CN113033931B

CN113033931B - Closed-loop self-adaptive individual and region allocation method and device and computing equipment

Info

Publication number: CN113033931B
Application number: CN201911349557.5A
Authority: CN
Inventors: 杨雨青; 郑欢; 陈勇; 余侃; 高琴; 傅泉辉; 许林甲
Original assignee: China Mobile Communications Group Co Ltd; China Mobile Group Zhejiang Co Ltd
Current assignee: China Mobile Communications Group Co Ltd; China Mobile Group Zhejiang Co Ltd
Priority date: 2019-12-24
Filing date: 2019-12-24
Publication date: 2023-12-29
Anticipated expiration: 2039-12-24
Also published as: CN113033931A

Abstract

The embodiment of the invention relates to the technical field of instant messaging, and discloses a closed-loop self-adaptive individual and region allocation method, a device and computing equipment, wherein the method comprises the following steps: applying a hierarchical attribution model to perform hierarchical partition processing on the user according to the instant user distribution data; performing region division on the core points according to a pre-stored region division strategy and a user region attribution data application region automatic division model of region processing; performing closed-loop verification on the hierarchical attribution model and the automatic regional division model according to service system data, and optimizing parameters; and outputting the user partition attribution data and the core point attribution data according to the optimized parameters. By means of the method, the embodiment of the invention supports multi-level scattered user summarization and diversified partition strategy control, can adapt to more complex partitioning conditions, can realize partition result verification closed-loop control, and continuously optimizes region allocation.

Description

Closed-loop self-adaptive individual and region allocation method and device and computing equipment

Technical Field

The embodiment of the invention relates to the technical field of instant messaging, in particular to a closed-loop self-adaptive individual and region allocation method, a device and computing equipment.

Background

The reasonable division of management areas and the attribution of matching service objects are the basis of intelligent management work. At present, an intelligent technical scheme is not available in the aspect of regional division, and the regional division mode is often determined based on administrative division and subjective judgment. The only area division algorithm, such as the Thiessen polygon (Voronoi) algorithm, has universality, but the algorithm relies on homogenized single data, and the application range is limited to the case of relatively uniform area distribution.

At present, the division of the areas is mainly realized by two methods: the method is simple and extensive, but lacks data support, and cannot realize fine management. One is a standard Voronoi algorithm that divides regions by a unique dependence on distance to core points, which cannot support more complex division conditions. Meanwhile, the algorithm can work normally only under the condition that the core points are distributed relatively uniformly.

Disclosure of Invention

In view of the foregoing, embodiments of the present invention provide a closed-loop adaptive individual and region allocation method, apparatus, and computing device that overcome or at least partially solve the foregoing problems.

According to an aspect of an embodiment of the present invention, there is provided a closed-loop adaptive individual and region allocation method, the method including: applying a hierarchical attribution model to perform hierarchical partition processing on the user according to the instant user distribution data; according to a pre-stored regional division strategy and a regional division model applied to the regional attribution data of the user in the regional division process, carrying out regional division on core points in a core point database; performing closed-loop verification on the hierarchical attribution model and the automatic regional division model according to service system data, and optimizing parameters of the hierarchical attribution model and the automatic regional division model; and outputting user partition attribution data and core point attribution data according to the optimized hierarchical attribution model and the parameters of the area automatic division model.

In an optional manner, the applying a hierarchical attribution model to the users according to the instant user distribution data performs hierarchical partitioning processing, including: and all users of the previous-level partition are attributed to the current-level partition according to the user resident data application level attribution model in the instant user distribution data.

In an optional manner, all users of the previous-level partition are attributed to the current-level partition according to the user resident data application-level attribution model in the instant user distribution data, including: calculating the partition gravity center according to the user resident data in the instant user distribution data and the current level partition list; calculating the Cartesian distance between the user grid and the center of gravity of the subarea according to the user resident data to form a Cartesian distance vector of the user; calculating a weighted Manhattan distance vector of the user grid and the partition gravity center according to the user resident data and the road network data; calculating a weighted double-distance vector of the user grid and the partition gravity center according to the user Cartesian distance vector and the weighted Manhattan distance vector; and extracting the weighted double-distance vectors of the user grid where the user resides and all the partition barycenters, and taking the partition corresponding to the partition barycenter with the shortest weighted double distance as the home partition of the user.

In an optional manner, the performing area division on the core points in the core point database according to a pre-stored area division policy and an automatic area division model of user partition attribution data of partition processing includes: calculating a weighted double-distance vector of the core point and a hierarchical region by applying the hierarchical attribution model according to the core point; establishing a to-be-allocated set of a whole to-be-allocated hierarchical area of a designated hierarchy of a policy center; a core point bottom-guaranteeing distribution model is applied to obtain a bottom-guaranteeing combination with the minimum number of weighted people of each core point in the set to be distributed according to the core point grade; acquiring a attribution combination with the minimum weighted number of each core point in the remaining hierarchical areas to be distributed of the set to be distributed according to the core point grade by applying a core point attribution distribution model; and merging the bottom protection combination and the attribution combination of each core point to form the merging attribution combination of the core points.

In an optional manner, the obtaining, by the application of the core point bottom-guard distribution model according to the core point level, a bottom-guard combination with the smallest weighted number of people for each core point in the to-be-distributed set includes: sequentially extracting the core points from the to-be-allocated set, and constructing all qualified bottom-guaranteeing combinations of each core point, so that the sum of the attribution groups of each qualified bottom-guaranteeing combination is just larger than the requirements of the number of bottom-guaranteeing people; calculating the weighted number of people of each qualified bottom-guard combination of the core points for each core point; and arranging the core points from high to low according to the grade, and sequentially taking the qualified bottom-guard combination with the smallest number of the weighted persons of each core point as the bottom-guard combination of each core point.

