CN110930244A - Method and device for calculating user credit investigation evaluation value - Google Patents

Method and device for calculating user credit investigation evaluation value Download PDF

Info

Publication number
CN110930244A
CN110930244A CN201911185326.5A CN201911185326A CN110930244A CN 110930244 A CN110930244 A CN 110930244A CN 201911185326 A CN201911185326 A CN 201911185326A CN 110930244 A CN110930244 A CN 110930244A
Authority
CN
China
Prior art keywords
target collection
collection dimension
data
feedback data
dimension
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.)
Granted
Application number
CN201911185326.5A
Other languages
Chinese (zh)
Other versions
CN110930244B (en
Inventor
乔燕峰
郭啸
于雪龙
何海清
史岩
刘斌
冯帅
武绩峰
王海生
周炎
申红彬
王荣
闫磊
姚崇明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Guotenglianxin Technology Co Ltd
Original Assignee
Beijing Guotenglianxin Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Guotenglianxin Technology Co Ltd filed Critical Beijing Guotenglianxin Technology Co Ltd
Priority to CN201911185326.5A priority Critical patent/CN110930244B/en
Publication of CN110930244A publication Critical patent/CN110930244A/en
Application granted granted Critical
Publication of CN110930244B publication Critical patent/CN110930244B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/03—Credit; Loans; Processing thereof

Landscapes

  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Engineering & Computer Science (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • Technology Law (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The application discloses a method and a device for calculating a credit investigation evaluation value of a user, comprising the following steps of: determining a problem set corresponding to a user to be evaluated, and displaying the problem set to the user to be evaluated through a man-machine interaction module to obtain feedback data of the problem set; performing characteristic extraction on the feedback data of the problem set to obtain a characteristic data set; inputting the feature data set into a pre-constructed confidence coefficient calculation model to obtain the overall confidence coefficient of each target collection dimension; displaying the verification problem corresponding to each target collection dimension; verifying the confidence degree of the feedback data of the problem corresponding to the target collection dimension by using the feedback data of the verification problem corresponding to each target collection dimension to obtain a verification result of the feedback data of the problem corresponding to each target collection dimension; correcting the overall confidence corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension; and obtaining the credit investigation evaluation value of the user to be evaluated according to the corrected overall confidence coefficient.

