CN113807749B - Object scoring method and device - Google Patents

Object scoring method and device Download PDF

Info

Publication number
CN113807749B
CN113807749B CN202111372688.2A CN202111372688A CN113807749B CN 113807749 B CN113807749 B CN 113807749B CN 202111372688 A CN202111372688 A CN 202111372688A CN 113807749 B CN113807749 B CN 113807749B
Authority
CN
China
Prior art keywords
scoring
vector
sample
sample object
score
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.)
Active
Application number
CN202111372688.2A
Other languages
Chinese (zh)
Other versions
CN113807749A (en
Inventor
温嘉瑶
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 Jindi Technology Co Ltd
Original Assignee
Beijing Jindi 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 Jindi Technology Co Ltd filed Critical Beijing Jindi Technology Co Ltd
Priority to CN202111372688.2A priority Critical patent/CN113807749B/en
Publication of CN113807749A publication Critical patent/CN113807749A/en
Application granted granted Critical
Publication of CN113807749B publication Critical patent/CN113807749B/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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/10—Office automation; Time management
    • G06Q10/103—Workflow collaboration or project management
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/10—Office automation; Time management
    • G06Q10/105—Human resources
    • G06Q10/1053—Employment or hiring

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Operations Research (AREA)
  • Marketing (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Development Economics (AREA)
  • Educational Administration (AREA)
  • Game Theory and Decision Science (AREA)
  • Image Analysis (AREA)

Abstract

The embodiment of the invention provides an object scoring method and device, wherein the method comprises the following steps: determining spatial vectors of a number of sample objects; wherein the spatial vector of the sample object is determined from the data of the sample object; determining scoring information for the number of sample objects; determining a scoring direction vector according to the scoring information and the space vector of each sample object; and determining the score of the current object according to the score direction vector. The embodiment of the invention can improve the scoring efficiency and reduce the cost.