In an optional manner, the applying a core point attribution allocation model to obtain, according to a core point level, an attribution combination with a minimum number of weighted persons of each core point in remaining hierarchical areas to be allocated in the to-be-allocated set includes: sequentially extracting the core points from the rest hierarchical areas to be allocated of the set to be allocated, and constructing all qualified attribution combinations of each core point, so that the sum of attribution groups of each qualified attribution combination is just larger than the requirement of the attribution number of the core point at the current level; calculating a weighted population of each of the qualified attribution combinations of the core points for each of the core points; and arranging the core points from high to low according to the grade, and sequentially taking the qualified attribution combination with the smallest number of the weighted persons of each core point as the attribution combination of each core point.

In an optional manner, the performing closed-loop verification on the hierarchical attribution model and the automatic area division model according to service system data, optimizing parameters of the hierarchical attribution model and the automatic area division model, includes: respectively acquiring corresponding parameter sequences according to the parameters of the hierarchical attribution model and the automatic regional division model; respectively calculating corresponding deviation functions according to the parameter sequences of the hierarchical attribution model and the regional automatic division model; respectively introducing optimization step parameters according to the deviation functions to calculate corresponding deviation gradient functions; and calculating a new parameter sequence and a new deviation function according to the deviation gradient function and the optimization parameter step length until the new deviation function meets a convergence condition.

According to another aspect of an embodiment of the present invention, there is provided a closed loop adaptive individual and region allocation apparatus, the apparatus comprising: the hierarchical partitioning unit is used for performing hierarchical partitioning processing on the user by applying a hierarchical attribution model according to the instant user distribution data; the regional division unit is used for carrying out regional division on core points in the core point database according to a pre-stored regional division strategy and an automatic regional division model of user regional attribution data of regional processing; the parameter verification unit is used for performing closed-loop verification on the hierarchical attribution model and the area automatic division model according to the service system data, and optimizing parameters of the hierarchical attribution model and the area automatic division model; and the data output unit is used for outputting the user partition attribution data and the core point attribution data according to the optimized hierarchical attribution model and the parameters of the area automatic division model.

According to another aspect of an embodiment of the present invention, there is provided a computing device including: the device comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete communication with each other through the communication bus;

The memory is used for storing at least one executable instruction which causes the processor to execute the steps of the closed loop adaptive individual and region allocation method.

According to yet another aspect of the embodiments of the present invention, there is provided a computer storage medium having stored therein at least one executable instruction for causing the processor to perform the steps of the above-described closed-loop adaptive individual and region allocation method.

According to the embodiment of the invention, the hierarchical partitioning treatment is carried out on the user by applying a hierarchical attribution model according to the instant user distribution data; according to a pre-stored regional division strategy and a regional division model applied to the regional attribution data of the user in the regional division process, carrying out regional division on core points in a core point database; performing closed-loop verification on the hierarchical attribution model and the automatic regional division model according to service system data, and optimizing parameters of the hierarchical attribution model and the automatic regional division model; and outputting user partition attribution data and core point attribution data according to the optimized parameters of the hierarchical attribution model and the region automatic partition model, supporting multi-hierarchical scattered user summarization and diversified partition strategy control, adapting to more complex partition conditions, realizing verification closed-loop control of partition results, and continuously optimizing region allocation.

The foregoing description is only an overview of the technical solutions of the embodiments of the present invention, and may be implemented according to the content of the specification, so that the technical means of the embodiments of the present invention can be more clearly understood, and the following specific embodiments of the present invention are given for clarity and understanding.

Drawings

Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to designate like parts throughout the figures. In the drawings:

fig. 1 is a schematic flow chart of a closed-loop adaptive individual and region allocation method according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of hierarchical partitioning of users according to a closed-loop adaptive individual and region allocation method according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a method of hierarchical attribution model of a closed loop adaptive individual and region allocation method according to an embodiment of the present invention;

fig. 4 is a schematic diagram showing area division of core points in a closed-loop adaptive individual and area allocation method according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of a method for automatically partitioning a region of a closed-loop adaptive individual and region allocation method according to an embodiment of the present invention;

fig. 6 is a schematic diagram of a complete flow of region division of core points in a closed-loop adaptive individual and region allocation method according to an embodiment of the present invention;

fig. 7 is a schematic flow chart of a parameter optimization method of a closed-loop adaptive individual and region allocation method according to an embodiment of the present invention;

FIG. 8 is a schematic diagram of a closed-loop adaptive individual and area distribution device according to an embodiment of the present invention;

FIG. 9 illustrates a schematic diagram of a computing device provided by an embodiment of the present invention.

Detailed Description

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

Fig. 1 is a schematic flow chart of a closed-loop adaptive individual and region allocation method according to an embodiment of the present invention. As shown in fig. 1, the closed-loop adaptive individual and region allocation method includes:

Step S11: and carrying out hierarchical partition processing on the users according to the instant user distribution data by applying a hierarchical attribution model.

In an embodiment of the invention, a hierarchical attribution model is applied to attribution all users to meaningful small partitions, such as residential communities. The small partitions will be further layered, i.e., nested. In the case that the partition is not fully paved, the hierarchical attribution model is processed in a targeted manner according to the condition. The data relied on by the hierarchical attribution model is partition data (position and boundary) of each hierarchy and user resident data. The user premises data distinguishes between different types, and the identity of the premises uses the latitude and longitude of the grid in which the user is located.

