CN115099418A

CN115099418A - Processing method, system, equipment and medium based on characteristic multi-time state

Info

Publication number: CN115099418A
Application number: CN202210859043.XA
Authority: CN
Inventors: 张炜林
Original assignee: Shanghai 2345 Network Technology Co ltd
Current assignee: Shanghai 2345 Network Technology Co ltd
Priority date: 2022-07-21
Filing date: 2022-07-21
Publication date: 2022-09-23

Abstract

The application discloses a processing method, a system, equipment and a medium based on feature multi-time state, wherein the processing method comprises the following steps: a 7-day week-level time wheel disc is constructed in the characteristic processor, the time wheel disc automatically rotates along with the lapse of time days, and a complete cycle is formed every other week; the time on each scale stores the shape data of the features of the current day, and is used for describing the change state of the features in one day; when the features are constructed, recording the distance relationship among the features; the distance relation among the characteristics is increased from two dimensions to three-dimensional space, the closer the relation among the characteristics is, the higher the association degree is, the denser the association degree is, and therefore the association relation among the characteristics is judged; and acquiring feature data and association relation of feature multi-time states. The method and the device can meet the requirement of accurately restoring the feature data in machine learning at any time state, accurately restore the features of the specific timeline under any condition, and simultaneously keep the incidence relation of the features.

Description

Processing method, system, equipment and medium based on characteristic multi-time state

Technical Field

The present invention relates to the field of data processing technologies, and in particular, to a processing method, system, device, and medium based on feature multiple time states.

Background

In machine learning, the quality of feature data greatly affects the quality of models and the training dependence of models on features, and model reasoning also depends on features. Therefore, data and features are important components in determining whether a model is of good quality.

The existing industrial-level features mainly solve the problem that at present, the feature forms of time and scenes are changed, and the form of the features is changed continuously along with the time so as to reduce the uncertainty of a model algorithm, wherein the mass feature states only keep the latest features.

How to solve the problem that the features in machine learning travel at any time and can flexibly shuttle back to any time node to restore the features under the evolution of time is a urgent need to be solved by technical personnel in the field.

Disclosure of Invention

The present invention is directed to a processing method based on feature multi-time states, so as to solve the problems mentioned in the above technical background.

In order to achieve the purpose, the invention adopts the following technical scheme:

the first aspect of the present application provides a processing method based on feature multiple time states, including:

s1, constructing a 7-day week-level time wheel disc in the feature processor, wherein the time wheel disc automatically rotates along with the lapse of time days, and a complete cycle is formed every other week; the time wheel disc comprises an annular data structure buffer circular queue which is connected end to end, the annular data structure buffer circular queue is divided into a plurality of unit grooves, and each unit groove corresponds to a time scale;

s2, storing the shape data of the feature of the current day by the time on each scale, and describing the change state of the feature in one day;

s3, recording the distance relation among the characteristics when the characteristics are constructed;

s4, increasing the distance relation between the characteristics from two dimensions to three dimensions, wherein the closer the relation between the characteristics is, the higher the association degree is, the more dense the relation is, thereby judging the association relation between the characteristics;

s5, acquiring feature data and association relation of feature multi-time states, comprising the following steps:

s51, selecting a space coordinate of a feature, and calculating the feature with the similar distance to the space coordinate of the selected feature through a distance formula to obtain an associated feature set of the selected feature;

s52, selecting a time coordinate of a feature, and filtering the associated feature set of the selected feature through a filter algorithm with the time degree of O (n) to obtain a target feature set.

Preferably, in the time wheel, the clockwise direction in the time wheel is defined as the direction from the tail of the line to the head of the line, and the counterclockwise direction is defined as the direction from the head of the line to the tail of the line.

Preferably, in step S2, the shape data of the feature of the current day is stored at each time on a scale, and is stored in a form of a linked list, and the shape data of the feature of the current day is periodically updated through a first-in first-out permutation algorithm.