Description

Method and device for calculating user credit investigation evaluation value
Technical Field
The present application relates to the field of computers, and in particular, to a method and an apparatus for calculating a user credit investigation evaluation value.
Background
At present, an important link of a risk assessment method for small and micro enterprises is a mode of on-site due diligence. Taking the current credit business as an example, the on-site due diligence refers to the way of on-site interviewing with customers by a skilled customer manager, and the loan purpose, repayment capability and repayment willingness of the loan enterprise are deeply known to be evaluated. The key risk points are identified by knowing the specific conditions of the client, the authenticity of the information provided by the client is judged, and finally, credit scores are formed through multi-aspect information to provide basis for credit decision making.
However, the on-site due diligence depends on the qualities of the customer managers themselves, and the cost for cultivating the customer managers is high, so that the standardization and replication difficulty is high, and the efficiency of risk review on the customers is seriously influenced.
Therefore, there is a need for an intelligent method for acquiring user credit data and calculating credit score, which is used to efficiently investigate customers and calculate credit score and ensure the accuracy of risk review.
Disclosure of Invention
In view of this, the embodiments of the present application provide a method and an apparatus for calculating a user credit investigation evaluation value, so as to achieve the purposes of efficiently investigating and calculating a client and ensuring accuracy of risk review.
The first aspect of the application discloses a method for calculating a user credit investigation evaluation value, which comprises the following steps:
determining a question set corresponding to a user to be evaluated; wherein the problem set comprises problems of at least one target collection dimension; each target collection dimension corresponds to at least one question;
displaying the question set to the user to be evaluated through a man-machine interaction module, and acquiring feedback data of the question set; wherein the feedback data of the problem set comprises: feedback data of all problems of each target collection dimension input by a user to be evaluated; feedback data of each problem of each target collection dimension are obtained through different source channels respectively;
performing characteristic extraction on the feedback data of the problem set to obtain a characteristic data set; wherein the feature data set comprises: feature data corresponding to the feedback data of each question of each target collection dimension;
inputting the feature data set into a pre-constructed confidence coefficient calculation model to obtain the overall confidence coefficient of each target collection dimension;
according to the overall confidence of each target collection dimension, displaying a verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module;
obtaining feedback data of the verification problem corresponding to each target collection dimension input by a user to be evaluated, verifying the confidence of the feedback data of the problem corresponding to the target collection dimension by using the feedback data of the verification problem corresponding to each target collection dimension, and obtaining a verification result of the feedback data of the problem corresponding to each target collection dimension;
correcting the overall confidence corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension;
and calculating the credit investigation evaluation value of the user to be evaluated according to the feature data set and the corrected overall confidence of each target collection dimension in the feature data set.
Optionally, the method for constructing the confidence coefficient calculation model includes:
acquiring a training sample set for training and testing a confidence level calculation model, wherein the training sample set comprises feedback data of a plurality of problems belonging to the same dimensionality, and the feedback data of the problems are acquired from different data sources respectively;
performing cross validation on the feedback data of the plurality of problems to divide the feedback data of the plurality of problems into a feedback data group without data consistency and a feedback data group with data consistency;
obtaining target data corresponding to each data in the feedback data group with data consistency based on each data in the feedback data group with data consistency, and replacing each data in the feedback data group corresponding to the target data in the training sample set with the target data;
calculating the probability of each data in the feedback data group without data consistency, the probability of each target data and the overall confidence of the training sample set after data replacement;
and taking the probability of each data, the probability of each target data and the overall confidence coefficient of the training sample set after data replacement as the input of the confidence coefficient calculation model, and carrying out training test on the confidence coefficient calculation model.
Optionally, the displaying, by the human-computer interaction module, the verification problem corresponding to each target collection dimension to the user to be evaluated according to the overall confidence of each target collection dimension includes:
for each target collection dimension, screening a plurality of verification problems in a local question bank to obtain a verification problem corresponding to the target collection dimension, or determining a verification problem corresponding to the target collection dimension from a cloud database to output; the verification problem corresponding to each target collection dimension is obtained by screening based on the feedback data of the collection problem corresponding to each target collection dimension;
and displaying the verification problem corresponding to each target collection dimension to the user to be evaluated through the man-machine interaction module.
Optionally, the verifying the confidence of the feedback data of the problem corresponding to the target collection dimension by using the feedback data of the verification problem corresponding to each target collection dimension to obtain the verification result of the feedback data of the problem corresponding to each target collection dimension includes:
verifying the correctness of the feedback data of the verification problem corresponding to each target collection dimension respectively;
if the correctness of the feedback data of the verification problem corresponding to the target collection dimension cannot be verified, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is unverifiable;
if the feedback data of the verification problem corresponding to the target collection dimension is verified to be wrong, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is low in confidence;
and if the feedback data of the verification problem corresponding to the target collection dimension is verified to be correct, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is high in confidence.
Optionally, the modifying the overall confidence corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension includes:
if the verification result of the feedback data of the problem corresponding to the target collection dimension is that the feedback data cannot be verified, calculating the product of a preset first correction coefficient corresponding to the verification problem and the overall confidence corresponding to the target collection dimension to obtain the overall confidence corresponding to the corrected target collection dimension; wherein the first correction coefficient is less than 1;
if the verification result of the feedback data of the problem corresponding to the target collection dimension is low in confidence, calculating the product of a preset second correction coefficient corresponding to the verification problem and the overall confidence corresponding to the target collection dimension to obtain the overall confidence corresponding to the corrected target collection dimension; wherein the second correction coefficient is less than 1;
if the verification result of the feedback data of the problem corresponding to the target collection dimension is higher in confidence, calculating the product of a preset third correction coefficient corresponding to the verification problem and the overall confidence corresponding to the target collection dimension to obtain the overall confidence corresponding to the corrected target collection dimension; wherein the third correction factor is equal to 1.
Optionally, the calculating, according to the feature data set and the corrected overall confidence of each target collection dimension in the feature data set, a credit investigation evaluation value of the user to be evaluated includes:
calculating to obtain a credit calculation value corresponding to each target collection dimension according to a preset scoring table of each target collection dimension, the question set and a probability distribution function corresponding to each target collection dimension in the question set; the scoring table of each target collection dimension is used for explaining credit calculation values corresponding to each question under each target collection dimension; the probability distribution function corresponding to each target collection dimension in the problem set is used for calculating the weight probability corresponding to each problem in the problem set;
adjusting the credit calculation value corresponding to each target collection dimension according to the corrected overall confidence of each target collection dimension in the feature data set to obtain the adjusted credit calculation value corresponding to each target collection dimension;
and summing the credit calculation values corresponding to each adjusted target collection dimension to obtain the credit investigation evaluation value of the user to be evaluated.
Optionally, a probability distribution function corresponding to each target collection dimension in the problem set is constructed according to each problem in each target collection dimension and a preset confidence level of a source channel corresponding to each problem in each target collection dimension; the confidence of the source channel corresponding to each question is used for explaining the reliability of the credit data acquired by the source channel.
The second aspect of the present application discloses a device for calculating a user credit investigation evaluation value, comprising:
the determining unit is used for determining a problem set corresponding to a user to be evaluated; wherein the problem set comprises problems of at least one target collection dimension; each target collection dimension corresponds to at least one question;
the execution unit is used for displaying the question set to the user to be evaluated through a human-computer interaction module and acquiring feedback data of the question set; wherein the feedback data of the problem set comprises: feedback data of all problems of each target collection dimension input by a user to be evaluated; feedback data of each problem of each target collection dimension are obtained through different source channels respectively;
the extraction unit is used for carrying out feature extraction on the feedback data of the problem set to obtain a feature data set; wherein the feature data set comprises: feature data corresponding to the feedback data of each question of each target collection dimension;
the first calculation unit is used for inputting the feature data set into a pre-constructed confidence coefficient calculation model to obtain the overall confidence coefficient of each target collection dimension;
the display unit is used for displaying the verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module according to the overall confidence of each target collection dimension;
the first obtaining unit is used for obtaining feedback data of the verification problem corresponding to each target collection dimension input by a user to be evaluated, verifying the confidence coefficient of the feedback data of the problem corresponding to the target collection dimension by using the feedback data of the verification problem corresponding to each target collection dimension, and obtaining the verification result of the feedback data of the problem corresponding to each target collection dimension;
the correction unit is used for correcting the overall confidence corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension;