In the embodiment of the invention, all users of the previous-level partition are attributed to the current-level partition according to the user resident data application level attribution model in the instant user distribution data. For example, as shown in fig. 2, a layer 1 partition is defined according to a layer attribution model applied to the user resident data in the instant user distribution data, all users are attributed to a layer 1 partition to generate a layer 1 partition position library, then a layer 2 partition is defined according to a layer attribution model applied to the user resident data, all users of the layer 1 partition are attributed to a layer 2 partition to generate a layer 2 partition position library, and so on, a layer n partition definition is performed according to a layer attribution model applied to a layer n-1 partition before the layer, and are attributed to a layer n partition to generate a layer n partition position library, and finally the user layered partition attribution library is obtained.

The hierarchical home model of the embodiments of the present invention employs a binary weighted distance (DualWeighted Distance with Cut-Off, DWDCO) model with cut-Off points. The data processing flow is as follows: calculating the partition gravity center according to the user resident data in the instant user distribution data and the current level partition list; calculating the Cartesian distance between the user grid and the center of gravity of the subarea according to the user resident data to form a Cartesian distance vector of the user; calculating a weighted Manhattan distance vector of the user grid and the partition gravity center according to the user resident data and the road network data; calculating a weighted double-distance vector of the user grid and the partition gravity center according to the user Cartesian distance vector and the weighted Manhattan distance vector; and extracting the weighted double-distance vectors of the user grid where the user resides and all the partition barycenters, and taking the partition corresponding to the partition barycenter with the shortest weighted double distance as the home partition of the user. When a weighted Manhattan distance vector is calculated, establishing a shortest path between the center of gravity of the subarea and the user grid according to the road network data, and dividing the shortest path into different straight line sections; and calculating the weighted Manhattan distance vector of the user grid and the partition gravity center according to the weight of each straight line section provided in the road network data.

More specifically, when all users of the previous hierarchical partition are assigned to the current hierarchical partition according to the user resident data application hierarchical assignment model, as shown in fig. 3, it includes:

step S110: and acquiring the user resident data in the instant user distribution data.

The customer premises data is the customer premises data of all the customers of the previous level partition.

Step S111: the partition center of gravity is calculated.

And calculating the partition barycenter according to the user resident data and the current hierarchical partition list. Center of gravity of partition jThe following relationship is satisfied:

wherein p is _i For the weight of user grid i, x _i 、y _i Is the longitude and latitude of the resident of the user grid i.

Step S112: a Cutoff distance vector is calculated.

The user grid of the current hierarchy is ensured to only calculate the distance from the current hierarchy partition according to the hierarchy partition, and a Cutoff distance vector is generated:

step S113: a user cartesian distance vector is calculated.

For each user grid i, calculating the Cartesian distance between the user grid and the center of gravity of the subarea according to the user resident data and the Cutoff distance vector, and forming a Cartesian distance vector of the user. Cartesian distance vectorThe following relationship is satisfied:

step S114: and (5) path segmentation is carried out.

For the center of gravity of each subarea j, establishing the shortest path between the center of gravity of the subarea j and each user grid i by depending on road network data, and dividing the shortest path into different straight line sections Wherein (1)>Representing the kth straight-line segment on the shortest path from partition j to user grid i.

Step S115: a weighted manhattan distance vector is calculated.

For each user grid i, path segmentation provided by the road network data is relied on, and the weight of each road section is consideredComputing a weighted Manhattan distance vector for the user grid i and all partitions j>The following relationship is satisfied:

step S116: a weighted double distance vector is calculated.

For each user grid i, calculating a weighted double distance vector of the user grid i and all partitions j according to the user Cartesian distance vector and the weighted Manhattan distance vectorThe following relationship is satisfied:

wherein,is a partition dual weight. />Weights for the user Cartesian distance vector, +.>Is the weight of the weighted manhattan distance vector.

Step S117: partition polling.

And extracting a resident user grid of the user k and weighted double distance vectors of all partitions j for each user k, and carrying out partition polling.

Step S118: judging whether the shortest distance is the shortest distance. If yes, step S119 is executed; if not, step S117 is performed.

And judging whether the weighted double distance vector of the resident user grid i of the user k and all the partitions j is shortest or not.

Step S119: forming the user partition attribution.

The weighted double distance vector of the resident user grid i of the user k and all the partitions j is shortest, the shortest weighted double distance is extracted, and the partition j is used as the attribution partition of the user k to form the attribution of the user partition. Finally, for each partition j, summarizing the number of its home subscribers mu _j 。

According to the embodiment of the invention, based on objective data, the partitioning data is generated based on a model algorithm, so that the fairness of the result is ensured, the partitioning scheme is based on the user resident data, and can adapt to more complex partitioning conditions, so that the feasibility and universality of the result are higher, the popularization is strong, the multi-level summarization is supported, the Cartesian distance and the Manhattan distance are integrated, and the fairness and the controllability of the distribution are ensured.

Step S12: and carrying out region division on the core points in the core point database according to a pre-stored region division strategy and a user partition attribution data application region automatic division model of partition processing.

The automatic region dividing model of the embodiment of the invention divides regions around core points. The object that operates when dividing a region is a small partition of a specified hierarchy. The user is indirectly assigned to the region by the region assignment of the region as an object assigned to each region. As shown in fig. 4, the input data of the automatic region division model is mainly a user hierarchical partition home base and a core point database, and the core point region allocation is performed by a policy center according to the control of the automatic region division model of the hierarchical region division policy application. The core point database is from other systems, and must contain the position of the core point and its priority for controlling the assigned weight. The area automatic partitioning model is preferably a guaranteed priority strategic allocation (Guaranteed Prioritized Strategic Allocation, GPSA) model, and each core point is automatically allocated to the partition under the control of a strategic center. Parameters controlled by the policy center include: the guard bottom attribution user number and the minimum attribution user number of each core point level; default bottom-preserving weight coefficient η for each core point level _n And default home weight coefficient ζ _n The method comprises the steps of carrying out a first treatment on the surface of the The hierarchy to be allocated is set.

