CN113393056B - Crowdsourcing service supply and demand gap prediction method and system based on time sequence - Google Patents

Crowdsourcing service supply and demand gap prediction method and system based on time sequence Download PDF

Info

Publication number
CN113393056B
CN113393056B CN202110774605.6A CN202110774605A CN113393056B CN 113393056 B CN113393056 B CN 113393056B CN 202110774605 A CN202110774605 A CN 202110774605A CN 113393056 B CN113393056 B CN 113393056B
Authority
CN
China
Prior art keywords
graph
demand
current
crowdsourcing service
residual block
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110774605.6A
Other languages
Chinese (zh)
Other versions
CN113393056A (en
Inventor
何伟
李招明
崔立真
郭伟
刘磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong University
Original Assignee
Shandong University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shandong University filed Critical Shandong University
Priority to CN202110774605.6A priority Critical patent/CN113393056B/en
Publication of CN113393056A publication Critical patent/CN113393056A/en
Application granted granted Critical
Publication of CN113393056B publication Critical patent/CN113393056B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/35—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/38—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06315—Needs-based resource requirements planning or analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Tourism & Hospitality (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Business, Economics & Management (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Operations Research (AREA)
  • Computational Linguistics (AREA)
  • Databases & Information Systems (AREA)
  • Biophysics (AREA)
  • Development Economics (AREA)
  • Biomedical Technology (AREA)
  • Game Theory and Decision Science (AREA)
  • Molecular Biology (AREA)
  • Mathematical Physics (AREA)
  • Quality & Reliability (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Primary Health Care (AREA)
  • Educational Administration (AREA)
  • Library & Information Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a method and a system for predicting a crowdsourcing service supply and demand gap based on a time sequence, wherein the method comprises the following steps of: acquiring current crowdsourcing service supply and demand time sequence data and current crowdsourcing service text data; converting the static information of the current crowdsourcing service supply and demand time sequence data into a static graph; converting dynamic information of the current crowdsourcing service supply and demand time sequence data into a recursive graph; and inputting the static graph, the recursive graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap. The method can effectively solve the problem of forecasting the supply and demand gaps in time and space in the prior art, so that mobile crowdsourcing workers (service providers) and users (service demand parties) can be in a balanced state in each area, and the mobile crowdsourcing service platform can be better operated.

Description

Crowdsourcing service supply and demand gap prediction method and system based on time sequence
Technical Field
The disclosure relates to the technical field of computer information communication and service calculation, in particular to a crowdsourcing service supply and demand gap prediction method and system based on time series.
Background
The statements in this section merely provide background information related to the present disclosure and may not constitute prior art.
Crowdsourcing is a new mode for solving problems through crowd sourcing, and generally refers to a distributed problem solving mode in which work tasks performed by full-time employees in the past are outsourced to unspecified solution provider groups in a voluntary manner through an open platform to complete the problem solving mode. The crowdsourcing examples such as dripping and taking a vehicle, beautifying and taking out and the like are applied to the life of people, so that the life quality of people is greatly improved, and the rapid development of the society is promoted.
The development and popularization of the mobile device provide a hardware condition for crowdsourcing service, and the mobile device can collect data required by the crowdsourcing service in time and can transmit the data to the cloud server, so that the mobile crowdsourcing service is easier to realize. For example, embedded sensors on smartphones allow the public to report traffic conditions, weather forecasts and any street problems through a simple mobile application.
The success of mobile crowdsourcing depends on supply and demand balance, namely, a crowdsourcing system can timely discover the demands of users, and enable the demands of the users to be met by crowdsourcing workers (service providers) through reasonable distribution, and if the demands cannot be met timely, the users can experience relatively poor service quality, so that the operation of a crowdsourcing platform is influenced. How crowdsourcing systems meet the upcoming demand with limited resources is a major challenge and recurring problem faced by mobile crowdsourcing. However, demand is often unpredictable, asymmetric, and constantly changing throughout the day. Other factors such as weather conditions, urban events can also lead to irregularities or asymmetries in user demand. How to predict the gap of supply and demand in time and space becomes an urgent problem to be solved in the field.
Disclosure of Invention
In order to solve the defects of the prior art, the disclosure provides a method and a system for predicting a crowdsourcing service supply and demand gap based on a time sequence; the method can effectively solve the problem of forecasting the supply and demand gaps in time and space in the prior art, so that mobile crowdsourcing workers (service providers) and users (service demand parties) can be in a balanced state in each area, and the mobile crowdsourcing service platform can be better operated.