Description

Object scoring method and device
Technical Field
The invention relates to the technical field of computers, in particular to a method and a device for scoring an object.
Background
To measure the competitiveness of an object in a certain aspect, the object is usually scored based on data of multiple dimensions of the object. The object may be an enterprise, a school, or the like. For example, to measure the competitiveness of a business, the business is scored based on dimensional data such as registered capital, social security number, etc. of the business.
The existing method scores objects based on preset expert rules. For example, the expert determines a scoring rule according to the contribution degree of each dimension to the scale of the enterprise, and the enterprise score is calculated based on the scoring rule.
However, this method is labor-intensive. In addition, when the scoring items change, for example, the enterprise scale scoring changes into the enterprise recruitment scoring, the experts need to readjust the contribution degree and scoring rule of each dimension, and the process takes a long time.
Disclosure of Invention
The embodiment of the invention provides a method and a device for scoring an object, which can improve scoring efficiency and reduce cost.
In a first aspect of the embodiments of the present invention, there is provided an object scoring method, including: determining spatial vectors of a number of sample objects; wherein the spatial vector of the sample object is determined from the data of the sample object; determining scoring information for the number of sample objects; determining a scoring direction vector according to the scoring information and the space vector of each sample object; and determining the score of the current object according to the score direction vector.
Preferably, the determining the spatial vectors of the plurality of sample objects comprises: for each of the sample objects: selecting data of a first dimension from the multi-dimensional data of the sample object; stitching the data of the first dimension to obtain a spatial vector of the sample object composed of elements of the first dimension;
preferably, the determining the spatial vectors of the plurality of sample objects comprises: for each of the sample objects: obtaining data of a second dimension from the multi-dimensional data of the sample object; converting the data of the second dimension into a space vector of the sample object according to a preset transformation model; wherein the spatial vector of the sample object is composed of elements of a first dimension.
Preferably, the space vector according to the scoring information and each sample object includes: determining the scoring direction vector according to a preset argmax function, the scoring information and the space vector of each sample object; the scoring direction vector is a variable value when the target function takes the maximum value.
Preferably, the argmax function is
Figure 535395DEST_PATH_IMAGE001
Wherein
Figure 228544DEST_PATH_IMAGE002
Figure 493303DEST_PATH_IMAGE003
for characterizing the scoring direction vector, and for characterizing the scoring direction vector,
Figure 980917DEST_PATH_IMAGE004
m is used to characterize the number of spatial vectors of the sample object,
Figure 65547DEST_PATH_IMAGE005
a spatial vector for characterizing a sample object i, which has a score higher than the score of sample object i + 1.
Preferably, the argmax function is
Figure 562388DEST_PATH_IMAGE001
Wherein
Figure 212812DEST_PATH_IMAGE006
Figure 136906DEST_PATH_IMAGE003
for characterizing the scoring direction vector, and for characterizing the scoring direction vector,
Figure 239991DEST_PATH_IMAGE004
m is used to characterize the number of spatial vectors of the sample object,
Figure 274943DEST_PATH_IMAGE007
a spatial vector for characterizing the sample object a,
Figure 65961DEST_PATH_IMAGE008
a spatial vector for characterizing a sample object b, the score of sample object a being higher than the score of sample object b.
Preferably, the determining the scoring direction vector according to a preset argmax function, the scoring information and the spatial vector of each sample object includes: and calculating the grading direction vector through a gradient descent algorithm according to the argmax function, the grading information and the space vector of each sample object.
Preferably, the scoring information includes: an identification and a rank of the sample object; determining the scoring direction vector according to a preset argmax function, the scoring information and the space vector of each sample object, including: calculating to obtain a plurality of candidate direction vectors according to the space vectors of every two adjacent sample objects; calculating the value of the objective function corresponding to each candidate direction vector; and calculating the scoring direction vectors according to the values of the target functions corresponding to the candidate direction vectors.
Preferably, the method further comprises: determining a spatial vector of the current object; determining the score of the current object according to the score direction vector comprises: and determining the score of the current object as the projection of the space vector of the current object in the direction of the score direction vector.
Preferably, the determining the score of the current object according to the score direction vector includes: determining first scores of the plurality of objects according to the score direction vectors; wherein the current object is included in the plurality of objects; determining a maximum value and a minimum value in the first scores of the plurality of objects; and determining a second score of the current object according to the maximum value, the minimum value, the first score of the current object and a preset score interval.
In a second aspect of the embodiments of the present invention, there is provided an object scoring apparatus, including: a first determination module configured to determine spatial vectors of a number of sample objects; wherein the spatial vector of the sample object is determined from the data of the sample object; a second determination module configured to determine scoring information for the number of sample objects; a third determining module configured to determine a scoring direction vector according to the scoring information and the spatial vector of each sample object; and the scoring module is configured to determine the score of the current object according to the scoring direction vector.
In a third aspect of the embodiments of the present invention, a computer storage medium is provided, where a computer executable program is stored on the computer storage medium, and the computer executable program is executed to implement the object scoring method according to any embodiment of the present invention.
In a fourth aspect of the embodiments of the present invention, an electronic device is provided, where the electronic device includes a memory and a processor, the memory is used for storing a computer-executable program, and the processor is used for running the computer-executable program to implement the object scoring method according to any embodiment of the present invention.
According to the embodiment of the invention, the corresponding space vector is determined through the data of the sample object, the incidence relation between the data of the sample object and the scoring information is learned based on the space vector and the scoring information of the sample object, and the scoring direction vector is obtained, can represent the scoring rule, and can be used for scoring different objects. According to the method, no expert is needed to determine the scoring rule, and when the scoring item is changed, the scoring direction vector can be adjusted by adjusting the data of the sample object, so that the score corresponding to the scoring item is obtained. Compared with the existing method, the method can improve the scoring efficiency.
Additional features and advantages of the disclosure will be set forth in the detailed description which follows.
Drawings
Fig. 1 is a schematic view of an application scenario provided by an embodiment of the present invention;
fig. 2 is a schematic flow chart of an object scoring method according to an embodiment of the present invention;