In an embodiment of the present invention, as shown in fig. 5, the method includes:

step S121: and calculating the weighted double-distance vector of the core point and the hierarchical region according to the hierarchical attribution model applied to the core point.

In the embodiment of the invention, according to the configuration of the policy center, a hierarchy area (such as a partition) of a hierarchy to be allocated is selected, and a weighted double-distance vector of each core point and each hierarchy area is calculated. The whole calculation process and method are consistent with the DWDCO model of the hierarchical attribution calculation, and are not repeated here.

Step S122: and establishing a to-be-allocated set of all to-be-allocated hierarchical areas of the designated hierarchy of the policy center.

To-be-allocated setWhere N is all layers belonging locally to the required hierarchyNumber of stage areas, A ⁱ Is the i-th hierarchical region.

Step S123: and obtaining the bottom-keeping combination with the minimum weighted number of people of each core point in the set to be distributed according to the core point grade by applying a core point bottom-keeping distribution model.

The core point policy distribution (Kernel Points Baseline Allocation, KPBA) model has the core goal of better meeting policy center policy requirements for the number of policies set by core point class.

In the embodiment of the invention, the specific flow is as follows: and sequentially extracting the core points from the to-be-allocated set, and constructing all qualified bottom-guaranteeing combinations of each core point, so that the sum of the attribution groups of each qualified bottom-guaranteeing combination is just larger than the requirements of the number of bottom-guaranteeing people. The set of all qualified bottom-guard combinations marking core point n is:wherein M is _n Is the qualified number of bottom-guard combinations of the core point n,N _mn is a combination of the bottom protection->The number of hierarchical regions involved in +.>Is a combination of the bottom protection->An inner ith hierarchy area.

Calculating, for each of the core points, a weighted population for each of the qualifying base combinations of the core points: is the core point n and in combinationWeighted double distance vector, μ for hierarchical region i _i The number of people to whom the inner level region i belongs is combined.

And arranging the core points from high to low according to the grade, and sequentially taking the qualified bottom-guard combination with the smallest number of the weighted persons of each core point as the bottom-guard combination of each core point. The qualified bottom-guard combination with the smallest weighted number of people at the core point n satisfies the relation:viewing assembly->Whether all regions within are in the set to be allocated +.>Until the available area condition is satisfied. Marking the bottom-protecting combination as +. >As a bottom guard combination of core points n.

Step S124: and obtaining the attribution combination with the minimum weighted number of each core point in the remaining hierarchical areas to be distributed of the set to be distributed according to the core point grade by applying a core point attribution distribution model.

The core objective of the core point attribution allocation (Kernel Points Supplemental Allocation, KPSA) model is to allocate all the remaining areas to be allocated in the set D to all the core points according to the priority and the allocation number requirements of each level of the core points, so as to complete the division of the core point attribution areas.

In the embodiment of the invention, the specific flow is as follows: sequentially extracting the core points from the rest hierarchical areas to be allocated of the set to be allocated, and constructing all qualified attribution combinations of each core point, so that the sum of attribution groups of each qualified attribution combination is just larger than the requirement of the attribution number of the core point at the current level. The set of all qualified home combinations for the marked core point n is: M _n as the number of qualified home combinations of core point n,N _mn for home combination->The number of hierarchical regions involved in +.>For home combination->An inner ith hierarchy area.

Calculating, for each of the core points, a weighted population for each of the qualified attribution combinations of the core points: Is the weighted double distance vector of the core point n and the combined inner layer level region i, mu _i The number of people to whom the inner level region i belongs is combined.

And arranging the core points from high to low according to the grade, and sequentially taking the qualified attribution combination with the smallest number of the weighted persons of each core point as the attribution combination of each core point. The qualified attribution combination with the smallest weighted number of people at the core point n satisfies the relation:viewing the home combination->Whether all regions within are at the level to be allocatedRegional collection->Until the available area condition is satisfied. This home combination is marked +.>As the home combination of core points n.

Step S125: and merging the bottom protection combination and the attribution combination of each core point to form the merging attribution combination of the core points.

Combination of core points n

In the embodiment of the present invention, a complete flow of performing region division on core points in a core point database by applying a region automatic division model is shown in fig. 6, and includes:

step S200: the hierarchical region center of gravity is calculated.

Specifically, the center of gravity of the hierarchical region is calculated from the center of gravity of the partition, the core point database, and the hierarchical region obtained from the regional home population.

Step S201: a Cutoff distance vector is calculated.

And calculating the distance between the core point and the hierarchical region according to the region data, and generating a Cutoff distance vector.

Step S202: a cartesian distance vector is calculated.

And calculating the Cartesian distance between the core point and the gravity center of the hierarchical region according to the gravity center of the hierarchical region, the core point database and the Cutoff distance vector, and forming the Cartesian distance vector.

Step S203: and (5) path segmentation is carried out.

And establishing the shortest path between the gravity center of the hierarchical region and each core point by depending on the road network data, and dividing the shortest path into different straight line sections.

Step S204: a weighted manhattan distance vector is calculated.

And calculating weighted Manhattan distance vectors of the core points and all the level areas by considering the road section weights according to the path segmentation provided by the road network data.

Step S205: a weighted double distance vector is calculated.