In a first aspect, the present disclosure provides a method for predicting a crowdsourcing service supply and demand gap based on a time series;
the method for predicting the crowdsourcing service supply and demand gap based on the time sequence comprises the following steps:
acquiring time sequence data of supply and demand of the current crowdsourcing service and text data of the current crowdsourcing service;
converting the static information of the current crowdsourcing service supply and demand time sequence data into a static graph;
converting dynamic information of the current crowdsourcing service supply and demand time sequence data into a recursive graph;
and inputting the static graph, the recursive graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap.
In a second aspect, the present disclosure provides a time series based crowd-sourced service supply and demand gap prediction system;
time series based crowd-sourced service supply and demand gap prediction system, comprising:
an acquisition module configured to: acquiring time sequence data of supply and demand of the current crowdsourcing service and text data of the current crowdsourcing service;
a static graph conversion module configured to: converting the static information of the current crowdsourcing service supply and demand time sequence data into a static graph;
a recursive graph conversion module configured to: converting dynamic information of the current crowdsourcing service supply and demand time sequence data into a recursive graph;
an output module configured to: and inputting the static graph, the recursive graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap.
In a third aspect, the present disclosure also provides an electronic device, including:
a memory for non-transitory storage of computer readable instructions; and
a processor for executing the computer readable instructions,
wherein the computer readable instructions, when executed by the processor, perform the method of the first aspect.
In a fourth aspect, the present disclosure also provides a storage medium storing non-transitory computer readable instructions, wherein the non-transitory computer readable instructions, when executed by a computer, perform the instructions of the method of the first aspect.
Compared with the prior art, the beneficial effect of this disclosure is:
according to the invention, dynamic information and static information in the time sequence data can be effectively extracted by converting the supply and demand gap time sequence data into the feature matrix; the use of the residual block can effectively reduce the occurrence probability of gradient disappearance or explosion; the feature matrix containing the textual data information further enhances the accuracy of the prediction results.
The loss function is used in the neural network training process, so that the training of the neural network is more controllable; the advantage of the convolutional neural network is fully utilized by the image representation feature matrix.
Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure and are not to limit the disclosure.
FIG. 1 is a flow chart of a method of the first embodiment;
fig. 2 is a schematic diagram of the internal structure of the residual block according to the first embodiment.
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present disclosure. As used herein, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise, and it should be understood that the terms "comprises" and "comprising", and any 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.
The embodiments and features of the embodiments in the present disclosure may be combined with each other without conflict.
All data are obtained according to the embodiment and are legally applied on the data on the basis of compliance with laws and regulations and user consent.
Example one
The embodiment provides a method for predicting a crowdsourcing service supply and demand gap based on a time sequence;
the method for predicting the crowdsourcing service supply and demand gap based on the time sequence comprises the following steps:
s101: acquiring time sequence data of supply and demand of the current crowdsourcing service and text data of the current crowdsourcing service;
s102: converting the static information of the current crowdsourcing service supply and demand time series data into a static graph;
s103: converting dynamic information of the current crowdsourcing service supply and demand time sequence data into a recursive graph;
s104: and inputting the static graph, the recursive graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap.
The invention can predict the supply and demand gap at the time t +1 according to the supply and demand gap time sequence before the time t of a set place.
Further, the step S101: acquiring time sequence data of supply and demand of the current crowdsourcing service and text data of the current crowdsourcing service; the current crowdsourcing service supply and demand time sequence data specifically comprises the following steps: a series of successive supply and demand differences for a region under study ending at the current time are represented as follows:
Gap＝{gap 1 ,gap 2 ,gap 3 ，…,gap T }，gap i is the supply and demand difference at time i.
Wherein the current crowdsourced service text data comprises: weather and temperature data from weather forecasts about the current time of the area to be studied and whether the current time belongs to a weekday or holiday type of data.