FIG. 3 is a schematic flow chart of a business scoring method according to an embodiment of the present invention;
FIG. 4 is a flowchart illustrating a business scoring method according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an object scoring apparatus according to an embodiment of the present invention;
fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make those skilled in the art better understand the technical solutions in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments obtained by a person skilled in the art based on the embodiments of the present invention shall fall within the scope of the protection of the embodiments of the present invention.
The following further describes specific implementation of the embodiments of the present invention with reference to the drawings.
As shown in fig. 1, fig. 1 is a schematic view of an application scenario in the first embodiment of the present invention; the application scene is directed at an object scoring system, the object scoring system comprises a terminal device 101 and an object scoring server 102, an object scoring device is arranged on the object scoring server 102, the object scoring server 102 can be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, and can also be a cloud server for providing basic cloud computing services such as cloud service, a cloud database, cloud computing, cloud functions, cloud storage, network service, cloud communication, middleware service, domain name service, security service, CDN (content delivery network) and a big data and artificial intelligence platform. The terminal device 101 may be, but is not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, and the like. The terminal device 101 and the object scoring server may be directly or indirectly connected through a wireless communication manner (e.g., a network), and the present invention is not limited thereto.
The following embodiment describes a specific implementation process of the object scoring method in detail by taking an enterprise as an example.
The enterprise data has the characteristics of large data volume, multiple data dimensions, strong specialization and the like. In the existing method, an expert determines a scoring rule, and in the process, the expert needs to have strong specialty and can determine the contribution degree of each dimensional data to enterprise scoring and determine a final scoring rule according to the contribution degree. If the scoring process involves more data dimensions, then the expert will spend longer time determining the scoring rules. In addition, as the scoring items change, the data dimensions related to the scoring rules may change, and the existing method needs to re-determine the contribution degree of each dimension data to the enterprise scoring by experts. Therefore, the existing scoring process is long in time and high in cost.
In view of this, as shown in fig. 2, an embodiment of the present invention provides an object scoring method, including the following steps:
step 201: determining spatial vectors of a number of sample objects; wherein the spatial vector of the sample object is determined from the data of the sample object.
Taking the example where the object is an enterprise, the data of the enterprise includes but is not limited to: registration capital, actual payment capital, establishment time, time to market, state to market, social security, participation and stock control, branch office, financing condition, number of customers, number of suppliers, etc. The spatial vector for a sample enterprise may be determined by data for one or more dimensions of the sample enterprise, for example, by social security population, or by registered capital, affiliates, and time to market. One element of the spatial vector of the object may correspond to data of one dimension, or may be a fusion of data of multiple dimensions.
Step 202: scoring information for a number of sample objects is determined.
The scoring information may be the identity and rank of the sample object, the rank being determined by the score of the sample object. The scoring information may also be the identification and scoring of the sample object.
The scoring information may be various types of lists related to scoring items, for example, if the popularity of a business is scored, a "big-business popularity ranking" list may be used.
The scoring information may be ranking information obtained by analyzing website data, such as, for example, an intra-site enterprise click rate ranking.
The scoring information may be ranking information obtained by experts according to a scoring rule.
The scoring information can reflect the scoring condition of the sample object under the scoring item.
In an actual application scenario, the scoring information may be obtained first, then the sample object involved in the scoring information is determined, and then the spatial vector of the sample object is determined. Specifically, the spatial vectors of a plurality of objects may be predetermined and stored in a database. For each scoring item, a spatial vector of a sample object corresponding to the scoring item may be obtained from the database. For example, the database stores the space vectors of 1000 objects, and when the scale of the target object is evaluated, the space vectors of 200 objects are used as the space vectors of the sample object. When the development potential of the target object is evaluated, the spatial vectors of 100 objects employed therein are taken as the spatial vectors of the sample objects.
Step 203: and determining a scoring direction vector according to the scoring information and the space vector of each sample object.
The embodiment of the invention adopts the space vectors of a plurality of sample objects to determine the scoring direction vector, and enough sample amount can ensure the accuracy of the scoring direction vector. The scoring direction vector can characterize the association between the data of the sample object and the score, i.e. the scoring rule. When the sample object or the data dimension for determining the scoring direction vector changes, the scoring direction vector may change, that is, the scoring rule changes. If the scoring rules need to be adjusted, the data dimensions used to determine the spatial vectors of the sample objects may be adjusted, or the sample objects used may be adjusted.
Step 204: and determining the score of the current object according to the score direction vector.
The embodiment of the invention learns the incidence relation between the data of the sample object and the grading information based on the space vector and the grading information of the sample object to obtain the grading direction vector, and the grading direction vector can represent the grading rule and can be used for grading different objects. According to the method, no expert is needed to determine the scoring rule, and when the scoring item is changed, the scoring direction vector can be adjusted by adjusting the data of the sample object, so that the score corresponding to the scoring item is obtained. Compared with the existing method, the method can improve the scoring efficiency.
Further, the method further comprises: the plurality of objects is ranked according to their scores. The ranking of the objects in the scoring items can be further obtained through the scoring of the objects, and the competitiveness of each object on the scoring items can be determined more intuitively.
In one embodiment of the invention, determining a spatial vector of a number of sample objects comprises:
for each sample object: selecting data of a first dimension from multi-dimensional data of a sample object; and splicing the data of the first dimension to obtain a space vector of the sample object formed by the elements of the first dimension.
In the embodiment of the invention, the dimension of the space vector can be preset, the method obtains data of corresponding dimension according to the preset dimension of the space vector, and the space vector of the sample object is obtained by splicing. For example, if the preset first dimension is 5, 5 elements exist in the space vector constituting the sample object, each element corresponds to one-dimensional data of the sample object, and the space vector of the sample enterprise may be (registered capital, real-time capital, established time, time to market, social security). If the preset first dimension is 3, the space vector of the sample enterprise may be (registered capital, established time, social security).