A second aspect of the present application provides a processing system based on feature multi-time states, comprising:

the time wheel disc construction module constructs a 7-day week-level time wheel disc through the feature processor, the time wheel disc automatically rotates along with the time of days, and a complete cycle is formed every other week; the time wheel disc comprises an annular data structure buffer circular queue which is connected end to end, the annular data structure buffer circular queue is divided into a plurality of unit grooves, and each unit groove corresponds to a time scale;

the scale storage module is used for storing the shape data of the features of the current day at the time of each scale and describing the change state of the features in one day;

the characteristic distance recording module is used for recording the distance relation among the characteristics when the characteristics are constructed;

the characteristic incidence relation determining module is used for increasing the distance relation among the characteristics from two dimensions to three-dimensional space, and the closer the relation among the characteristics is, the higher the incidence degree is, the denser the relation among the characteristics is, so that the incidence relation among the characteristics is determined;

the characteristic multi-time-state obtaining module is used for obtaining characteristic data and incidence relation of characteristic multi-time states, and comprises the following steps: an associated feature set acquisition submodule and a target feature set acquisition submodule; the associated feature set acquisition submodule is used for selecting a spatial coordinate of a feature, calculating the feature with a distance similar to that of the spatial coordinate of the selected feature through a distance formula, and obtaining an associated feature set of the selected feature; the target feature set acquisition submodule is used for selecting a time coordinate of one feature, and filtering the associated feature set of the selected feature through a filter algorithm with the time degree of O (n) to obtain a target feature set.

A third aspect of the present application discloses an electronic device, comprising: at least one processor; and a memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform the feature multi-temporal state based processing method as described above.

A fourth aspect of the present application discloses a computer-readable storage medium, on which a computer program is stored, which, when executed, can implement the processing method based on feature multi-time-states as described above.

Compared with the prior art, the technical scheme of the invention has the following beneficial effects:

the method and the device can meet the requirement of accurately restoring the feature data in machine learning at any time state, accurately restore the features of the specific timeline under any condition, and simultaneously keep the incidence relation of the features.

Drawings

The accompanying drawings, which form a part of the present application, are included to provide a further understanding of the present application, and the description and illustrative embodiments of the present application are provided to explain the present application and not to limit the present application. In the drawings:

FIG. 1 is a flow chart of a feature-based multi-temporal-state processing method of the present invention;

FIG. 2 is a schematic diagram of a time wheel configuration;

FIG. 3 is an example of a storage manner of storing the shape data of the feature of the present day at each time on a scale;

FIG. 4 is an exemplary graph of distance relationships between features as they are constructed in accordance with the present invention;

FIG. 5 is an exemplary graph of distance relationships between features as they are constructed according to the present invention, wherein circles refer to features and numbers between circles refer to distances between features;

fig. 6(a) to (d) are exemplary diagrams for raising the distance relationship between the features from two dimensions into a three-dimensional space;

FIG. 7 is a block diagram of a feature multi-time state based processing system of the present invention.

Detailed Description

In order to make the objects, technical solutions and effects of the present invention clearer and clearer, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.

It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order, it being understood that the data so used may be interchanged under appropriate circumstances. Furthermore, the terms "comprises," "comprising," and any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

Example (b):

the existing industrial-level features mainly solve the problem that at present, the feature forms of time and scenes are mainly solved, and mass feature states only keep the latest features, so that the forms of the features are continuously changed along with the passage of time, so that the uncertainty of a model algorithm is reduced, as shown in table 1:

TABLE 1 exemplary change in feature morphology over time

Time line	Feature(s)	Characteristic information	Characteristic relationship
				00:01	FeatureA	5	A and B
02:12	FeatureA	19	A and B
				03:23	FeatureB	21	B and A
04:50	FeatureB	27	B and A
				05:30	FeatureC	19	C and A

The characteristics may change over time. The characteristic state changes in the process of continuous iteration, and information is lost gradually along with the time. When there are N features that are changing constantly and there is an association relationship between the features, we need to be able to draw the association relationship of the features at any time and at any node.

Fig. 1 is a flowchart of a processing method based on feature multi-time-state according to the present application.

As shown in fig. 1, a processing method based on feature multi-time-state specifically includes the following steps:

step S1: a7-day week-level time wheel is constructed in the feature processor, and the time wheel automatically rotates along with the time of days, and every other week is a complete cycle.