For each core point, a weighted bi-distance vector of the core point and all hierarchical regions is calculated from the Cartesian distance vector and the weighted Manhattan distance vector.

Step S206: and generating a bottom-protecting combined set.

And establishing a to-be-allocated set of a hierarchical region of the whole designated hierarchy of the policy center, sequentially extracting core points from the to-be-allocated set according to the weighted double-distance vectors, constructing all qualified bottom-protecting combinations of each core point, and generating a bottom-protecting combination set.

Step S207: the weighted population is calculated.

The weighted population for each eligible bottom guard combination of core points is calculated.

Step S208: a minimum weighted combination is generated.

And (3) arranging the core points from high to low according to the grades, sequentially taking the qualified bottom-guard combination with the minimum number of weighted persons of each core point, and generating the minimum weighted combination.

Step S209: the region is attributed.

When all the hierarchical areas in the minimum weighted combination are in the set to be allocated, the minimum weighted combination is taken as a bottom-keeping combination of the core point, namely the core point is attributed to all the hierarchical areas in the minimum weighted combination.

Step S210: a home combination set is generated.

Core points are sequentially extracted from the remaining hierarchical areas to be allocated of the set to be allocated, all qualified attribution combinations of each core point are constructed, and an attribution combination set is generated.

Step S211: the weighted population is calculated.

For each core point, a weighted population of each qualified attribution combination of core points is calculated.

Step S212: a minimum weighted combination is generated.

And (3) arranging the core points from high to low according to the grades, sequentially taking the qualified attribution combination with the smallest weighted number of each core point, and generating the minimum weighted combination. When all the hierarchical regions in the minimum weighted combination are in the hierarchical region set to be allocated, the minimum weighted combination is taken as the attribution combination of each core point.

Step S213: and combining the bottom protection combination with the attribution combination.

And combining the bottom protection combination and the attribution combination of each core point to form a combined attribution combination of the core points, and completing the regional attribution of the core points.

The embodiment of the invention supports multi-level summarization, diversified partition strategy control, supports policy center to control diversified core point hierarchical bottom-protection attribution allocation strategies, ensures the fairness and controllability of allocation, has high automation degree, is compatible with manual strategy configuration, can be flexibly coupled with a service system, and has high application flexibility.

Step S13: and performing closed-loop verification on the hierarchical attribution model and the automatic regional division model according to the service system data, and optimizing parameters of the hierarchical attribution model and the automatic regional division model.

In the embodiment of the invention, core point attribution data of authenticated user partition attribution and hierarchical area is extracted from a butted service system, so that parameters in a DWDCO model and a GPSA model are optimizedThe output data of the closed-loop self-adaptive individual and region allocation method of the embodiment of the invention comprises the regional attribution of the user, the attribution region of the core point and the attribution user group of the core point, and can be connected to a service system for use. The service system can verify the data in the process of service operation and support, and the verification conclusion is automatically fed back to perform parameter optimization for further optimizing the hierarchical attribution model and the automatic region division model.

In the embodiment of the invention, because the parameter space to be optimized is huge and cannot be traversed, the parameter optimization is realized by adopting a gradient descent (Distance BasedGradient Descent, DBGD) model based on distance. As shown in fig. 7, the specific procedure is as follows:

step S131: and respectively acquiring corresponding parameter sequences according to the parameters of the hierarchical attribution model and the automatic regional division model.

Specifically, the parameter sequences in the two models are reformed to obtain the parameter sequence of the DWDCO modelParameter sequence of GPSA model ++>

Step S132: and respectively calculating corresponding deviation functions according to the parameter sequences of the hierarchical attribution model and the automatic regional division model.

The bias function includes a user attribution bias function corresponding to the hierarchical attribution model and a hierarchical region bias function corresponding to the region automatic partitioning model. Defining a user attribution deviation function as:

where Θ is a system function representing the DWDCO model,and N is the total number of users, and is the distance between the center of gravity of the current attribution zone of the ith user and the center of gravity of the actual attribution zone.

Defining a hierarchical region departure function as:

wherein, the xi is a system function representing the GPSA model, For the j-th hierarchy area, M is the target hierarchyIs a hierarchical region total number of (1).

Step S133: and respectively introducing optimization step parameters according to the deviation functions to calculate corresponding deviation gradient functions.

Introducing optimized step length parameter lambda to user attribution deviation function ₁ Introducing an optimized step size parameter lambda into the hierarchical region deviation function ₂ . Respectively calculating the deviation function J _μ (Θ) and G _v Deviation gradient function of (xi):

step S134: and calculating a new parameter sequence and a new deviation function according to the deviation gradient function and the optimization parameter step length until the new deviation function meets a convergence condition.

New parameter sequences { μ '} and { v' } are calculated, respectively. Wherein,

new parameter sequence calculates a new departure function J _μ′ (Θ) and G _ν′ (xi). Judging the new deviation function J _μ′ (Θ) and G _ν′ (xi) whether the convergence condition is satisfied: i J _μ′ (Θ)-J _μ (Θ)|<ε ₁ And |G _v′ (Ξ)-G _v (Ξ)|<ε ₂ ). If the convergence condition is not met, the new parameter sequence and the new deviation function are repeatedly calculated. And if the convergence condition is met, stopping optimizing, wherein the current parameter sequence is the optimized parameter sequence.

According to the embodiment of the invention, the verification closed loop is realized by interfacing the service system, a large number of parameters of the hierarchical attribution model and the regional automatic division model are automatically optimized, the verification data collected by the service system form an optimized closed loop, the self-verification is supported, the current situation is reflected in time, and the continuous improvement of the model can be realized without additional data collection and verification cost.