Considering that a sample object generally has data with multiple dimensions, in an actual application scenario, a spatial vector dimension and a data dimension may also be preset. For example, the preset space vector dimension is 2, the preset data dimension comprises registered capital and social security number, and at this time, according to the preset information, the method obtains the numerical values of the registered capital and the social security number from the multidimensional data of the sample enterprise, and splices to obtain the space vector of the sample enterprise. For example, the registered capital is 20 thousands, the social security number is 10 persons, and the obtained space vector of the sample enterprise can be (20, 10) or (10, 20). The splicing sequence of the data of different dimensions of the sample object can be preset. According to the embodiment of the invention, the space vector of the sample object can be obtained through a splicing mode according to the set space vector dimension, the mode can rapidly convert the data of the sample object into the space vector of the sample object, and the object scoring efficiency is improved.
In one embodiment of the invention, determining a spatial vector of a number of sample objects comprises:
for each sample object: obtaining data of a second dimension from the multi-dimensional data of the sample object; converting the data of the second dimension into a space vector of the sample object according to a preset transformation model; wherein the spatial vector of the sample object is composed of elements of the first dimension.
In the embodiment of the present invention, data of a sample object of a second dimension, which is different from the set first dimension, that is, the dimension of the space vector of the sample object, may be randomly acquired. When the second dimension is different from the first dimension, the method may convert the data of the second dimension into a spatial vector of the sample object through a preset transformation model. Specifically, the transformation model may first stitch the second-dimensional data into a target vector, and then convert the target vector into a spatial vector of the sample object. Or, the data of the second dimension is spliced into a target vector in other ways, and then the target vector is converted into a space vector of the sample object through the transformation model. The transformation model may be a linear model, an exponential model, a logarithmic model, or the like.
In one embodiment of the present invention, determining a scoring direction vector according to the scoring information and the spatial vector of each sample object includes:
determining a scoring direction vector according to a preset argmax function, scoring information and the space vector of each sample object; the scoring direction vector is the variable value when the objective function takes the maximum value.
According to the scoring direction vector, the bit number of each sample object in the scoring sequence is obtained according to the scoring information, and the scoring direction vector is determined according to the argmax function, the bit number of each sample object and the space vector. In the embodiment of the present invention, the number of space vectors of sample objects is m, the dimension of the space vector of each sample object is n, and the space vector of sample object i is
Figure 160956DEST_PATH_IMAGE005
The higher the score of the sample object, the more advanced the rank, i.e. the
Figure 548075DEST_PATH_IMAGE009
。
Figure 386718DEST_PATH_IMAGE010
A score for characterizing the sample object i.
argmax is a function that parameterizes (or sets of) a function.
The argmax function adopted by the embodiment of the invention is
Figure 214996DEST_PATH_IMAGE001
，
Figure 480893DEST_PATH_IMAGE001
Is that make
Figure 558570DEST_PATH_IMAGE011
Obtaining the variable point corresponding to the maximum value
Figure 935325DEST_PATH_IMAGE003
(or
Figure 414848DEST_PATH_IMAGE003
A collection of). Thus, the scoring direction vector is
Figure 648383DEST_PATH_IMAGE011
The value of the variable or set of values of the variable at which the maximum is taken. It is an object of embodiments of the invention to determine
Figure 213356DEST_PATH_IMAGE003
The product is
Figure 393802DEST_PATH_IMAGE003
Make it
Figure 727831DEST_PATH_IMAGE011
Taking the maximum value.
Figure 69951DEST_PATH_IMAGE011
As shown in formula (1):
Figure 122221DEST_PATH_IMAGE012
（1）
wherein,
Figure 903095DEST_PATH_IMAGE003
for characterizing the scoring direction vector, and for characterizing the scoring direction vector,
Figure 91631DEST_PATH_IMAGE004
。
Figure 870231DEST_PATH_IMAGE011
the following forms may also be adopted:
Figure 406867DEST_PATH_IMAGE013
（2）
wherein,
Figure 929115DEST_PATH_IMAGE007
a spatial vector for characterizing the sample object a,
Figure 972157DEST_PATH_IMAGE008
a spatial vector for characterizing the sample object b. The sample object a has a higher score than the sample object b, the rank of sample object a precedes the rank of sample object b.
According to the embodiment of the invention, through the argmax function, the method is small in calculation amount, and the scoring direction vector can be determined efficiently. It should be noted that, in an actual application scenario, other manners such as argmin function may also be used to implement, for example
Figure 921659DEST_PATH_IMAGE014
。
In solving argmax function
Figure 948521DEST_PATH_IMAGE003
There are various methods of which two solutions will be described below by means of two examples.
In an embodiment of the present invention, determining a scoring direction vector according to a preset argmax function, scoring information, and a spatial vector of each sample object includes:
and calculating a grading direction vector through a gradient descent algorithm according to the argmax function, the grading information and the space vector of each sample object.
According to the embodiment of the invention, the bit number of each sample object in the grading sequence is obtained according to the grading information, and the grading direction vector is calculated through a gradient descent algorithm according to the argmax function, the space vector and the bit number of the sample object.
The embodiment of the invention realizes the gradient descent method through the preset loss function, and the loss function is shown as a formula (3).
Figure 805618DEST_PATH_IMAGE015
（3）
Wherein,
Figure 968746DEST_PATH_IMAGE016
for characterizing the constraint parameters.
Step size of gradient descent algorithm adopted by embodiment of the invention
Figure 26832DEST_PATH_IMAGE017
Distance of termination
Figure 540990DEST_PATH_IMAGE018
，
Figure 405041DEST_PATH_IMAGE016
=1。
The gradient descent algorithm is implemented as follows:
a1: initialized by calculation of equation (4)
Figure 688255DEST_PATH_IMAGE019
。
Figure 979559DEST_PATH_IMAGE020
（4）
A2: from equations (5) and (6), a loss function gradient is calculated.
Figure 715434DEST_PATH_IMAGE021
（5）
Figure 586438DEST_PATH_IMAGE022
（6）
A3: updating
Figure 721228DEST_PATH_IMAGE023
；
Figure 183434DEST_PATH_IMAGE024
。
A4: repeating A2 and A3 until
Figure 672184DEST_PATH_IMAGE025
Are all less than
Figure 878037DEST_PATH_IMAGE026
Obtained at this time
Figure 870264DEST_PATH_IMAGE003
Is a scoring direction vector.
In one embodiment of the invention, the scoring information comprises: the identity and rank of the sample object;
determining a scoring direction vector according to a preset argmax function, scoring information and a space vector of each sample object, wherein the scoring direction vector comprises the following steps:
b1: and calculating to obtain a plurality of candidate direction vectors according to the space vectors of every two adjacent sample objects.
The candidate direction vector is calculated using equation (7).
Figure 768950DEST_PATH_IMAGE027
（7）
Wherein,
Figure 479417DEST_PATH_IMAGE028
for characterizing the candidate direction vector i, in an embodiment of the invention,
Figure 223382DEST_PATH_IMAGE029
。
b2: and calculating the value of the target function corresponding to each candidate direction vector.
Calculated according to equations (1) and (7)
Figure 335694DEST_PATH_IMAGE030
。