and the second calculation unit is used for calculating the credit investigation evaluation value of the user to be evaluated according to the feature data set and the corrected overall confidence of each target collection dimension in the feature data set.
Optionally, the computing apparatus for user credit investigation and evaluation value disclosed in the second aspect of the present application further includes a unit for constructing a confidence coefficient computing model; the construction unit of the confidence coefficient calculation model comprises:
the second acquisition unit is used for acquiring a training sample set used for training and testing the confidence coefficient calculation model, wherein the training sample set comprises feedback data of a plurality of problems belonging to the same dimensionality, and the feedback data of the problems are acquired from different data sources respectively;
the cross validation unit is used for cross validation of the feedback data of the problems so as to divide the feedback data of the problems into a feedback data group without data consistency and a feedback data group with data consistency;
a replacing unit, configured to obtain target data corresponding to each data in the feedback data group with data consistency based on each data in the feedback data group with data consistency, and replace each data in the feedback data group corresponding to the target data in the training sample set with the target data;
the third calculating unit is used for calculating the probability of each data in the feedback data group without data consistency, the probability of each target data and the overall confidence of the training sample set after data replacement;
and the training test unit is used for taking the probability of each piece of data, the probability of each piece of target data and the overall confidence coefficient of the training sample set subjected to data replacement as the input of the confidence coefficient calculation model and carrying out training test on the confidence coefficient calculation model.
Optionally, the display unit includes:
the screening unit is used for screening the verification problems corresponding to the target collection dimensionality from a plurality of verification problems in a local question bank aiming at each target collection dimensionality, or determining the verification problems corresponding to the target collection dimensionality from a cloud database and outputting the verification problems; the verification problem corresponding to each target collection dimension is obtained by screening based on the feedback data of the collection problem corresponding to each target collection dimension;
and the display subunit is used for displaying the verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module.
Optionally, the first obtaining unit includes:
the first obtaining subunit is configured to obtain feedback data of the verification problem corresponding to each target collection dimension, which is input by a user to be evaluated;
the verification unit is used for verifying the correctness of the feedback data of the verification problem corresponding to each target collection dimension respectively;
if the correctness of the feedback data of the verification problem corresponding to the target collection dimension cannot be verified, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is unverifiable;
if the feedback data of the verification problem corresponding to the target collection dimension is verified to be wrong, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is low in confidence;
and if the feedback data of the verification problem corresponding to the target collection dimension is verified to be correct, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is high in confidence.
Optionally, the modifying unit includes:
a first correcting subunit, configured to calculate, if a verification result of the feedback data of the problem corresponding to the target collection dimension is that verification cannot be performed, a product of a preset first correction coefficient corresponding to the verification problem and an overall confidence corresponding to the target collection dimension, so as to obtain an overall confidence corresponding to the target collection dimension after correction; wherein the first correction coefficient is less than 1;
a second correcting subunit, configured to calculate, if a verification result of the feedback data of the problem corresponding to the target collection dimension is that a confidence degree is low, a product of a preset second correction coefficient corresponding to the verification problem and an overall confidence degree corresponding to the target collection dimension, so as to obtain an overall confidence degree corresponding to the target collection dimension after correction; wherein the second correction coefficient is less than 1;
a third correction subunit, configured to, if a verification result of the feedback data of the problem corresponding to the target collection dimension is higher than a confidence level, calculate a product of a preset third correction coefficient corresponding to the verification problem and an overall confidence level corresponding to the target collection dimension, to obtain an overall confidence level corresponding to the target collection dimension after correction; wherein the third correction factor is equal to 1.
Optionally, the second computing unit includes:
the second calculating subunit is used for calculating to obtain a credit calculation value corresponding to each target collection dimension according to a preset score table of each target collection dimension, the question set and a probability distribution function corresponding to each target collection dimension in the question set; the scoring table of each target collection dimension is used for explaining credit calculation values corresponding to each question under each target collection dimension; the probability distribution function corresponding to each target collection dimension in the problem set is used for calculating the weight probability corresponding to each problem in the problem set;
the adjusting unit is used for adjusting the credit calculation value corresponding to each target collection dimension according to the corrected overall confidence coefficient of each target collection dimension in the feature data set to obtain the adjusted credit calculation value corresponding to each target collection dimension;
and the summing unit is used for summing the credit calculation values corresponding to each adjusted target collection dimension to obtain the credit investigation evaluation value of the user to be evaluated.
According to the technical scheme, in the calculation method of the user credit investigation evaluation value provided by the embodiment of the application, the problem set corresponding to the user to be evaluated is determined; wherein the problem set comprises problems of at least one target collection dimension; each target collection dimension corresponds to at least one question; displaying the question set to the user to be evaluated through a man-machine interaction module, and acquiring feedback data of the question set; wherein the feedback data of the problem set comprises: feedback data of all problems of each target collection dimension input by a user to be evaluated; feedback data of each problem of each target collection dimension are obtained through different source channels respectively; performing feature extraction according to the feedback data of the problem set to obtain a feature data set; wherein the feature data set comprises: feature data corresponding to the feedback data of each question of each target collection dimension; then inputting the feature data set into a pre-constructed confidence coefficient calculation model to obtain the overall confidence coefficient of each target collection dimension; then, according to the overall confidence of each target collection dimension, displaying a verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module; then obtaining feedback data of the verification problem corresponding to each target collection dimension input by a user to be evaluated, verifying the confidence of the feedback data of the problem corresponding to the target collection dimension by using the feedback data of the verification problem corresponding to each target collection dimension, and obtaining a verification result of the feedback data of the problem corresponding to each target collection dimension; correcting the overall confidence corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension; (ii) a And finally, calculating to obtain the credit investigation evaluation value of the user to be evaluated according to the feature data set and the corrected overall confidence of each target collection dimension in the feature data set. So as to achieve the purpose of efficiently investigating the client and calculating the credit score.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a flowchart illustrating a method for calculating a user credit investigation evaluation value according to an embodiment of the present disclosure;
FIG. 2 is a flowchart illustrating a method for calculating a credit investigation evaluation value of a user according to another embodiment of the present application;
FIG. 3 is a flowchart illustrating a method for calculating a credit investigation evaluation value of a user according to another embodiment of the present application;
FIG. 4 is a flowchart illustrating a method for calculating a credit investigation evaluation value of a user according to another embodiment of the present application;
FIG. 5 is a schematic structural diagram of a computing apparatus for user credit investigation and evaluation values according to another embodiment of the present application;
fig. 6 is a schematic structural diagram of a construction unit of a confidence computation model according to another embodiment of the present application;
FIG. 7 is a schematic structural diagram of a display unit according to another embodiment of the present application;
fig. 8 is a schematic structural diagram of a modification unit according to another embodiment of the present application;
fig. 9 is a schematic structural diagram of a second computing unit according to another embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
In this application, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
Referring to fig. 1, the embodiment of the present application discloses a method for calculating a user credit investigation evaluation value, which specifically includes the following steps:
s101, determining a problem set corresponding to a user to be evaluated.
The user to be evaluated can be a personal user, an enterprise or other various organizations; the problem set comprises at least one problem in a target collection dimension; each target collection dimension corresponds to at least one question;
it should be noted that, in the embodiment of the present application, a local question bank including a plurality of questions needs to be established in advance. Specifically, a plurality of questions are set for each target collection dimension, the attribute of each question is set, and then the questions and the attribute are stored in a local question bank. Optionally, the attributes of the question include an output mode, a question feedback mode, a purpose, a target collection dimension corresponding to the question, an importance coefficient and a support coefficient of the question, an association relationship with other questions, and a confidence confirmation mode.
Specifically, the process of determining the problem set for the user to be evaluated may be to perform random screening in a pre-established local problem library to determine the problem set that needs to be presented to the user to be evaluated.
And S102, displaying the problem set to the user to be evaluated through the man-machine interaction module, and acquiring feedback data of the problem set.
Wherein the feedback data of the problem set comprises: feedback data of all problems of each target collection dimension input by a user to be evaluated; and feedback data of each question of each target collection dimension are obtained through different source channels respectively.