In the application, the time Wheel disc adopts a Hash Wheel Timer. A Hash Wheel Timer is a ring structure, which can be thought of as a clock, divided into a number of bins, one bin representing a period of time (the shorter the Timer becomes more accurate), and holds all tasks that expire on that bin with a linked list, while a pointer rotates one bin after another as time passes and performs all tasks that expire in the corresponding linked list. The task decides which bin should be put in by taking the modulus.

Taking fig. 2 as an example, assuming that a grid is a day, the time period that the whole time wheel can represent is 8 days, and if the current pointer points to 2, a task that is executed after 3 days needs to be scheduled, it is obvious that the current pointer should be added to a square of (2+3 ═ 5), and the pointer can be executed after 3 times of walking; if the task is to be executed after 10 days, the task should wait for the pointer to walk through one round zero 2 grids and then execute, so 4 should be put in, and round (1) is saved in the task. When checking the due task, only round of 0 should be executed, and round of other tasks on the grid should be reduced by 1.

The time wheel disc comprises an annular data structure buffer circular queue (namely a circular buffer) which is connected end to end, the annular data structure buffer circular queue is divided into a plurality of unit grooves, each unit groove corresponds to a time scale, and morphological data of the characteristics of the current day are filled in the unit grooves and used for describing the change state of the characteristics of the current day. In the time wheel disc, the clockwise direction in the time wheel disc is defined as the direction from the tail of the team to the head of the team, and the anticlockwise direction is defined as the direction from the head of the team to the tail of the team. And a pointer pointing to the queue tail unit slot is arranged in the circular data structure buffer circular queue.

Step S2: the time on each scale stores the shape data of the feature of the current day, and is used for describing the change state of the feature in one day.

Referring to fig. 3, the time on each scale stores the shape data of the feature of the current day, and the shape data is stored in a form of a linked list, and is periodically updated through a first-in first-out permutation algorithm.

Step S3: the distance relationship between the features is recorded when the features are constructed, and recorded as shown in fig. 4 and 5. In fig. 5, circles refer to features, and numbers between circles refer to distances between features.

Step S4: referring to fig. 6, the distance relationship between the features is raised from two dimensions to three dimensions, and the closer the relationship between the features is, the higher the degree of association is, the denser the relationship between the features is, and the farther the relationship between the features is, the smaller the degree of association is, the more sparse the relationship between the features is, thereby determining the association relationship between the features.

Step S5: and acquiring feature data and association relation of feature multi-time states. The method specifically comprises the following steps:

at S51, the spatial coordinates of a feature, for example, the x1, y1(FeatureA) feature, are selected.

S52, calculating features with similar spatial coordinate distance to the selected features by the distance formula, and obtaining an associated feature set of the selected features, for example, the set of featureb.

Wherein, the distance formula for calculating the distance is as follows:

the distances of the spatial coordinates are similar, a distance threshold value can be preset, and when the calculated distance value between the spatial coordinate of the selected feature and the spatial coordinates of other features is within the range of the preset distance threshold value, the spatial coordinate distances are judged to be similar.

S53, the time coordinate of a feature is selected, e.g., t1(04: 50).

S54, filtering the associated feature set of the selected features by a filter algorithm with a time scale of o (n) to obtain a target feature set, for example, a feature set related to featureb.

On the other hand, the application also discloses a processing system based on the characteristic multi-time state, which comprises: the time wheel acquisition module comprises a time wheel construction module 100, a scale storage module 200, a feature distance recording module 300, a feature association relation judgment module 400 and a feature multi-time-state acquisition module 500.

The time wheel disc construction module 100 constructs a 7-day time wheel disc at a week level in the feature processor, the time wheel disc automatically rotates along with the lapse of time days, and a complete cycle is formed every other week; the time wheel disc comprises an annular data structure buffer circular queue which is connected end to end, the annular data structure buffer circular queue is divided into a plurality of unit grooves, and each unit groove corresponds to one time scale.

The scale storage module 200 is configured to store shape data of the feature of the current day at each time on the scale, and describe a change state of the feature of the current day.

The feature distance recording module 300 is configured to record a distance relationship between features when the features are constructed.

The feature association relation determining module 400 is configured to increase the distance relation between the features from two dimensions to a three-dimensional space, where the closer the relation between the features is, the higher the association degree is, the denser the relation between the features is, so as to determine the association relation between the features.