Further, the S102: converting the static information of the current crowdsourcing service supply and demand time sequence data into a static graph; the method specifically comprises the following steps:
s1021: assuming that the current time is t time, segmenting the crowdsourcing service supply and demand time sequence data within the range from t-N time to t-1 time into time sequence data blocks with equal size, wherein a set overlapping proportion exists between the time sequence data blocks; n is a positive integer greater than 1;
s1022: scaling the numerical value in each time sequence data block after segmentation into a [0,1] interval;
s1023: converting the scaled numerical value and the corresponding time into an angle and a radius under a polar coordinate;
s1024: drawing a polar coordinate graph according to the converted polar coordinates;
s1025: generating two Graham Angular Field (GAF) matrices, which are a Graham Angular Sum Field (GASF) matrix and a Graham Angular Difference Field (GADF) matrix, respectively, according to the polar coordinate diagram;
s1026: and converting the GASF matrix and the GADF matrix into a GASF graph and a GADF graph.
Exemplary, S1022: scaling the numerical value in each time sequence data block after segmentation into a [0,1] interval; the method specifically comprises the following steps:
when the static information of the time series is converted into the image, the value in each divided block is scaled to the [0,1] interval according to the scaling formula:
Figure BDA0003154209950000061
wherein X represents a set of supply and demand gap values within a block, X i A single supply and demand gap value (supply and demand difference) within a block is indicated. The value x as a whole i After making a difference with the minimum value min (X) in the set, the quotient is made with the difference between the maximum value max (X) and the minimum value min (X) in the set.
Illustratively, the S1023: converting the scaled numerical value and the corresponding time into an angle and a radius under a polar coordinate; the method specifically comprises the following steps:
converting the scaled gap value and the corresponding time into an angle and a radius of a polar coordinate, wherein the conversion formula is as follows:
Figure BDA0003154209950000062
Figure BDA0003154209950000063
wherein psi i Is a converted angle value; rho i Is the pole diameter corresponding to the angle value;
Figure BDA0003154209950000064
is a notched value after scaling; arccos () is an inverse cosine function in an inverse trigonometric function; i is a time point, cst is a fixed constant, and can be set as appropriate according to the data situation.
Exemplary, S1025: generating a GASF matrix and a GADF matrix according to the polar coordinate graph; the method specifically comprises the following steps:
the conversion formulas of the GASF matrix and the GADF matrix are respectively as follows:
Figure BDA0003154209950000065
Figure BDA0003154209950000066
wherein psi i Is the angle value at time point i, likewise,. Phi j Is the angle value at time point j; i-j | = k; GASF i,j Reflecting the dependence of the superposition of directions in a given k time, which is expressed in the formula as the sum of two angles: psi i +ψ j ，GADF i,j Reflected is the correlation of the directional difference over a given time k, represented in the formula as the difference between two angles: psi i -ψ j . cos () and sin () are sine function, cosine function, gap 'of the trigonometric function, respectively' b The method is a block formed by continuous supply and demand difference values after scaling, namely a supply and demand difference value block, I is a unit vector, and Tr is a transposition operator.
S1026: and converting the GASF matrix and the GADF matrix into a GASF graph and a GADF graph.
Illustratively, the S1026: converting the GASF matrix and the GADF matrix into a GASF graph and a GADF graph; the transformation tool uses the Python environment function pyts image gramian angularfield (),
the call details for this function may look at the official document or other relevant material for the pyts package, a Python package sorted on a time series.
Further, the step S103: converting dynamic information of the current crowdsourcing service supply and demand time sequence data into a recursive graph; the method specifically comprises the following steps:
s1031: assuming that the current time is t time, segmenting the crowdsourcing service supply and demand time sequence data within the range from t-N time to t-1 time into time sequence data blocks with equal size, wherein a set overlapping proportion exists between the time sequence data blocks; n is a positive integer greater than 1;
s1032: calculating the norm of the difference between the time series data blocks according to the segmented time series data blocks;
s1033: and obtaining a two-dimensional matrix with the element of 1 or 0 according to the step function and the norm, representing 0 and 1 by different colors, and further obtaining a colored graph with the color representing the element value of the two-dimensional matrix, wherein the graph is a recursive graph.
For example, because the time series data blocks can be represented by vectors, and the values of the vectors are supply and demand difference values, the difference between the time series data blocks is a vector obtained by subtracting two vectors, the norm of the difference is the norm of the vector, and the norm of the vector is calculated in three ways, namely 1-norm, 2-norm and infinity norm, which is intuitively understood that the norm of each vector corresponds to a value.
It should be understood that the step function is a special continuous time function, is a process of jumping from 0 to 1 and belongs to a singular function
It should be understood that converting the time series into a recursive graph (Recurrence Plot) is to map the states in the time series phase space to a two-dimensional matrix, and the conversion of the states between time blocks is described by the two-dimensional matrix.