It should be noted that the human-computer interaction module may be a device that simulates human-to-human conversation through an artificial intelligence technique, and displays each problem in the problem set in a way of conversation with the user to be evaluated, so as to achieve the whole process of simulating the manual review link in the whole process. The human-computer interaction module can be composed of a display device, a sound output device, an input device, a voice synthesis device, a virtual character synthesis device, a sound acquisition device, a video acquisition device, a storage device and a voice recognition device. The display device is used for displaying virtual characters and assisting in displaying conversation contents, such as a display and the like; sound output devices for outputting sound, such as speakers; the input device is a device which can be manually input or confirm the problem by a client, such as a keyboard, a touch screen, a mouse and the like; the voice synthesis device is used for synthesizing the question into natural voice and outputting the natural voice through the sound output equipment; the virtual character synthesis device is used for generating a virtual role, adjusting the action expression according to the conversation content and displaying the action expression to the client through the display equipment; the sound collection equipment is used for collecting voice information of a client, such as a microphone and the like; the video acquisition equipment is used for collecting image information of a client, such as a camera; the storage device is used for storing the collected data and carrying out subsequent analysis.
Specifically, after the problem set is displayed to the user to be evaluated through the human-computer interaction module, different types of feedback data can be collected through a microphone, a touch screen, video acquisition equipment and the like in the human-computer interaction module.
And S103, performing feature extraction on the feedback data of the problem set to obtain a feature data set.
Wherein the feature data set comprises: and characteristic data corresponding to the feedback data of each question of each target collection dimension.
Specifically, feedback data of the user to be evaluated for the problem set can be obtained through the polymorphic feature recognition module, and feature extraction is performed on the obtained feedback data of the user to be evaluated for the problem set to obtain a feature data set.
It should be noted that the polymorphic feature recognition module collects various features through technical means, and carries out risk prompt on potential fraud of the answer questions of the user to be evaluated from the aspect that the traditional investigation cannot effectively recognize the multiple features, wherein the risk prompt includes but is not limited to a micro-expression feature module, a body language feature module, a voiceprint feature module, a reaction speed recognition module, a biological information feature module and the like.
For example, the micro-expression feature module may present a fraud risk by recognizing an expression reaction prompt when the user to be evaluated answers a question; the body language feature module prompts possible fraud risks by identifying body reactions when the user to be evaluated answers questions; the voiceprint recognition module prompts possible fraud risks by recognizing the voice, tone and speed when the investigated object answers the question; the response speed identification module prompts possible fraud risks by identifying the thinking speed of the user to be evaluated for answering the questions; the biometric feature module alerts of possible fraud risk by identifying certain physiological characteristics of the user to be assessed, such as heart rate, respiration, etc.
And S104, inputting the feature data set into a pre-constructed confidence coefficient calculation model to obtain the overall confidence coefficient of each target collection dimension.
Optionally, in another embodiment of the present application, as shown in fig. 2, the method for constructing the confidence coefficient calculation model includes the following steps:
s201, obtaining a training sample set used for training and testing the confidence coefficient calculation model.
The training sample set comprises feedback data of a plurality of problems belonging to the same dimensionality, and the feedback data of the problems are collected from different data sources respectively.
There may be a plurality of dimensions for training and testing the model, and the training sample set to which the dimension belongs is processed by the present embodiment, and then the training sample set can be used as the input of the model to train and test the model.
For any dimension, the feedback data of the problem data of the dimension can be acquired from all feedback data sources recording the problem of the dimension, and the feedback data of the problem data of the dimension from all the feedback data sources of the dimension can be stored in the training sample set of the dimension. However, in an actual business scenario, the values of some data sources are reliable/authoritative, and the values of other data sources may be ignored if the values can be obtained from these data sources.
S202, performing cross validation on the feedback data of the plurality of problems to divide the feedback data of the plurality of problems into a feedback data group without data consistency and a feedback data group with data consistency.
The feedback data with data consistency is found out from the feedback data with multiple problems, and the feedback data group with data consistency indicates that the difference between the data can be ignored in subsequent applications, for example, for the dimension of profit, if the profit collected from different data sources is-500 ten thousand yuan and +700 ten thousand yuan respectively, the difference between the profit collected from different data sources can be indicated to be too large through the two values, the difference between the two values cannot be ignored, and further the two data are not consistent, the two data are divided into the data groups without data consistency; if the profits collected from different data sources are +700 ten thousand yuan and +650 ten thousand yuan respectively, the two data are close to each other, the difference between the two data can be ignored, the two data have data consistency, and the two data can be divided into the same data group with data consistency.
And S203, obtaining target data corresponding to each data in the feedback data group with data consistency based on each data in the feedback data group with data consistency, and replacing each data in the feedback data group corresponding to the target data in the training sample set with the target data.
The way of replacing with the target data is: target data is added to the set of training samples, and each data in the set of feedback data corresponding to the target data is deleted from the set of training samples. The target data can be replaced because the target data is derived based on the respective data in its corresponding feedback data set, which can characterize the respective data in its corresponding feedback data set.
And S204, calculating the probability of each data in the feedback data group without data consistency, the probability of each target data and the overall confidence of the training sample set after data replacement.
In this embodiment, the data distribution in the training sample set after data replacement is represented by the probability of each data and the probability of each target data. And replacing each data in the feedback data group in the training sample set by the target data of the feedback data group to add a reliable target data in the training sample set so as to increase the confidence degree of the training sample set, and for the feedback data group without data consistency, omitting the steps of calculating the target data and replacing (namely, keeping the original data and the confidence degree of each data in the feedback data group without data consistency), so that each data in the feedback data group without data consistency does not reduce the confidence degree of the training sample set, and the accuracy and the reliability of the confidence degree of the training sample set are improved by the feedback data group with data consistency and the data group without data consistency.
Specifically, one way to calculate the probability of each data and the probability of each target data is: and calculating the probability of each data in the feedback data group without data consistency, the probability of each target data and the confidence of the training sample set after data replacement based on the confidence of the data source corresponding to each data in the feedback data group with data consistency and the confidence of each target data.
And S205, taking the probability of each data, the probability of each target data and the overall confidence coefficient of the training sample set after data replacement as the input of the confidence coefficient calculation model, and carrying out training test on the confidence coefficient calculation model.
Through the training test of the confidence coefficient calculation model, the probability of each data and the probability of each target data can represent the probability distribution of the feedback data of a plurality of problems belonging to the same dimensionality, and the accuracy of the model is improved in a mode of passing through the probability distribution of the feedback data of the plurality of problems of the same dimensionality and the confidence coefficient of the training sample set.
And S105, displaying the verification problem corresponding to each target collection dimension to the user to be evaluated through the man-machine interaction module according to the overall confidence of each target collection dimension.
It should be noted that the verification problem is a problem of feedback data for verifying the problem of the target collection dimension. For example, where the feedback data is the name of a university that the user has read at the subject, the verification question may be asking the user for a workout of the school that the user has read at the subject. Therefore, the verification problem needs to be obtained by screening according to the feedback data of the problem of the target collection dimension.
Therefore, after step S104 is executed, based on the feedback data of the problem corresponding to each target collection dimension, the verification problem corresponding to the feedback data of the problem corresponding to each target collection dimension is obtained by screening, and is displayed to the user to be evaluated through the human-computer interaction module.
It should be noted that, the number of verification questions corresponding to the output feedback data of the question corresponding to each target collection dimension may specifically be determined according to the total confidence of the target collection dimensions obtained in step S104. Therefore, the lower the overall confidence of the target collection dimension is, the more the number of verification problems corresponding to the determined feedback data of the problem corresponding to the target collection dimension to be output is.
Accordingly, when the overall confidence of the target collection dimension is high enough, the number of verification problems corresponding to the determined feedback data of the problem corresponding to the target collection dimension to be output may be zero.
Optionally, in another embodiment of the present application, an implementation manner of step S105, as shown in fig. 3, includes the following steps:
s301, aiming at each target collection dimension, screening the verification problems corresponding to the target collection dimension from the verification problems in the local question bank, or determining the verification problems corresponding to the target collection dimension from the cloud database to output.
And the verification problem corresponding to each target collection dimension is obtained by screening the feedback data of the collection problem corresponding to each target collection dimension.
That is to say, in the embodiment of the present application, the verification problem may be obtained by screening from a local question bank, or may be obtained by screening from a cloud database. Specifically, if the verification problem is obtained by screening from the local question bank, a plurality of verification problems need to be set in the local question bank in advance according to the collection problem, and an association relationship between the verification problem and the collection problem is established. Subsequently, feedback data of the problem of the dimension can be collected according to different targets, and the verification problem corresponding to the feedback data of the problem in the dimension is obtained by screening from the local question bank.