B3: and calculating the scoring direction vectors according to the values of the target functions corresponding to the candidate direction vectors.
Go through
Figure 139702DEST_PATH_IMAGE031
Taking the maximum first k of B2
Figure 540728DEST_PATH_IMAGE030
Calculate the k
Figure 885122DEST_PATH_IMAGE030
Correspond to
Figure 586361DEST_PATH_IMAGE032
Mean value of
Figure 826850DEST_PATH_IMAGE033
Calculating
Figure 511909DEST_PATH_IMAGE034
。
For example, B2 can give
Figure 863256DEST_PATH_IMAGE035
、
Figure 681652DEST_PATH_IMAGE036
、
Figure 968408DEST_PATH_IMAGE037
……
Figure 78446DEST_PATH_IMAGE038
。
When k =1, determining
Figure 967905DEST_PATH_IMAGE035
~
Figure 909316DEST_PATH_IMAGE038
Middle first value, assuming
Figure 491607DEST_PATH_IMAGE037
When the value of (1) is maximum, then
Figure 885679DEST_PATH_IMAGE039
=
Figure 578829DEST_PATH_IMAGE040
，
Figure 109167DEST_PATH_IMAGE041
=
Figure 862360DEST_PATH_IMAGE037
；
k =2, determining
Figure 540466DEST_PATH_IMAGE035
~
Figure 37306DEST_PATH_IMAGE038
Middle rank first two values, suppose
Figure 684801DEST_PATH_IMAGE042
When the value of (1) is maximum, then
Figure 343315DEST_PATH_IMAGE043
=
Figure 711979DEST_PATH_IMAGE044
，
Figure 746932DEST_PATH_IMAGE045
=
Figure 517441DEST_PATH_IMAGE046
；
k =3, determining
Figure 550120DEST_PATH_IMAGE035
~
Figure 671659DEST_PATH_IMAGE038
Middle first three values, assuming
Figure 244723DEST_PATH_IMAGE047
When the value of (1) is maximum, then
Figure 604160DEST_PATH_IMAGE048
=
Figure 870057DEST_PATH_IMAGE049
，
Figure 947734DEST_PATH_IMAGE050
=
Figure 590068DEST_PATH_IMAGE051
；
……
By analogy, obtain
Figure 69591DEST_PATH_IMAGE041
、
Figure 506388DEST_PATH_IMAGE045
、
Figure 868100DEST_PATH_IMAGE050
……
Figure 48545DEST_PATH_IMAGE052
。
Determining
Figure 379645DEST_PATH_IMAGE041
~
Figure 987344DEST_PATH_IMAGE052
Maximum value of (1), maximum value corresponds to
Figure 774034DEST_PATH_IMAGE053
Is a scoring direction vector.
In one embodiment of the invention, the method further comprises: determining a spatial vector of a current object;
determining the score of the current object according to the score direction vector, wherein the step of determining the score of the current object comprises the following steps:
and determining the score of the current object as the projection of the space vector of the current object in the direction of the score direction vector.
The method for determining the space vector of the current object is the same as the method for determining the space vector of the sample object, and the description thereof is omitted here.
Figure 492591DEST_PATH_IMAGE054
（8）
And (4) calculating the projection of the space vector of the current object in the scoring direction vector direction according to the formula (8) to obtain the space vector of the current object.
The embodiment of the invention realizes the scoring calculation through vector projection, considers the contribution of different dimensional data to the scoring and ensures the consistency with the scoring rule of the sample object. Particularly, when the object comprises data of multiple dimensions, the calculation process can be accelerated through operation among vectors, and the scoring efficiency is improved.
In an actual application scenario, the score of the current object may also be calculated in other manners, for example, a dot product of a space vector and a score direction vector of the current object is used as the score of the current object.
In one embodiment of the present invention, determining the score of the current object according to the score direction vector comprises:
determining first scores of the plurality of objects according to the scoring direction vectors; wherein the plurality of objects comprise a current object;
determining a maximum value and a minimum value in the first scores of the plurality of objects;
and determining a second score of the current object according to the maximum value, the minimum value, the first score of the current object and a preset score interval.
In the embodiment of the invention, a plurality of current objects can be scored, and in order to more intuitively measure the score of each current object, the score of the current object can be mapped to a specified scoring interval.
Specifically, the second score of the current object may be determined according to equation (9).
Figure 946706DEST_PATH_IMAGE055
（9）
Wherein,
Figure 725307DEST_PATH_IMAGE056
a first score for characterizing the object i,
Figure 530452DEST_PATH_IMAGE057
a second score for characterizing the object i,
Figure 52700DEST_PATH_IMAGE058
for characterizing the minimum value in the first score,
Figure 361321DEST_PATH_IMAGE059
for characterizing the maximum value in the first score, p for characterizing the left end point value of the score interval, and q for characterizing the right end point value of the score interval. Wherein f (x) is from [0,1 ]]Interval mapping to [0,1]The monotonically increasing function of the interval may be a linear or non-linear function.
As shown in fig. 3, an embodiment of the present invention provides an enterprise scoring method, including:
step 301: for each sample enterprise: data of a first dimension is selected from the multidimensional data of the sample enterprise.
The embodiment of the invention aims to score the enterprise scale. Acquiring data of 100 sample enterprises, wherein each sample enterprise comprises 50-dimensional data, and step 301 selects 5-dimensional data related to the enterprise scale from the 50-dimensional data according to preset dimensions, such as registered capital, real-payment capital, social security number, customer number and branch number.
Step 302: and splicing the data of the first dimension to obtain a space vector of the sample enterprise formed by the elements of the first dimension.
And sequentially splicing the registration capital, the real payment capital, the social security number, the number of clients and the number of branches to obtain the space vector of the sample enterprise.
Step 303: scoring information for a number of sample businesses is determined.
And acquiring a list of 'large-scale ranking of hundred enterprises', and determining the ranking of 100 sample enterprises in the list.
Step 304: and calculating a grading direction vector through a gradient descent algorithm according to the argmax function, the grading information and the space vectors of all the sample enterprises.
Wherein, the scoring direction vector is a variable value when the target function takes the maximum value.
Step 305: a spatial vector of the current enterprise is determined.
And acquiring data of the current registered capital, the actual payment capital, the social security number, the number of clients and the number of branches of the enterprise, and sequentially splicing to obtain the space vector of the current enterprise.
Step 306: and determining the score of the current enterprise as the projection of the space vector of the current enterprise in the direction of the scoring direction vector.
The score of the current enterprise can measure the competitiveness of the current enterprise on the enterprise scale, and the scale ranking of each enterprise can be distinguished by scoring the enterprise on the scale.
In a general case, if enterprise scale ranking is to be calculated, experts are required to evaluate the contribution degree of each dimension data, such as registered capital, social security number and the like, of the enterprise data dimension to the scale, and then a scoring rule is formulated for calculation. In the embodiment of the invention, only one enterprise scale ranking sample is required to be given, such as enterprise g, enterprise h, enterprise j, g > h > j, and the scheme can automatically learn the scoring direction vector so as to calculate the score. In addition, when new scoring items are met, such as computing enterprise recruitment competitiveness ranking, the original method needs experts to evaluate the data dimension again, but the scoring direction vector can be quickly determined again only by giving a recruitment competitiveness ranking list.
As shown in fig. 4, an embodiment of the present invention provides an enterprise scoring method, including:
step 401: for each sample enterprise: and acquiring data of a second dimension from the multidimensional data of the sample enterprise.
The embodiment of the invention aims to score the recruitment competitiveness of an enterprise. Data was obtained for 50 sample enterprises, each of which included 20-dimensional data.