The feature multi-time-state obtaining module 500 is configured to obtain feature data and an association relationship of feature multi-time states, and includes: and the associated feature set acquisition submodule and the target feature set acquisition submodule. The associated feature set acquisition submodule is used for selecting a spatial coordinate of a feature, calculating the feature with a distance similar to that of the spatial coordinate of the selected feature through a distance formula, and obtaining an associated feature set of the selected feature; the target feature set acquisition submodule is used for selecting a time coordinate of one feature, and filtering the associated feature set of the selected feature through a filter algorithm with the time degree of O (n) to obtain a target feature set.

On the other hand, the application also discloses an electronic device. An electronic device disclosed herein may include a processor and a memory. A memory for storing a computer program; wherein the processor executes the computer program in the memory to implement the method provided by the above-described method embodiments. For a specific implementation process, reference may be made to the related description above, and details are not described herein again.

In a preferred embodiment, the electronic device may include, but is not limited to: personal computers, server computers, workstations, desktop computers, laptop computers, notebook computers, mobile computing devices, smart phones, tablet computers, cellular phones, Personal Digital Assistants (PDAs), handheld devices, messaging devices, wearable computing devices, consumer electronics, and so forth.

The processor may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

The memory may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. Volatile memory can include, for example, Random Access Memory (RAM), cache memory (or the like). The non-volatile memory may include, for example, Read Only Memory (ROM), a hard disk, flash memory, and the like. One or more computer program instructions may be stored on a computer-readable storage medium and executed by a processor to implement the methods of the various embodiments of the present application above and/or other desired functions.

Furthermore, the present invention also provides a computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, is adapted to carry out the method provided by the method embodiments described above.

In practice, the computer program in the present embodiment may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages, for performing the operations of the embodiments of the present application. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.

In practice, the computer-readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

Those of skill would further appreciate that the various illustrative logical blocks, modules, and algorithm steps described in connection with the disclosure herein may be implemented as electronic hardware, computer software, or combinations of both.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus and methods according to embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

To sum up, the technical scheme of the application can meet the requirement of accurately reducing the feature data in machine learning in any time state, accurately reducing the features of a specific timeline in any condition and simultaneously keeping the association relation of the features. The method and the device improve the quality of the feature data in machine learning.

The embodiments of the present invention have been described in detail, but the embodiments are only examples, and the present invention is not limited to the embodiments described above. Any equivalent modifications and substitutions to those skilled in the art are also within the scope of the present invention. Accordingly, equivalent changes and modifications made without departing from the spirit and scope of the present invention should be covered by the present invention.

Claims

1. A processing method based on feature multi-time state is characterized by comprising the following steps:

s51, selecting the space coordinate of a feature, calculating the feature with similar distance with the space coordinate of the selected feature through a distance formula, and obtaining the associated feature set of the selected feature;

2. The processing method according to claim 1, wherein the clockwise direction in the time wheel is defined as the direction from the end of the line to the head of the line, and the counterclockwise direction is defined as the direction from the head of the line to the end of the line.

3. The method according to claim 1, wherein in step S2, the time on each scale is used to store the feature shape data of the current day, and the feature shape data is stored in a linked list, and the feature shape data of the current day is updated periodically by a first-in-first-out permutation algorithm.

4. A feature-based multi-temporal processing system, comprising:

the scale storage module is used for storing shape data of the features of the current day at the time of each scale and describing the change state of the features in one day;

the characteristic multi-time-state acquisition module is used for acquiring characteristic data and association relation of characteristic multi-time states, and comprises: an associated feature set acquisition submodule and a target feature set acquisition submodule; the associated feature set acquisition submodule is used for selecting a spatial coordinate of a feature, calculating the feature with a distance similar to that of the spatial coordinate of the selected feature through a distance formula, and obtaining an associated feature set of the selected feature; the target feature set acquisition submodule is used for selecting a time coordinate of one feature, and filtering the associated feature set of the selected feature through a filter algorithm with the time degree of O (n) to obtain a target feature set.

5. An electronic device, comprising: at least one processor; and a memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform a method of feature multi-temporal state based processing according to any of claims 1-3.

6. A computer-readable storage medium, on which a computer program is stored, wherein the computer program, when executed, implements a method for feature-based multi-temporal-state processing according to any of claims 1-3.