Specifically, the verification problems are screened from the cloud database, after feedback data of the problem of the target collection dimension is obtained, a plurality of verification problems corresponding to the feedback data are collected from the network according to the feedback data to form the cloud database, and then the verification problems are screened from the cloud database.
S302, displaying the verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module.
Specifically, the manner of presenting the verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module may be the manner in step S102, which is not described herein again.
S106, obtaining feedback data of the verification problems corresponding to each target collection dimension input by a user to be evaluated, verifying the confidence degree of the feedback data of the problems corresponding to the target collection dimension by using the feedback data of the verification problems corresponding to each target collection dimension, and obtaining a verification result of the feedback data of the problems corresponding to each target collection dimension.
Specifically, whether the confidence of the obtained feedback data of the verification problem corresponding to the target collection dimension is enough to be credible is verified through the received feedback data of the verification problem corresponding to each target collection dimension input by the user.
Optionally, in another embodiment of the present application, an implementation manner of step S106 includes the following steps:
and verifying the correctness of the feedback data of the verification problem corresponding to each target collection dimension respectively.
Specifically, if the correctness of the feedback data of the verification problem corresponding to the target collection dimension cannot be verified, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is unverifiable; if the feedback data of the verification problem corresponding to the target collection dimension is verified to be wrong, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is low in confidence; and if the feedback data of the verification problem corresponding to the target collection dimension is verified to be correct, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is high in confidence.
S107, based on the verification result of the feedback data of the problem corresponding to each target collection dimension, correcting the overall confidence corresponding to the target collection dimension.
When the confidence degree of the feedback data of the verification problem corresponding to the target collection dimension is verified to be unreliable, the overall confidence degree corresponding to the target collection dimension is corrected, namely the overall confidence degree corresponding to the target collection dimension is correspondingly increased or decreased based on the verification result of the feedback data of the verification problem corresponding to each target collection dimension, or the overall confidence degree corresponding to the target collection dimension is not changed. When the verification result shows that the credibility of the feedback data of the verification problem corresponding to each target collection dimension is high, the overall confidence corresponding to the target collection dimension is not changed or increased correspondingly, and when the verification result shows that the credibility of the feedback data of the verification problem corresponding to the target collection dimension is low, the overall confidence corresponding to the target collection dimension is reduced.
It should be noted that, when it is verified whether the confidence of the feedback data of the verification question corresponding to the obtained target collection dimension is sufficient and reliable, the correctness of the feedback data of the verification question corresponding to each target collection dimension can be determined by searching the correct answer of the verification question and comparing the correct answer of the verification question with the feedback data input by the user.
Optionally, in another embodiment of the present application, an implementation manner of step S107 includes the following steps:
and if the feedback data of the verification problem corresponding to the target collection dimension cannot be verified, calculating the product of a first correction coefficient of the verification problem corresponding to the preset target collection dimension and the overall confidence coefficient of the target collection dimension to obtain the overall confidence coefficient of the corrected target collection dimension.
And if the confidence of the feedback data of the verification problem corresponding to the target collection dimension is low, calculating the product of a second correction coefficient of the verification problem corresponding to the preset target collection dimension and the overall confidence of the target collection dimension to obtain the overall confidence of the corrected target collection dimension.
And if the confidence of the feedback data of the verification problem corresponding to the target collection dimension is higher, calculating the product of a third correction coefficient of the verification problem corresponding to the preset target collection dimension and the overall confidence of the first problem set to obtain the overall confidence of the corrected target collection dimension.
Wherein the first correction coefficient and the second correction coefficient are both smaller than 1, and the first correction coefficient is larger than the second correction coefficient; the third correction factor is equal to 1.
That is to say, in the embodiment of the present application, when the verification result is both of the non-verification and the error, the overall confidence corresponding to the target collection dimension is reduced, and the reduction amplitude when the verification result is the error is larger than the reduction amplitude when the verification result is the non-verification. And when the verification result is correct, not changing the overall confidence corresponding to the target collection dimension. Of course, this is only one alternative way, and other correction coefficients may also be set to perform corresponding correction on the overall confidence corresponding to the target collection dimension, which all fall within the protection scope of the present application.
And S108, calculating to obtain the credit investigation evaluation value of the user to be evaluated according to the corrected overall confidence of the feature data set and each target collection dimension in the feature data set.
Specifically, according to a preset calculation method, the credit investigation evaluation value of the user to be evaluated is calculated by using the acquired feature data set and the corrected overall confidence of each target collection dimension in the feature data set.
Optionally, in another embodiment of the present application, an implementation manner of step S108, as shown in fig. 4, includes the following steps:
s401, calculating to obtain a credit calculation value corresponding to each target collection dimension according to a preset scoring table of each target collection dimension, a preset problem set and a preset probability distribution function corresponding to each target collection dimension in the problem set.
The scoring table of each target collection dimension is used for explaining credit calculation values corresponding to the problems under each target collection dimension; the probability distribution function corresponding to each target collection dimension in the problem set is used for calculating the weight probability corresponding to each problem in the problem set.
It should be noted that the probability distribution function corresponding to each target collection dimension in the problem set is constructed according to each problem in each target collection dimension and the preset confidence level of the source channel corresponding to each problem in each target collection dimension; the confidence of the source channel corresponding to each question is used for explaining the reliability of the credit data acquired by the source channel.
Specifically, a corresponding mapping function may be constructed according to the scoring table for each target collection dimension. The scoring table of the ith target collection dimension constructs a corresponding mapping function Fi(x) Where x is credit data in the ith target collection dimension. The credit data x in the ith target collection dimension may be divided into k sets, S1, S2, … …, Sk, respectively. Substitution of credit data belonging to the set S1 into the mapping function Fi(x) The credit calculation value V1 is obtained, i.e. the credit calculation value corresponding to the credit data in S1 is V1, and the credit calculation value corresponding to the credit data in Sk is Vk. For example, the credit data x in the revenue target collection dimension is divided into 6 sets of 1000-2000, 2000-4000, 4000-6000, 6000-10000, 10000-20000 and more than 20000, wherein when the credit data belongs to 1000-2000, the corresponding credit calculation value is 3 points, when the credit data belongs to 2000-4000, the corresponding credit calculation value is 4 points, when the credit data belongs to 4000-6000, the corresponding credit calculation value is 6 points, when the credit data belongs to 6000-10000, the corresponding credit calculation value is 8 points, when the credit data belongs to 10000-20000, the corresponding credit calculation value is 10 points, and when the credit calculation value belongs to 20000, the corresponding credit calculation value is 15 points.
In practical applications, if a credit evaluation policy of some products changes, etc., the scoring table of each target collection dimension does not necessarily cover the credit calculation values corresponding to the credit data of all target collection dimensions. For example, for an age target collection dimension, credit data may be divided into sets of 18-24, 25-30, 30-45, 45-55, and 55-70 years old, but credit calculations for under 18, and over 70 years old are not covered. And if the credit calculation value corresponding to the credit data is not found in the corresponding scoring table, making the credit calculation value corresponding to the credit data a preset default credit calculation value.
Optionally, if the user to be evaluated does not have credit data under part of the feature parameters in the credit parameter set, the credit calculation value of the feature parameter for which the user to be evaluated lacks credit data may be set as the default credit calculation value.
S402, adjusting the credit calculation value corresponding to each target collection dimension according to the corrected overall confidence of each target collection dimension in the feature data set to obtain the credit calculation value corresponding to each adjusted target collection dimension.
Specifically, the corrected overall confidence of each target collection dimension and the credit calculation value corresponding to each target collection dimension are substituted into a preset formula, so as to obtain the adjusted credit calculation value corresponding to each target collection dimension.
Wherein, the preset formula is as follows: a'i＝Ai×σ(αi)+Vi 0×(1-σ(αi))；A‘iCalculating a credit value corresponding to the adjusted ith target collection dimension; a. theiCredit calculation value corresponding to ith target collection dimension αiCorrected overall confidence for the ith target collection dimension, σ (α)i) Collecting corrected overall confidence of dimensionality for the adjusted ith target; vi0 is the default credit calculation value of the preset ith target collection dimension.
And S403, summing the credit calculation values corresponding to each adjusted target collection dimension to obtain the credit investigation evaluation value of the user to be evaluated.