Step S14: and outputting user partition attribution data and core point attribution data according to the optimized hierarchical attribution model and the parameters of the area automatic division model.

In the embodiment of the invention, the user partition attribution data after the hierarchical attribution model is used for carrying out hierarchical partition processing on the user and the core point attribution data after the core point is subjected to region division by the application region automatic division model are output, wherein the data comprise the partition attribution of the user, the attribution partition of the core point and the attribution user group of the core point.

Fig. 8 is a schematic structural diagram of a closed-loop adaptive individual and area distribution device according to an embodiment of the present invention. As shown in fig. 8, the closed-loop adaptive individual and region allocation apparatus includes: hierarchical partitioning unit 801, region dividing unit 802, parameter verification unit 803, and data output unit 804. Wherein:

the hierarchical partitioning unit 801 is configured to apply a hierarchical attribution model to perform hierarchical partitioning processing on a user according to instant user distribution data; the area dividing unit 802 is configured to perform area division on core points in the core point database according to a pre-stored area dividing policy and an automatic user partition attribution data application area dividing model of partition processing; the parameter verification unit 803 is configured to perform closed-loop verification on the hierarchical attribution model and the automatic regional division model according to service system data, and optimize parameters of the hierarchical attribution model and the automatic regional division model; the data output unit 804 is configured to output user partition attribution data and core point attribution data according to the optimized hierarchical attribution model and the parameters of the area automatic division model.

In an alternative way, the hierarchical partitioning unit 801 is configured to: and all users of the previous-level partition are attributed to the current-level partition according to the user resident data application level attribution model in the instant user distribution data.

In an alternative way, the hierarchical partitioning unit 801 is configured to: calculating the partition gravity center according to the user resident data in the instant user distribution data and the current level partition list; calculating the Cartesian distance between the user grid and the center of gravity of the subarea according to the user resident data to form a Cartesian distance vector of the user; calculating a weighted Manhattan distance vector of the user grid and the partition gravity center according to the user resident data and the road network data; calculating a weighted double-distance vector of the user grid and the partition gravity center according to the user Cartesian distance vector and the weighted Manhattan distance vector; and extracting the weighted double-distance vectors of the user grid where the user resides and all the partition barycenters, and taking the partition corresponding to the partition barycenter with the shortest weighted double distance as the home partition of the user.

In an alternative manner, the area dividing unit 802 is configured to: calculating a weighted double-distance vector of the core point and a hierarchical region by applying the hierarchical attribution model according to the core point; establishing a to-be-allocated set of a whole to-be-allocated hierarchical area of a designated hierarchy of a policy center; a core point bottom-guaranteeing distribution model is applied to obtain a bottom-guaranteeing combination with the minimum number of weighted people of each core point in the set to be distributed according to the core point grade; acquiring a attribution combination with the minimum weighted number of each core point in the remaining hierarchical areas to be distributed of the set to be distributed according to the core point grade by applying a core point attribution distribution model; and merging the bottom protection combination and the attribution combination of each core point to form the merging attribution combination of the core points.

In an alternative manner, the area dividing unit 802 is configured to: sequentially extracting the core points from the to-be-allocated set, and constructing all qualified bottom-guaranteeing combinations of each core point, so that the sum of the attribution groups of each qualified bottom-guaranteeing combination is just larger than the requirements of the number of bottom-guaranteeing people; calculating the weighted number of people of each qualified bottom-guard combination of the core points for each core point; and arranging the core points from high to low according to the grade, and sequentially taking the qualified bottom-guard combination with the smallest number of the weighted persons of each core point as the bottom-guard combination of each core point.

In an alternative manner, the area dividing unit 802 is configured to: sequentially extracting the core points from the rest hierarchical areas to be allocated of the set to be allocated, and constructing all qualified attribution combinations of each core point, so that the sum of attribution groups of each qualified attribution combination is just larger than the requirement of the attribution number of the core point at the current level; calculating a weighted population of each of the qualified attribution combinations of the core points for each of the core points; and arranging the core points from high to low according to the grade, and sequentially taking the qualified attribution combination with the smallest number of the weighted persons of each core point as the attribution combination of each core point.

In an alternative way, the parameter verification unit 803 is configured to: respectively acquiring corresponding parameter sequences according to the parameters of the hierarchical attribution model and the automatic regional division model; respectively calculating corresponding deviation functions according to the parameter sequences of the hierarchical attribution model and the regional automatic division model; respectively introducing optimization step parameters according to the deviation functions to calculate corresponding deviation gradient functions; and calculating a new parameter sequence and a new deviation function according to the deviation gradient function and the optimization parameter step length until the new deviation function meets a convergence condition.

Embodiments of the present invention provide a non-volatile computer storage medium storing at least one executable instruction that may perform the closed-loop adaptive individual and region allocation method of any of the above method embodiments.

The executable instructions may be particularly useful for causing a processor to:

applying a hierarchical attribution model to perform hierarchical partition processing on the user according to the instant user distribution data;

according to a pre-stored regional division strategy and a regional division model applied to the regional attribution data of the user in the regional division process, carrying out regional division on core points in a core point database;

performing closed-loop verification on the hierarchical attribution model and the automatic regional division model according to service system data, and optimizing parameters of the hierarchical attribution model and the automatic regional division model;

and outputting user partition attribution data and core point attribution data according to the optimized hierarchical attribution model and the parameters of the area automatic division model.

In one alternative, the executable instructions cause the processor to:

and all users of the previous-level partition are attributed to the current-level partition according to the user resident data application level attribution model in the instant user distribution data.