Illustratively, calculating a norm of a difference between time-series data blocks from the segmented time-series data blocks; obtaining a two-dimensional matrix with the element of 1 or 0 according to the step function and the norm, and converting the two-dimensional matrix into a recursion graph; the method specifically comprises the following steps:
Figure BDA0003154209950000081
wherein
Figure BDA0003154209950000082
For the function of order, attribute e>0. The step function formula is:
Figure BDA0003154209950000083
further, as shown in fig. 1, the S104: inputting the static graph, the recursion graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap; wherein, the network structure of the convolutional neural network specifically comprises:
four branches in parallel: the first branch circuit, the second branch circuit, the third branch circuit and the fourth branch circuit;
wherein, first branch road includes: a convolution layer a1, a residual block a2 and a residual block a3 connected in sequence; wherein, the input end of the convolution layer a1 is used for inputting a GASF graph;
wherein, the second branch road includes: a convolution layer b1, a residual block b2 and a residual block b3 connected in sequence; wherein, the input end of the convolution layer b1 is used for inputting a GADF graph;
wherein, the third branch road includes: a convolution layer c1, a residual block c2 and a residual block c3 connected in sequence;
wherein, the fourth branch road includes: the embedded layer d1 and the full connection layer d2 are connected in sequence;
the output end of the residual block a3, the output end of the residual block b3 and the output end of the residual block c3 are connected with the input end of the first connecting layer; the first connection layer fuses the output characteristics of the residual block a3, the residual block b3 and the residual block c3 together to obtain a first fusion characteristic matrix;
the output end of the first connecting layer and the output end of the full connecting layer d2 are both connected with the input end of the second connecting layer; the second connection layer fuses the first fusion characteristic matrix output by the first connection layer and the output value of the full connection layer d2 together to obtain a second fusion characteristic matrix;
the output end of the second connection layer is connected with the input end of the full connection layer, and the full connection layer realizes classification of the second fusion characteristic matrix; the full connection layer is connected with the output layer.
Further, the internal structures of the residual block a2, the residual block a3, the residual block b2, the residual block b3, and the residual block c3 are all consistent.
The introduction of the residual block can avoid the gradient vanishing/explosion problem while being able to train a very deep neural network.
Further, the internal structure of the residual block a2 includes: the device comprises a residual block input end, a convolution layer e1, a nonlinear activation function RELU layer, a convolution layer e2, an adder and a residual block output end which are connected in sequence; wherein the input of the residual block is further connected to the input of the adder. The internal structure of the residual block is shown in fig. 2.
The convolution operation of the kth neuron of the l layer in the convolutional layer is as follows:
Figure BDA0003154209950000091
where a is the input, τ is the filter used to extract the features of the input data, with the size n × m, b is a bias vector that centers the data on the origin of coordinates to facilitate application of the activation function, and f is a nonlinear activation function:
RELU(x)＝max(0,x)。
the introduction of the residual block can avoid the gradient vanishing/explosion problem while being able to train a very deep neural network.
In contrast to the previous internal structure diagram of the residual block, it is seen that a residual block includes two convolutional layers, and an activation function is applied between the two convolutional layers. The input of the residual block is X im Obtaining F (X) after residual block training im ) The final output result is
H(X im )＝F(X im )+X im 。
Through the training of the neural network, meaningful and distinctive features can be extracted, and the features can be used for predicting supply and demand gap values.
Further, the S104: inputting the static graph, the recursion graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap; wherein, the training step of the trained convolutional neural network comprises:
s104a1: constructing a training set; the training set includes: historical crowdsourcing service supply and demand time sequence data and historical crowdsourcing service text data of known supply and demand gap labels;
s104a2: converting static information of the historical crowdsourcing service supply and demand time sequence data of the training set into a static graph;
s104a3: converting dynamic information of historical crowdsourcing service supply and demand time sequence data of a training set into a recursion graph;
s104a4: and inputting the static graph, the recursive graph, the historical crowdsourcing service text data and the known supply and demand gap labels into a convolutional neural network, training the convolutional neural network, and stopping training when the loss function of the convolutional neural network obtains a minimum value to obtain the trained convolutional neural network.
And comparing the difference between the actual value and the predicted value in the training process by using the loss function so as to assist the training of the neural network.