According to the technical scheme, in the calculation method of the user credit investigation evaluation value provided by the embodiment of the application, the problem set corresponding to the user to be evaluated is determined; wherein the problem set comprises problems of at least one target collection dimension; each target collection dimension corresponds to at least one question; displaying the question set to the user to be evaluated through a man-machine interaction module, and acquiring feedback data of the question set; wherein the feedback data of the problem set comprises: feedback data of all problems of each target collection dimension input by a user to be evaluated; feedback data of each problem of each target collection dimension are obtained through different source channels respectively; performing feature extraction according to the feedback data of the problem set to obtain a feature data set; wherein the feature data set comprises: feature data corresponding to the feedback data of each question of each target collection dimension; then inputting the feature data set into a pre-constructed confidence coefficient calculation model to obtain the overall confidence coefficient of each target collection dimension; then, according to the overall confidence of each target collection dimension, displaying a verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module; then, feedback data of the verification problem corresponding to each target collection dimension, which is input by a user, is obtained, and the overall confidence of the target collection dimension is corrected based on the feedback data of the verification problem corresponding to each target collection dimension; and finally, calculating to obtain the credit investigation evaluation value of the user to be evaluated according to the feature data set and the corrected overall confidence of each target collection dimension in the feature data set. So as to achieve the purpose of efficiently investigating the client and calculating the credit score. .
Referring to fig. 5, an embodiment of the present application discloses a device for calculating a user credit investigation evaluation value, including:
the determining unit 501 is configured to determine a problem set corresponding to a user to be evaluated.
Wherein the problem set comprises problems of at least one target collection dimension; each target collection dimension corresponds to at least one question.
And the execution unit 502 is used for displaying the problem set to the user to be evaluated through the human-computer interaction module and acquiring feedback data of the problem set.
Wherein the feedback data of the problem set comprises: feedback data of all problems of each target collection dimension input by a user to be evaluated; and feedback data of each question of each target collection dimension are obtained through different source channels respectively.
The extracting unit 503 is configured to perform feature extraction on the feedback data of the problem set to obtain a feature data set.
Wherein the feature data set comprises: and characteristic data corresponding to the feedback data of each question of each target collection dimension.
The first calculating unit 504 is configured to input the feature data set into a pre-constructed confidence calculating model, so as to obtain an overall confidence of each target collection dimension.
Optionally, in another embodiment of the present application, the computing device for user credit investigation evaluation value may further include a unit for constructing a confidence level computing model.
As shown in fig. 6, the unit for constructing the confidence calculation model includes:
an obtaining unit 601, configured to obtain a training sample set used for performing a training test on the confidence level calculation model.
The training sample set comprises feedback data of a plurality of problems belonging to the same dimensionality, and the feedback data of the problems are collected from different data sources respectively.
And a cross validation unit 602, configured to cross-validate the feedback data of the multiple problems to divide the feedback data of the multiple problems into a feedback data group without data consistency and a feedback data group with data consistency.
A replacing unit 603, configured to obtain target data corresponding to each data in the feedback data set with data consistency based on each data in the feedback data set with data consistency, and replace each data in the feedback data set corresponding to the target data in the training sample set with the target data.
And a third calculating unit 604, configured to calculate probabilities of the respective data in the feedback data group without data consistency, probabilities of the respective target data, and an overall confidence of the training sample set after data replacement.
The training test unit 605 is configured to perform a training test on the confidence level calculation model by using the probability of each data, the probability of each target data, and the overall confidence level of the training sample set after data replacement as input of the confidence level calculation model.
For a specific working process of the unit disclosed in the above embodiment of the present application, reference may be made to the content of the corresponding method embodiment, as shown in fig. 2, which is not described herein again.
And the display unit 505 is configured to display, according to the overall confidence of each target collection dimension, the verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module.
Optionally, in another embodiment of the present application, an implementation manner of the display unit 505, as shown in fig. 7, includes:
the screening unit 701 is configured to, for each target collection dimension, screen a plurality of verification problems in the local question bank to obtain a verification problem corresponding to the target collection dimension, or determine a verification problem corresponding to the target collection dimension from the cloud database and output the verification problem.
And the verification problem corresponding to each target collection dimension is obtained by screening the feedback data of the collection problem corresponding to each target collection dimension.
And the display subunit 702 is configured to display, to the user to be evaluated, the verification problem corresponding to each target collection dimension through the human-computer interaction module.
The first obtaining unit 506 is configured to obtain feedback data of the verification problem corresponding to each target collection dimension input by the user to be evaluated, and verify the confidence of the feedback data of the problem corresponding to the target collection dimension by using the feedback data of the verification problem corresponding to each target collection dimension to obtain a verification result of the feedback data of the problem corresponding to each target collection dimension.
A correcting unit 507, configured to correct the overall confidence corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension.
Optionally, in another embodiment of the present application, an implementation manner of the correcting unit 507, as shown in fig. 8, includes:
a first modification subunit 801, configured to, if the feedback data of the verification problem corresponding to the target collection dimension cannot be verified, calculate a product of a first modification coefficient of the verification problem corresponding to the preset target collection dimension and an overall confidence of the target collection dimension, to obtain an overall confidence of the modified target collection dimension; wherein the first correction coefficient is smaller than 1.
A second modification subunit 802, configured to calculate, if the confidence of the feedback data of the verification problem corresponding to the target collection dimension is low, a product of a second modification coefficient of the verification problem corresponding to the preset target collection dimension and the overall confidence of the target collection dimension, so as to obtain an overall confidence of the modified target collection dimension; wherein the second correction coefficient is smaller than 1.
A third modification subunit 803, configured to, if the confidence of the feedback data of the verification problem corresponding to the target collection dimension is higher, calculate a product of a third modification coefficient of the verification problem corresponding to the preset target collection dimension and the overall confidence of the first problem set, to obtain an overall confidence of the modified target collection dimension; wherein the third correction factor is equal to 1.
For specific working processes of the units disclosed in the above embodiments of the present application, reference may be made to the contents of the corresponding method embodiments, which are not described herein again.
And a second calculating unit 508, configured to calculate, according to the feature data set and the corrected overall confidence of each target collection dimension in the feature data set, a credit investigation evaluation value of the user to be evaluated.
For a specific working process of the unit disclosed in the above embodiment of the present application, reference may be made to the content of the corresponding method embodiment, as shown in fig. 1, which is not described herein again.
Optionally, in another embodiment of the present application, an implementation manner of the second calculating unit 508, as shown in fig. 9, includes:
the second calculating subunit 901 is configured to calculate a credit calculation value corresponding to each target collection dimension according to a preset score table of each target collection dimension, the question set, and a probability distribution function corresponding to each target collection dimension in the question set.
The scoring table of each target collection dimension is used for explaining credit calculation values corresponding to the problems under each target collection dimension; the probability distribution function corresponding to each target collection dimension in the problem set is used for calculating the weight probability corresponding to each problem in the problem set.
An adjusting unit 902, configured to adjust, according to the corrected overall confidence of each target collection dimension in the feature data set, a credit calculation value corresponding to each target collection dimension to obtain a credit calculation value corresponding to each adjusted target collection dimension.
And the summing unit 903 is configured to sum the credit calculation values corresponding to each adjusted target collection dimension to obtain a credit investigation evaluation value of the user to be evaluated.
For a specific working process of the unit disclosed in the above embodiment of the present application, reference may be made to the content of the corresponding method embodiment, as shown in fig. 4, which is not described herein again.
According to the technical scheme, in the computing device for the user credit investigation evaluation value provided by the embodiment of the application, the problem set corresponding to the user to be evaluated is determined through the determining unit 501; wherein the problem set comprises problems of at least one target collection dimension; each target collection dimension corresponds to at least one question; displaying the question set to the user to be evaluated by using a man-machine interaction module through an execution unit 502, and acquiring feedback data of the question set; wherein the feedback data of the problem set comprises: feedback data of all problems of each target collection dimension input by a user to be evaluated; feedback data of each problem of each target collection dimension are obtained through different source channels respectively; extracting features according to the feedback data of the problem set by using an extraction unit 503 to obtain a feature data set; wherein the feature data set comprises: feature data corresponding to the feedback data of each question of each target collection dimension; then, the feature data set is input to a pre-constructed confidence coefficient calculation model by using a first calculation unit 504, so that the overall confidence coefficient of each target collection dimension is obtained; then, a display unit 505 is used for displaying the verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module according to the overall confidence of each target collection dimension; feedback data of the verification problem corresponding to each target collection dimension input by a user to be evaluated is obtained by the first obtaining unit 506, the confidence of the feedback data of the problem corresponding to the target collection dimension is verified by the feedback data of the verification problem corresponding to each target collection dimension, and a verification result of the feedback data of the problem corresponding to each target collection dimension is obtained; then, the correction unit 507 is used for correcting the overall confidence corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension; finally, the credit investigation evaluation value of the user to be evaluated is calculated by using the second calculating unit 508 according to the feature data set and the corrected overall confidence of each target collection dimension in the feature data set, so as to achieve the purpose of efficiently investigating the client and calculating the credit score.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (13)