Step 402: and converting the data of the second dimension into a space vector of the sample enterprise according to a preset transformation model.
According to the preset dimensionality, 8-dimensional data is selected from the 20-dimensional data, for example, the selected 8-dimensional data is registered capital, real payment capital, establishment time, time to market, social security number, branch number, customer number and supplier number.
According to the transformation model, 8-dimensional data is converted into a space vector of a 4-dimensional sample enterprise. For example, 4 elements of the space vector of the sample enterprise are obtained by calculating the registered capital + the real payment capital, the established time + the time to market, the social security number + the number of branches, the number of customers + the number of suppliers through a transformation model.
Step 403: determining grading information of a plurality of sample enterprises; the scoring information comprises the identification and the rank of the sample enterprise.
Click-through rate rankings for a number of sample businesses within a specified website are determined.
Step 404: and calculating to obtain a plurality of candidate direction vectors according to the space vectors of every two adjacent sample enterprises.
Step 405: and calculating the value of the target function corresponding to each candidate direction vector.
Step 406: and calculating the scoring direction vectors according to the values of the target functions corresponding to the candidate direction vectors.
Step 407: spatial vectors for a number of businesses are determined.
Step 408: determining a first score of each enterprise according to the scoring direction vector; the first score of the enterprise is the projection of the space vector of the enterprise in the direction of the score direction vector.
Step 409: a maximum and a minimum of the first scores for the plurality of businesses are determined.
Step 410: and determining a second grade of the enterprise according to the maximum value, the minimum value, the first grade of the enterprise and a preset grade interval.
According to the embodiment of the invention, the scoring direction vector is determined through the multidimensional data and the scoring information of the sample enterprise, and the scoring direction vector can represent the scoring rule of the recruitment competitiveness of the enterprise. And scoring the recruitment competitiveness of the enterprise based on the scoring direction vector, and mapping the score to a designated scoring interval.
As shown in fig. 5, an embodiment of the present invention provides an object scoring apparatus, including:
a first determining module 501 configured to determine spatial vectors of a number of sample objects; wherein the spatial vector of the sample object is determined from the data of the sample object;
a second determining module 502 configured to determine scoring information for the number of sample objects;
a third determining module 503, configured to determine a scoring direction vector according to the scoring information and the spatial vectors of the respective sample objects;
a scoring module 504 configured to determine a score of the current object according to the scoring direction vector.
In one embodiment of the invention, the first determining module 501 is configured to, for each sample object: selecting data of a first dimension from multi-dimensional data of a sample object; and splicing the data of the first dimension to obtain a space vector of the sample object formed by the elements of the first dimension.
In one embodiment of the invention, the first determining module 501 is configured to, for each sample object: obtaining data of a second dimension from the multi-dimensional data of the sample object; converting the data of the second dimension into a space vector of the sample object according to a preset transformation model; wherein the spatial vector of the sample object is composed of elements of the first dimension.
In an embodiment of the present invention, the third determining module 503 is configured to determine a scoring direction vector according to a preset argmax function, scoring information, and a spatial vector of each sample object; the scoring direction vector is the variable value when the objective function takes the maximum value.
In one embodiment of the invention, the argmax function is
Figure 310823DEST_PATH_IMAGE001
Wherein
Figure 806526DEST_PATH_IMAGE060
Figure 132465DEST_PATH_IMAGE003
for characterizing the scoring direction vector, and for characterizing the scoring direction vector,
Figure 295593DEST_PATH_IMAGE004
m is used to characterize the number of spatial vectors of the sample object,
Figure 212734DEST_PATH_IMAGE005
a spatial vector for characterizing a sample object i, which has a score higher than the score of sample object i + 1.
In one embodiment of the invention, the argmax function is
Figure 726892DEST_PATH_IMAGE001
Wherein
Figure 325363DEST_PATH_IMAGE006
Figure 342998DEST_PATH_IMAGE003
for characterizing the scoring direction vector, and for characterizing the scoring direction vector,
Figure 696619DEST_PATH_IMAGE004
m is used to characterize the number of spatial vectors of the sample object,
Figure 429564DEST_PATH_IMAGE007
a spatial vector for characterizing the sample object a,
Figure 97306DEST_PATH_IMAGE008
a spatial vector for characterizing the sample object b. The score of sample object a is higher than the score of sample object b.
In an embodiment of the invention, the third determining module 503 is configured to calculate the scoring direction vector by a gradient descent algorithm according to the argmax function, the scoring information and the spatial vectors of the respective sample objects.
In one embodiment of the invention, the scoring information comprises: the identity and rank of the sample object; a third determining module 503, configured to calculate a plurality of candidate direction vectors according to the space vectors of every two bit-wise neighboring sample objects; calculating the value of the target function corresponding to each candidate direction vector; and calculating the scoring direction vectors according to the values of the target functions corresponding to the candidate direction vectors.
In one embodiment of the present invention, the first determining module 501 is configured to determine a spatial vector of a current object;
a third determining module 503 configured to determine the score of the current object as a projection of the spatial vector of the current object in the direction of the score direction vector.
In one embodiment of the invention, the scoring module 504 is configured to determine a first score of the plurality of objects according to the scoring direction vector; wherein the plurality of objects comprise a current object; determining a maximum value and a minimum value in the first scores of the plurality of objects; and determining a second score of the current object according to the maximum value, the minimum value, the first score of the current object and a preset score interval.
The embodiment of the invention also provides a computer storage medium, wherein a computer executable program is stored on the computer storage medium, and the computer executable program is operated to implement the object scoring method in any embodiment of the invention.
As shown in fig. 6, fig. 6 is a schematic structural diagram of an electronic device according to a sixth embodiment of the present invention; the electronic device comprises a memory 601 and a processor 602, wherein the memory is used for storing a computer executable program, and the processor is used for running the computer executable program to implement the object scoring method according to any embodiment of the invention.
The object scoring server may be an electronic device shown in fig. 6.
The above-mentioned embodiments are only specific embodiments of the present invention, which are used for illustrating the technical solutions of the present invention and not for limiting the same, and the protection scope of the present invention is not limited thereto, although the present invention is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present invention, and they should be construed as being included therein. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (9)

