-06-08 | 2020-10-02 | 北京灵蚌科技有限公司 | Semantic recognition method and device, storage medium and processor |
| CN113065946A (en) * | 2021-03-17 | 2021-07-02 | 上海浦东发展银行股份有限公司 | Classification updating promoting method and device for overdue credit card certificate clients and storage medium |
| CN114398903A (en) * | 2022-01-21 | 2022-04-26 | 平安科技（深圳）有限公司 | Intention recognition method and device, electronic equipment and storage medium |
| CN115393077A (en) * | 2022-10-28 | 2022-11-25 | 深圳市人马互动科技有限公司 | Data processing method based on loan transaction man-machine conversation system and related device |
| CN116153330A (en) * | 2023-04-04 | 2023-05-23 | 杭州度言软件有限公司 | Intelligent telephone voice robot control method |
-
2019
- 2019-01-02 CN CN201910001811.6A patent/CN109815489A/en active Pending
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110599324A (en) * | 2019-07-25 | 2019-12-20 | 阿里巴巴集团控股有限公司 | Method and device for predicting refund rate |
| CN110570295A (en) * | 2019-07-25 | 2019-12-13 | 深圳壹账通智能科技有限公司 | Resource collection method and device, computer equipment and storage medium |
| CN111061948B (en) * | 2019-11-25 | 2023-05-23 | 成都铂锡金融信息技术有限公司 | User tag recommendation method and device, computer equipment and storage medium |
| CN111061948A (en) * | 2019-11-25 | 2020-04-24 | 成都铂锡金融信息技术有限公司 | User label recommendation method and device, computer equipment and storage medium |
| CN111552827A (en) * | 2020-04-28 | 2020-08-18 | 腾讯科技（深圳）有限公司 | Labeling method and device, and behavior willingness prediction model training method and device |
| CN111552827B (en) * | 2020-04-28 | 2023-06-16 | 腾讯科技（深圳）有限公司 | Labeling method and device, behavior willingness prediction model training method and device |
| CN111739537B (en) * | 2020-06-08 | 2023-01-24 | 北京灵蚌科技有限公司 | Semantic recognition method and device, storage medium and processor |
| CN111739537A (en) * | 2020-06-08 | 2020-10-02 | 北京灵蚌科技有限公司 | Semantic recognition method and device, storage medium and processor |
| CN113065946A (en) * | 2021-03-17 | 2021-07-02 | 上海浦东发展银行股份有限公司 | Classification updating promoting method and device for overdue credit card certificate clients and storage medium |
| CN114398903A (en) * | 2022-01-21 | 2022-04-26 | 平安科技（深圳）有限公司 | Intention recognition method and device, electronic equipment and storage medium |
| CN114398903B (en) * | 2022-01-21 | 2023-06-20 | 平安科技（深圳）有限公司 | Intention recognition method, device, electronic equipment and storage medium |
| CN115393077A (en) * | 2022-10-28 | 2022-11-25 | 深圳市人马互动科技有限公司 | Data processing method based on loan transaction man-machine conversation system and related device |
| CN116153330A (en) * | 2023-04-04 | 2023-05-23 | 杭州度言软件有限公司 | Intelligent telephone voice robot control method |
| CN116153330B (en) * | 2023-04-04 | 2023-06-23 | 杭州度言软件有限公司 | Intelligent telephone voice robot control method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109815489A (en) | Collection information generating method, device, computer equipment and storage medium | |
| CN111104495B (en) | Information interaction method, device, equipment and storage medium based on intention recognition | |
| CN111028827B (en) | Interaction processing method, device, equipment and storage medium based on emotion recognition | |
| WO2020140665A1 (en) | Method and apparatus for quality detection of double-recorded video, and computer device and storage medium | |
| WO2020024389A1 (en) | Method for collecting overdue payment, device, computer apparatus, and storage medium | |
| CN110162633A (en) | Voice data is intended to determine method, apparatus, computer equipment and storage medium | |
| CN110472224B (en) | Quality of service detection method, apparatus, computer device and storage medium | |
| CN109766766A (en) | Employee work condition monitoring method, device, computer equipment and storage medium | |
| CN109783338A (en) | Recording method, device and computer equipment based on business information | |
| CN109800335A (en) | Generation method, device, computer equipment and the storage medium of enterprise's map | |
| CN108446341A (en) | Service condition querying method, device, computer equipment and storage medium | |
| CN110032732A (en) | A kind of text punctuate prediction technique, device, computer equipment and storage medium | |
| CN109766474A (en) | Inquest signal auditing method, device, computer equipment and storage medium | |
| CN111597818B (en) | Call quality inspection method, device, computer equipment and computer readable storage medium | |
| CN110265032A (en) | Conferencing data analysis and processing method, device, computer equipment and storage medium | |
| CN109886110A (en) | Micro- expression methods of marking, device, computer equipment and storage medium | |
| CN108920640A (en) | Context acquisition methods and equipment based on interactive voice | |
| CN109325118A (en) | Uneven sample data preprocess method, device and computer equipment | |
| CN109857373A (en) | Business data processing method, device, computer equipment and storage medium | |
| CN109829640A (en) | Recognition methods, device, computer equipment and the storage medium of enterprise's default risk | |
| CN110598008A (en) | Data quality inspection method and device for recorded data and storage medium | |
| TWI752486B (en) | Training method, feature extraction method, device and electronic device | |
| US11699437B2 (en) | System and method for quantifying meeting effectiveness using natural language processing | |
| CN109711301A (en) | Driver's checking method, device, computer equipment and storage medium | |
| CN113571096B (en) | Speech emotion classification model training method and device, computer equipment and medium |
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 |