1. A method for calculating a user credit investigation evaluation value is characterized by comprising the following steps:
determining a question set corresponding to a user to be evaluated; wherein the problem set comprises problems of at least one target collection dimension; each target collection dimension corresponds to at least one question;
displaying the question set to the user to be evaluated through a man-machine interaction module, and acquiring feedback data of the question set; wherein the feedback data of the problem set comprises: feedback data of all problems of each target collection dimension input by a user to be evaluated; feedback data of each problem of each target collection dimension are obtained through different source channels respectively;
performing characteristic extraction on the feedback data of the problem set to obtain a characteristic data set; wherein the feature data set comprises: feature data corresponding to the feedback data of each question of each target collection dimension;
inputting the feature data set into a pre-constructed confidence coefficient calculation model to obtain the overall confidence coefficient of each target collection dimension;
according to the overall confidence of each target collection dimension, displaying a verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module;
obtaining feedback data of the verification problem corresponding to each target collection dimension input by a user to be evaluated, verifying the confidence of the feedback data of the problem corresponding to the target collection dimension by using the feedback data of the verification problem corresponding to each target collection dimension, and obtaining a verification result of the feedback data of the problem corresponding to each target collection dimension;
correcting the overall confidence corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension;
and calculating the credit investigation evaluation value of the user to be evaluated according to the feature data set and the corrected overall confidence of each target collection dimension in the feature data set.
2. The method of claim 1, wherein the confidence computation model is constructed by:
acquiring a training sample set for training and testing a confidence level calculation model, wherein the training sample set comprises feedback data of a plurality of problems belonging to the same dimensionality, and the feedback data of the problems are acquired from different data sources respectively;
performing cross validation on the feedback data of the plurality of problems to divide the feedback data of the plurality of problems into a feedback data group without data consistency and a feedback data group with data consistency;
obtaining target data corresponding to each data in the feedback data group with data consistency based on each data in the feedback data group with data consistency, and replacing each data in the feedback data group corresponding to the target data in the training sample set with the target data;
calculating the probability of each data in the feedback data group without data consistency, the probability of each target data and the overall confidence of the training sample set after data replacement;
and taking the probability of each piece of data, the probability of each piece of target data and the overall confidence coefficient of the training sample set after data replacement as the input of the confidence coefficient calculation model, and training the confidence coefficient calculation model.
3. The method according to claim 1, wherein the presenting, by the human-computer interaction module, the verification problem corresponding to each target collection dimension to the user to be evaluated according to the overall confidence of each target collection dimension comprises:
for each target collection dimension, screening a plurality of verification problems in a local question bank to obtain a verification problem corresponding to the target collection dimension, or determining a verification problem corresponding to the target collection dimension from a cloud database to output; the verification problem corresponding to each target collection dimension is obtained by screening based on the feedback data of the collection problem corresponding to each target collection dimension;
and displaying the verification problem corresponding to each target collection dimension to the user to be evaluated through the man-machine interaction module.
4. The method according to claim 1, wherein the verifying the confidence of the feedback data of the problem corresponding to the target collection dimension by using the feedback data of the verification problem corresponding to each target collection dimension to obtain the verification result of the feedback data of the problem corresponding to each target collection dimension comprises:
verifying the correctness of the feedback data of the verification problem corresponding to each target collection dimension respectively;
if the correctness of the feedback data of the verification problem corresponding to the target collection dimension cannot be verified, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is unverifiable;
if the feedback data of the verification problem corresponding to the target collection dimension is verified to be wrong, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is low in confidence;
and if the feedback data of the verification problem corresponding to the target collection dimension is verified to be correct, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is high in confidence.
5. The method of claim 4, wherein the modifying the overall confidence level corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension comprises:
if the verification result of the feedback data of the problem corresponding to the target collection dimension is that the feedback data cannot be verified, calculating the product of a preset first correction coefficient corresponding to the verification problem and the overall confidence corresponding to the target collection dimension to obtain the overall confidence corresponding to the corrected target collection dimension; wherein the first correction coefficient is less than 1;
if the verification result of the feedback data of the problem corresponding to the target collection dimension is low in confidence, calculating the product of a preset second correction coefficient corresponding to the verification problem and the overall confidence corresponding to the target collection dimension to obtain the overall confidence corresponding to the corrected target collection dimension; wherein the second correction coefficient is less than 1;
if the verification result of the feedback data of the problem corresponding to the target collection dimension is higher in confidence, calculating the product of a preset third correction coefficient corresponding to the verification problem and the overall confidence corresponding to the target collection dimension to obtain the overall confidence corresponding to the corrected target collection dimension; wherein the third correction factor is equal to 1.
6. The method of claim 1, wherein the calculating a credit investigation evaluation value of the user to be evaluated according to the feature data set and the revised overall confidence of each target collection dimension in the feature data set comprises:
calculating to obtain a credit calculation value corresponding to each target collection dimension according to a preset scoring table of each target collection dimension, the question set and a probability distribution function corresponding to each target collection dimension in the question set; the scoring table of each target collection dimension is used for explaining credit calculation values corresponding to each question under each target collection dimension; the probability distribution function corresponding to each target collection dimension in the problem set is used for calculating the weight probability corresponding to each problem in the problem set;
adjusting the credit calculation value corresponding to each target collection dimension according to the corrected overall confidence of each target collection dimension in the feature data set to obtain the adjusted credit calculation value corresponding to each target collection dimension;
and summing the credit calculation values corresponding to each adjusted target collection dimension to obtain the credit investigation evaluation value of the user to be evaluated.
7. The method according to claim 6, wherein the probability distribution function corresponding to each target collection dimension in the problem set is constructed according to each problem in each target collection dimension and a preset confidence level of a source channel corresponding to each problem in each target collection dimension; the confidence of the source channel corresponding to each question is used for explaining the reliability of the credit data acquired by the source channel.
8. A device for calculating a credit investigation evaluation value of a user, comprising:
the determining unit is used for determining a problem set corresponding to a user to be evaluated; wherein the problem set comprises problems of at least one target collection dimension; each target collection dimension corresponds to at least one question;
the execution unit is used for displaying the question set to the user to be evaluated through a human-computer interaction module and acquiring feedback data of the question set; wherein the feedback data of the problem set comprises: feedback data of all problems of each target collection dimension input by a user to be evaluated; feedback data of each problem of each target collection dimension are obtained through different source channels respectively;
the extraction unit is used for carrying out feature extraction on the feedback data of the problem set to obtain a feature data set; wherein the feature data set comprises: feature data corresponding to the feedback data of each question of each target collection dimension;
the first calculation unit is used for inputting the feature data set into a pre-constructed confidence coefficient calculation model to obtain the overall confidence coefficient of each target collection dimension;
the display unit is used for displaying the verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module according to the overall confidence of each target collection dimension;
the first obtaining unit is used for obtaining feedback data of the verification problem corresponding to each target collection dimension input by a user to be evaluated, verifying the confidence coefficient of the feedback data of the problem corresponding to the target collection dimension by using the feedback data of the verification problem corresponding to each target collection dimension, and obtaining the verification result of the feedback data of the problem corresponding to each target collection dimension;
the correction unit is used for correcting the overall confidence corresponding to the target collection dimension based on the verification result of the feedback data of the problem corresponding to each target collection dimension;
and the second calculation unit is used for calculating the credit investigation evaluation value of the user to be evaluated according to the feature data set and the corrected overall confidence of each target collection dimension in the feature data set.
9. The apparatus of claim 8, further comprising a confidence computation model building unit; the construction unit of the confidence coefficient calculation model comprises:
the second acquisition unit is used for acquiring a training sample set used for training and testing the confidence coefficient calculation model, wherein the training sample set comprises feedback data of a plurality of problems belonging to the same dimensionality, and the feedback data of the problems are acquired from different data sources respectively;
the cross validation unit is used for cross validation of the feedback data of the problems so as to divide the feedback data of the problems into a feedback data group without data consistency and a feedback data group with data consistency;
a replacing unit, configured to obtain target data corresponding to each data in the feedback data group with data consistency based on each data in the feedback data group with data consistency, and replace each data in the feedback data group corresponding to the target data in the training sample set with the target data;
the third calculating unit is used for calculating the probability of each data in the feedback data group without data consistency, the probability of each target data and the overall confidence of the training sample set after data replacement;
and the training test unit is used for taking the probability of each piece of data, the probability of each piece of target data and the overall confidence coefficient of the training sample set subjected to data replacement as the input of the confidence coefficient calculation model and carrying out training test on the confidence coefficient calculation model.
10. The apparatus of claim 8, wherein the display unit comprises:
the screening unit is used for screening the verification problems corresponding to the target collection dimensionality from a plurality of verification problems in a local question bank aiming at each target collection dimensionality, or determining the verification problems corresponding to the target collection dimensionality from a cloud database and outputting the verification problems; the verification problem corresponding to each target collection dimension is obtained by screening based on the feedback data of the collection problem corresponding to each target collection dimension;
and the display subunit is used for displaying the verification problem corresponding to each target collection dimension to the user to be evaluated through the human-computer interaction module.
11. The apparatus of claim 8, wherein the first obtaining unit comprises:
the first obtaining subunit is configured to obtain feedback data of the verification problem corresponding to each target collection dimension, which is input by a user to be evaluated;
the verification unit is used for verifying the correctness of the feedback data of the verification problem corresponding to each target collection dimension respectively;
if the correctness of the feedback data of the verification problem corresponding to the target collection dimension cannot be verified, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is unverifiable;
if the feedback data of the verification problem corresponding to the target collection dimension is verified to be wrong, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is low in confidence;
and if the feedback data of the verification problem corresponding to the target collection dimension is verified to be correct, determining that the verification result of the feedback data of the problem corresponding to the target collection dimension is high in confidence.
12. The apparatus of claim 11, wherein the modification unit comprises:
a first correcting subunit, configured to calculate, if a verification result of the feedback data of the problem corresponding to the target collection dimension is that verification cannot be performed, a product of a preset first correction coefficient corresponding to the verification problem and an overall confidence corresponding to the target collection dimension, so as to obtain an overall confidence corresponding to the target collection dimension after correction; wherein the first correction coefficient is less than 1;
a second correcting subunit, configured to calculate, if a verification result of the feedback data of the problem corresponding to the target collection dimension is that a confidence degree is low, a product of a preset second correction coefficient corresponding to the verification problem and an overall confidence degree corresponding to the target collection dimension, so as to obtain an overall confidence degree corresponding to the target collection dimension after correction; wherein the second correction coefficient is less than 1;
a third correction subunit, configured to, if a verification result of the feedback data of the problem corresponding to the target collection dimension is higher than a confidence level, calculate a product of a preset third correction coefficient corresponding to the verification problem and an overall confidence level corresponding to the target collection dimension, to obtain an overall confidence level corresponding to the target collection dimension after correction; wherein the third correction factor is equal to 1.
13. The apparatus of claim 8, wherein the second computing unit comprises:
the second calculating subunit is used for calculating to obtain a credit calculation value corresponding to each target collection dimension according to a preset score table of each target collection dimension, the question set and a probability distribution function corresponding to each target collection dimension in the question set; the scoring table of each target collection dimension is used for explaining credit calculation values corresponding to each question under each target collection dimension; the probability distribution function corresponding to each target collection dimension in the problem set is used for calculating the weight probability corresponding to each problem in the problem set;
the adjusting unit is used for adjusting the credit calculation value corresponding to each target collection dimension according to the corrected overall confidence coefficient of each target collection dimension in the feature data set to obtain the adjusted credit calculation value corresponding to each target collection dimension;
and the summing unit is used for summing the credit calculation values corresponding to each adjusted target collection dimension to obtain the credit investigation evaluation value of the user to be evaluated.
CN201911185326.5A 2019-11-27 2019-11-27 Method and device for calculating credit investigation evaluation value of user Active CN110930244B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911185326.5A CN110930244B (en) 2019-11-27 2019-11-27 Method and device for calculating credit investigation evaluation value of user