1. An object scoring method, comprising:
determining spatial vectors of a number of sample objects; wherein the spatial vector of the sample object is determined by the property data of the sample object; one element of the space vector of the sample object corresponds to attribute data of one dimension or the fusion of the attribute data of a plurality of dimensions;
determining scoring information for the number of sample objects;
determining a scoring direction vector according to the scoring information and the space vector of each sample object, wherein the scoring direction vector can represent the incidence relation between the attribute data of the sample object and the score;
determining the score of the current object according to the score direction vector;
determining the scoring direction vector according to the scoring information and the spatial vector of each sample object, including:
determining the scoring direction vector according to a preset argmax function, the scoring information and the space vector of each sample object; or, determining the scoring direction vector according to a preset argmin function, the scoring information and the space vector of each sample object;
the argmax function is
Figure 707137DEST_PATH_IMAGE001
The argmin function is
Figure 203978DEST_PATH_IMAGE002
；
Wherein
Figure 854402DEST_PATH_IMAGE003
Figure 512916DEST_PATH_IMAGE004
For characterizing the scoring direction vector, and for characterizing the scoring direction vector,
Figure 147160DEST_PATH_IMAGE005
m is used to characterize the number of spatial vectors of the sample object,
Figure 182112DEST_PATH_IMAGE006
a spatial vector for characterizing a sample object i, the sample object i having a score higher than the score of the sample object i +1,
or;
Figure 943833DEST_PATH_IMAGE007
Figure 773249DEST_PATH_IMAGE004
for characterizing the scoring direction vector, and for characterizing the scoring direction vector,
Figure 629209DEST_PATH_IMAGE005
m is the number of spatial vectors characterizing the sample object, m is the number of spatial vectors characterizing the sample object a,
Figure 733432DEST_PATH_IMAGE008
a spatial vector for characterizing a sample object b, the score of sample object a being higher than the score of sample object b.
2. The method of claim 1,
the determining spatial vectors for a number of sample objects comprises:
for each of the sample objects:
selecting attribute data of a first dimension from the multi-dimensional attribute data of the sample object;
splicing the attribute data of the first dimension to obtain a space vector of the sample object composed of the elements of the first dimension;
and/or the presence of a gas in the gas,
the determining spatial vectors for a number of sample objects comprises:
for each of the sample objects:
acquiring attribute data of a second dimension from the multi-dimensional attribute data of the sample object;
converting the attribute data of the second dimension into a space vector of the sample object according to a preset transformation model; wherein the spatial vector of the sample object is composed of elements of a first dimension.
3. The method of claim 1,
determining the scoring direction vector according to a preset argmax function, the scoring information and the space vector of each sample object, including:
and calculating the grading direction vector through a gradient descent algorithm according to the argmax function, the grading information and the space vector of each sample object.
4. The method of claim 1,
the scoring information comprises: an identification and a rank of the sample object;
determining the scoring direction vector according to a preset argmax function, the scoring information and the space vector of each sample object, including:
calculating to obtain a plurality of candidate direction vectors according to the space vectors of every two adjacent sample objects;
calculating the value of an objective function corresponding to each candidate direction vector;
and calculating the scoring direction vectors according to the values of the target functions corresponding to the candidate direction vectors.
5. The method of any of claims 1-4, further comprising:
determining a spatial vector of the current object;
determining the score of the current object according to the score direction vector comprises:
and determining the score of the current object as the projection of the space vector of the current object in the direction of the score direction vector.
6. The method according to any one of claims 1 to 4,
determining the score of the current object according to the score direction vector comprises:
determining first scores of the plurality of objects according to the score direction vectors; wherein the current object is included in the plurality of objects;
determining a maximum value and a minimum value in the first scores of the plurality of objects;
and determining a second score of the current object according to the maximum value, the minimum value, the first score of the current object and a preset score interval.
7. An object scoring device, comprising:
a first determination module configured to determine spatial vectors of a number of sample objects; wherein the spatial vector of the sample object is determined by the property data of the sample object; one element of the space vector of the sample object corresponds to attribute data of one dimension or the fusion of the attribute data of a plurality of dimensions;
a second determination module configured to determine scoring information for the number of sample objects;
a third determining module, configured to determine a scoring direction vector according to the scoring information and the spatial vector of each sample object, where the scoring direction vector can represent an association relationship between attribute data of the sample object and a score;
the scoring module is configured to determine the score of the current object according to the scoring direction vector;
determining the scoring direction vector according to the scoring information and the spatial vector of each sample object, including:
determining the scoring direction vector according to a preset argmax function, the scoring information and the space vector of each sample object; or, determining the scoring direction vector according to a preset argmin function, the scoring information and the space vector of each sample object;
the argmax function is, the argmin function is
Figure 92869DEST_PATH_IMAGE002
；
Wherein
Figure 624344DEST_PATH_IMAGE003
Figure 702022DEST_PATH_IMAGE004
For characterizing the direction of scoringThe amount of the compound (A) is,
Figure 344356DEST_PATH_IMAGE005
m is used to characterize the number of spatial vectors of the sample object,
Figure 89458DEST_PATH_IMAGE006
a spatial vector for characterizing a sample object i, the sample object i having a score higher than the score of the sample object i +1,
or;
Figure 526255DEST_PATH_IMAGE007
Figure 356808DEST_PATH_IMAGE004
for characterizing the scoring direction vector, and for characterizing the scoring direction vector,
Figure 537254DEST_PATH_IMAGE005
m is used to characterize the number of spatial vectors of the sample object,
Figure 871283DEST_PATH_IMAGE009
a spatial vector for characterizing the sample object a,
Figure 478982DEST_PATH_IMAGE008
a spatial vector for characterizing a sample object b, the score of sample object a being higher than the score of sample object b.
8. A computer storage medium having stored thereon a computer-executable program that is executed to implement the object scoring method of any one of claims 1-6.
9. An electronic device, comprising a memory for storing a computer-executable program and a processor for executing the computer-executable program to implement the object scoring method of any one of claims 1-6.
CN202111372688.2A 2021-11-19 2021-11-19 Object scoring method and device Active CN113807749B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111372688.2A CN113807749B (en) 2021-11-19 2021-11-19 Object scoring method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111372688.2A CN113807749B (en) 2021-11-19 2021-11-19 Object scoring method and device