In one alternative, the executable instructions cause the processor to:

calculating the partition gravity center according to the user resident data in the instant user distribution data and the current level partition list;

calculating the Cartesian distance between the user grid and the center of gravity of the subarea according to the user resident data to form a Cartesian distance vector of the user;

calculating a weighted Manhattan distance vector of the user grid and the partition gravity center according to the user resident data and the road network data;

calculating a weighted double-distance vector of the user grid and the partition gravity center according to the user Cartesian distance vector and the weighted Manhattan distance vector;

and extracting the weighted double-distance vectors of the user grid where the user resides and all the partition barycenters, and taking the partition corresponding to the partition barycenter with the shortest weighted double distance as the home partition of the user.

In one alternative, the executable instructions cause the processor to:

calculating a weighted double-distance vector of the core point and a hierarchical region by applying the hierarchical attribution model according to the core point;

establishing a to-be-allocated set of a whole to-be-allocated hierarchical area of a designated hierarchy of a policy center;

A core point bottom-guaranteeing distribution model is applied to obtain a bottom-guaranteeing combination with the minimum number of weighted people of each core point in the set to be distributed according to the core point grade;

acquiring a attribution combination with the minimum weighted number of each core point in the remaining hierarchical areas to be distributed of the set to be distributed according to the core point grade by applying a core point attribution distribution model;

and merging the bottom protection combination and the attribution combination of each core point to form the merging attribution combination of the core points.

In one alternative, the executable instructions cause the processor to:

sequentially extracting the core points from the to-be-allocated set, and constructing all qualified bottom-guaranteeing combinations of each core point, so that the sum of the attribution groups of each qualified bottom-guaranteeing combination is just larger than the requirements of the number of bottom-guaranteeing people;

calculating the weighted number of people of each qualified bottom-guard combination of the core points for each core point;

and arranging the core points from high to low according to the grade, and sequentially taking the qualified bottom-guard combination with the smallest number of the weighted persons of each core point as the bottom-guard combination of each core point.

In one alternative, the executable instructions cause the processor to:

Sequentially extracting the core points from the rest hierarchical areas to be allocated of the set to be allocated, and constructing all qualified attribution combinations of each core point, so that the sum of attribution groups of each qualified attribution combination is just larger than the requirement of the attribution number of the core point at the current level;

calculating a weighted population of each of the qualified attribution combinations of the core points for each of the core points;

and arranging the core points from high to low according to the grade, and sequentially taking the qualified attribution combination with the smallest number of the weighted persons of each core point as the attribution combination of each core point.

In one alternative, the executable instructions cause the processor to:

respectively acquiring corresponding parameter sequences according to the parameters of the hierarchical attribution model and the automatic regional division model;

respectively calculating corresponding deviation functions according to the parameter sequences of the hierarchical attribution model and the regional automatic division model;

respectively introducing optimization step parameters according to the deviation functions to calculate corresponding deviation gradient functions;

and calculating a new parameter sequence and a new deviation function according to the deviation gradient function and the optimization parameter step length until the new deviation function meets a convergence condition.

Embodiments of the present invention provide a computer program product comprising a computer program stored on a computer storage medium, the computer program comprising program instructions which, when executed by a computer, cause the computer to perform the closed loop adaptive individual and region allocation method of any of the method embodiments described above.

In one alternative, the executable instructions cause the processor to:

FIG. 9 is a schematic structural diagram of a computing device according to an embodiment of the present invention, and the specific embodiment of the present invention is not limited to the specific implementation of the device.

As shown in fig. 9, the computing device may include: a processor 902, a communication interface (Communications Interface), a memory 906, and a communication bus 908.

Wherein: processor 902, communication interface 904, and memory 906 communicate with each other via a communication bus 908. A communication interface 904 for communicating with network elements of other devices, such as clients or other servers. The processor 902 is configured to execute the program 910, and may specifically perform relevant steps in the above-described embodiments of the closed-loop adaptive individual and region allocation method.

In particular, the program 910 may include program code including computer-operating instructions.

The processor 902 may be a central processing unit, CPU, or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement embodiments of the present invention. The device includes one or each processor, which may be the same type of processor, such as one or each CPU; but may also be different types of processors such as one or each CPU and one or each ASIC.

A memory 906 for storing a program 910. Memory 906 may comprise high-speed RAM memory or may also include non-volatile memory (non-volatile memory), such as at least one disk memory.

The program 910 may be used to cause the processor 902 to perform operations comprising:

In an alternative, the program 910 causes the processor to:

The algorithms or displays presented herein are not inherently related to any particular computer, virtual system, or other apparatus. Various general-purpose systems may also be used with the teachings herein. The required structure for a construction of such a system is apparent from the description above. In addition, embodiments of the present invention are not directed to any particular programming language. It will be appreciated that the teachings of the present invention described herein may be implemented in a variety of programming languages, and the above description of specific languages is provided for disclosure of enablement and best mode of the present invention.

In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.

Similarly, it should be appreciated that in the above description of exemplary embodiments of the invention, various features of the embodiments of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be construed as reflecting the intention that: i.e., the claimed invention requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.

Those skilled in the art will appreciate that the modules in the apparatus of the embodiments may be adaptively changed and disposed in one or more apparatuses different from the embodiments. The modules or units or components of the embodiments may be combined into one module or unit or component and, furthermore, they may be divided into a plurality of sub-modules or sub-units or sub-components. Any combination of all features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or units of any method or apparatus so disclosed, may be used in combination, except insofar as at least some of such features and/or processes or units are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings), may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.

Furthermore, those skilled in the art will appreciate that while some embodiments herein include some features but not others included in other embodiments, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination.