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911185326.5A CN110930244B (en) 2019-11-27 2019-11-27 Method and device for calculating credit investigation evaluation value of user

Publications (2)

Publication Number Publication Date
CN110930244A true CN110930244A (en) 2020-03-27
CN110930244B CN110930244B (en) 2022-11-25

Family

ID=69846717

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911185326.5A Active CN110930244B (en) 2019-11-27 2019-11-27 Method and device for calculating credit investigation evaluation value of user

Country Status (1)

Country Link
CN (1) CN110930244B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117973903A (en) * 2024-03-29 2024-05-03 河北省农林科学院谷子研究所 Online intelligent evaluation management method and system for brand value

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104333530A (en) * 2013-07-22 2015-02-04 深圳市腾讯计算机系统有限公司 Information credibility verifying method and apparatus
CN104463603A (en) * 2014-12-05 2015-03-25 中国联合网络通信集团有限公司 Credit assessment method and system
CN107369034A (en) * 2017-06-14 2017-11-21 广东数相智能科技有限公司 A kind of user investigates the sincere method and apparatus judged
CN109492076A (en) * 2018-09-20 2019-03-19 西安交通大学 A kind of network-based community's question and answer website answer credible evaluation method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104333530A (en) * 2013-07-22 2015-02-04 深圳市腾讯计算机系统有限公司 Information credibility verifying method and apparatus
CN104463603A (en) * 2014-12-05 2015-03-25 中国联合网络通信集团有限公司 Credit assessment method and system
CN107369034A (en) * 2017-06-14 2017-11-21 广东数相智能科技有限公司 A kind of user investigates the sincere method and apparatus judged
CN109492076A (en) * 2018-09-20 2019-03-19 西安交通大学 A kind of network-based community's question and answer website answer credible evaluation method

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117973903A (en) * 2024-03-29 2024-05-03 河北省农林科学院谷子研究所 Online intelligent evaluation management method and system for brand value

Also Published As

Publication number Publication date
CN110930244B (en) 2022-11-25

Similar Documents

Publication Publication Date Title
Morrison et al. Consensus on validation of forensic voice comparison
US20190332676A1 (en) Human-to-human conversation analysis
US8893024B2 (en) User evaluation in a collaborative online forum
CN110458697A (en) Method and apparatus for assessing risk
JP3645901B2 (en) Test system and control method thereof
EP4097720A1 (en) Systems and methods for automatic candidate assessments in an asynchronous video setting
US20170069326A1 (en) System and method for validating natural language content using crowdsourced validation jobs
TW200837717A (en) Apparatus and method to reduce recognization errors through context relations among dialogue turns
US11450095B2 (en) Machine learning for video analysis and feedback
Bass et al. Utilizing multidimensional computer adaptive testing to mitigate burden with patient reported outcomes
JP2020086853A (en) Information processing system, program, and information processing method
Moses et al. Perceived service quality and user satisfaction in library environment
CN113035232A (en) Psychological state prediction system, method and device based on voice recognition
Fauzia et al. Implementation of chatbot on university website using RASA framework
Molder et al. What do extreme weather events say about climate change? Comparing politicization and climate policy in US wildfire and hurricane news coverage
CN111915224B (en) Teaching evaluation system
CN110930244B (en) Method and device for calculating credit investigation evaluation value of user
CN112669936A (en) Social network depression detection method based on texts and images
Kobrin et al. An investigation of the Fit of linear regression models to data from an SAT® validity study
He et al. A data-driven innovation model of big data digital learning and its empirical study
CN113919983A (en) Test question portrait method, device, electronic equipment and storage medium
CN111708889A (en) Score authentication service device, electronic score sheet device, and score authentication service system
CN111915227B (en) Teaching evaluation data analysis method
CN116563067B (en) Big data-based graduate crowd employment analysis method, system and medium
CN111915225B (en) Teaching evaluation data processing method

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
GR01 Patent grant
GR01 Patent grant