Publications (2)

Publication Number Publication Date
CN113807749A CN113807749A (en) 2021-12-17
CN113807749B true CN113807749B (en) 2022-02-15

Family

ID=78938464

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111372688.2A Active CN113807749B (en) 2021-11-19 2021-11-19 Object scoring method and device

Country Status (1)

Country Link
CN (1) CN113807749B (en)

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7324110B2 (en) * 2004-12-09 2008-01-29 Image Metrics Limited Method and system for cleaning motion capture data
CN108733778B (en) * 2018-05-04 2022-05-17 百度在线网络技术（北京）有限公司 Industry type identification method and device of object

Also Published As

Publication number Publication date
CN113807749A (en) 2021-12-17

Similar Documents

Publication Publication Date Title
CN110321422B (en) Method for training model on line, pushing method, device and equipment
CN110008397B (en) Recommendation model training method and device
CN112395487B (en) Information recommendation method and device, computer readable storage medium and electronic equipment
CN112231592B (en) Graph-based network community discovery method, device, equipment and storage medium
CN109471978B (en) Electronic resource recommendation method and device
CN110879864A (en) Context recommendation method based on graph neural network and attention mechanism
CN113742492B (en) Insurance scheme generation method and device, electronic equipment and storage medium
CN108717407A (en) Entity vector determines method and device, information retrieval method and device
CN113656699B (en) User feature vector determining method, related equipment and medium
CN111429161B (en) Feature extraction method, feature extraction device, storage medium and electronic equipment
CN113657087B (en) Information matching method and device
KR101605654B1 (en) Method and apparatus for estimating multiple ranking using pairwise comparisons
CN109241442A (en) Item recommendation method, readable storage medium storing program for executing and terminal based on predicted value filling
CN110457387B (en) Method and related device applied to user tag determination in network
CN105045827A (en) Familiarity based information recommendation method and apparatus
CN112559640A (en) Training method and device of atlas characterization system
CN110262906B (en) Interface label recommendation method and device, storage medium and electronic equipment
CN113807749B (en) Object scoring method and device
CN110992127A (en) Article recommendation method and device
CN113591986B (en) Method for generating object weight of recommendation model and personalized recommendation method
CN111291196B (en) Knowledge graph perfecting method and device, and data processing method and device
KR102323424B1 (en) Rating Prediction Method for Recommendation Algorithm Based on Observed Ratings and Similarity Graphs
CN113094584A (en) Method and device for determining recommended learning resources
CN111860870A (en) Training method, device, equipment and medium for interactive behavior determination model
CN117076962B (en) Data analysis method, device and equipment applied to artificial intelligence field

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