It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, third, etc. do not denote any order. These words may be interpreted as names. The steps in the above embodiments should not be construed as limiting the order of execution unless specifically stated.

Claims

1. A closed loop adaptive individual and region allocation method, the method comprising:

applying a hierarchical attribution model to perform hierarchical partitioning processing on the user according to the instant user distribution data, including: all users of the previous-level partition are attributed to the current-level partition according to a user resident data application level attribution model in the instant user distribution data; the step of attributing all users of the previous level partition to the current level partition according to the user resident data application level attribution model in the instant user distribution data comprises the following steps: calculating the partition gravity center according to the user resident data in the instant user distribution data and the current level partition list; calculating the Cartesian distance between the user grid and the center of gravity of the subarea according to the user resident data to form a Cartesian distance vector of the user; calculating a weighted Manhattan distance vector of the user grid and the partition gravity center according to the user resident data and the road network data; calculating a weighted double-distance vector of the user grid and the partition gravity center according to the user Cartesian distance vector and the weighted Manhattan distance vector; extracting the weighted double-distance vectors of the user grid where the user resides and all the partition barycenters, and taking the partition corresponding to the partition barycenter with the shortest weighted double distance as the attribution partition of the user;

According to a pre-stored regional division strategy and a regional automatic division model applied to user regional attribution data of regional processing, carrying out regional division on core points in a core point database, wherein the regional division method comprises the following steps: calculating a weighted double-distance vector of the core point and a hierarchical region by applying the hierarchical attribution model according to the core point; establishing a to-be-allocated set of a whole to-be-allocated hierarchical area of a designated hierarchy of a policy center; a core point bottom-guaranteeing distribution model is applied to obtain a bottom-guaranteeing combination with the minimum number of weighted people of each core point in the set to be distributed according to the core point grade; acquiring a attribution combination with the minimum weighted number of each core point in the remaining hierarchical areas to be distributed of the set to be distributed according to the core point grade by applying a core point attribution distribution model; combining the bottom protection combination and the attribution combination of each core point to form a combined attribution combination of the core points;

2. The method of claim 1, wherein the applying the core point guard allocation model to obtain a guard combination with a minimum number of weighted persons for each core point in the to-be-allocated set according to a core point level comprises:

3. The method according to claim 2, wherein said applying a core point attribution allocation model to obtain an attribution combination with a minimum number of weighted persons for each of the core points in the remaining hierarchical areas to be allocated of the set to be allocated according to a core point level comprises:

4. The method of claim 1, wherein the performing closed loop verification on the hierarchical attribution model and the area automatic division model according to service system data, optimizing parameters of the hierarchical attribution model and the area automatic division model, comprises:

respectively calculating corresponding deviation functions according to the parameter sequences of the hierarchical attribution model and the automatic regional division model;

and calculating a new parameter sequence and a new deviation function according to the deviation gradient function and the optimization step length parameter until the new deviation function meets a convergence condition.

5. A closed loop adaptive individual and region allocation apparatus, the apparatus comprising:

the hierarchical partitioning unit is used for performing hierarchical partitioning processing on the user by applying a hierarchical attribution model according to the instant user distribution data, and comprises the following steps: all users of the previous-level partition are attributed to the current-level partition according to a user resident data application level attribution model in the instant user distribution data; the step of attributing all users of the previous level partition to the current level partition according to the user resident data application level attribution model in the instant user distribution data comprises the following steps: calculating the partition gravity center according to the user resident data in the instant user distribution data and the current level partition list; calculating the Cartesian distance between the user grid and the center of gravity of the subarea according to the user resident data to form a Cartesian distance vector of the user; calculating a weighted Manhattan distance vector of the user grid and the partition gravity center according to the user resident data and the road network data; calculating a weighted double-distance vector of the user grid and the partition gravity center according to the user Cartesian distance vector and the weighted Manhattan distance vector; extracting the weighted double-distance vectors of the user grid where the user resides and all the partition barycenters, and taking the partition corresponding to the partition barycenter with the shortest weighted double distance as the attribution partition of the user;

The area dividing unit is used for carrying out area division on core points in the core point database according to a pre-stored area dividing strategy and an automatic area dividing model of user partition attribution data of partition processing, and comprises the following steps: calculating a weighted double-distance vector of the core point and a hierarchical region by applying the hierarchical attribution model according to the core point; establishing a to-be-allocated set of a whole to-be-allocated hierarchical area of a designated hierarchy of a policy center; a core point bottom-guaranteeing distribution model is applied to obtain a bottom-guaranteeing combination with the minimum number of weighted people of each core point in the set to be distributed according to the core point grade; acquiring a attribution combination with the minimum weighted number of each core point in the remaining hierarchical areas to be distributed of the set to be distributed according to the core point grade by applying a core point attribution distribution model; combining the bottom protection combination and the attribution combination of each core point to form a combined attribution combination of the core points;

the parameter verification unit is used for performing closed-loop verification on the hierarchical attribution model and the area automatic division model according to the service system data, and optimizing parameters of the hierarchical attribution model and the area automatic division model;

And the data output unit is used for outputting the user partition attribution data and the core point attribution data according to the optimized hierarchical attribution model and the parameters of the area automatic division model.

6. A computing device, comprising: the device comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete communication with each other through the communication bus;

the memory is configured to hold at least one executable instruction that causes the processor to perform the steps of the closed loop adaptive individual and region allocation method according to any one of claims 1-4.

7. A computer storage medium having stored therein at least one executable instruction for causing a processor to perform the steps of the closed loop adaptive individual and region allocation method according to any one of claims 1-4.