The loss function required for neural network training is:
Figure BDA0003154209950000111
wherein, X actual For practical values, θ is the attribute that the entire neural network needs to be trained. In the neural network training, the number of cycles and the number of batches (batch) are set as appropriate according to the data.
Further, the S104: inputting the static graph, the recursion graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting a prediction result of a current crowdsourcing service supply and demand gap; the working principle comprises the following steps:
s104b1: performing feature extraction on the GASF graph by a first branch of the trained convolutional neural network to obtain a first feature matrix;
s104b2: performing feature extraction on the GADF image by a second branch of the trained convolutional neural network to obtain a second feature matrix;
s104b3: performing feature extraction on the recursive graph by a third branch of the trained convolutional neural network to obtain a third feature matrix;
s104b4: performing feature extraction on the current crowdsourcing service text data by a fourth branch of the trained convolutional neural network to obtain a fourth feature matrix;
s104b5: performing feature fusion on the first, second and third feature matrices by the trained convolutional neural network to obtain a first fusion feature matrix;
s104b6: the trained convolutional neural network performs characteristic fusion on the first fusion characteristic matrix and the fourth characteristic matrix to obtain a fifth characteristic matrix; and classifying the fifth characteristic matrix by the trained convolutional neural network to obtain a supply and demand gap prediction result.
Further, the S104b4: performing feature extraction on the current crowdsourcing service text data by a fourth branch of the trained convolutional neural network to obtain a fourth feature matrix; the method specifically comprises the following steps:
inputting the content of the text data into an embedding layer, and extracting a characteristic matrix;
and inputting the features in the feature matrix output by the embedding layer into the full-connection layer for further classification, and outputting by the full-connection layer to obtain a fourth feature matrix.
It should be understood that since the crowd-sourced service supply and demand of a set area is actually affected not only by various external factors but also by the difference between weekends and weekdays, required text information is collected.
When the text embedding technology generates the feature matrix, since the crowd-sourced service supply and demand of a specific area is actually influenced by not only various external factors but also the difference between weekends and weekdays, features in text data including weather, temperature and day types (weekends and weekdays) need to be extracted for predicting supply and demand gap values.
The content of the text data is input to an embedding layer for the purpose of representing the text data features in the hidden space. The embedding layer is a way to represent text data using the latent factor space and allows words to be represented using dense vectors, where a vector is the projection of a word in a continuous vector space. With regard to the implementation of the embedding layer, text data is embedded and features implicit therein are extracted with relevant APIs in a deep learning framework (e.g., hundreds of degrees paddlefold, google's tensflow, facebook's Pytorch, etc.).
The embedding layer is followed by a complete connection layer for further classifying the features in the matrix output by the embedding layer to obtain a feature matrix X ext 。
Further, the S104b5: the trained convolutional neural network performs characteristic fusion on the first characteristic matrix, the second characteristic matrix and the third characteristic matrix to obtain a first fusion characteristic matrix; the method specifically comprises the following steps:
three feature matrixes X obtained after training GASF 、X GADF 、X Rec Through a connection layer, the characteristics contained in the three matrixes are subjected to characteristic combination to obtain X F 。
Further, the S104b6: the trained convolutional neural network performs characteristic fusion on the first fusion characteristic matrix and the fourth characteristic matrix to obtain a fifth characteristic matrix; the trained convolutional neural network classifies the fifth characteristic matrix to obtain a supply and demand gap prediction result; the method specifically comprises the following steps:
the feature matrix X F And X ext Inputting the feature combination into a connection layer;
then, the feature classification is carried out on the full connection layer, and the final prediction result X is output pred 。
According to the invention, dynamic information and static information in the time sequence data can be effectively extracted by converting the supply and demand gap time sequence data into the characteristic matrix; the use of the residual block can effectively reduce the occurrence probability of gradient disappearance or explosion; the feature matrix containing the text data information further enhances the accuracy of the prediction result; a loss function is used in the neural network training process, so that the training of the neural network is more controllable; the advantage of the convolutional neural network is fully utilized by the image representation feature matrix.
Example two
The embodiment provides a crowdsourcing service supply and demand gap prediction system based on a time sequence;
time series based crowd-sourced service supply and demand gap prediction system, comprising:
an acquisition module configured to: acquiring time sequence data of supply and demand of the current crowdsourcing service and text data of the current crowdsourcing service;
a static graph conversion module configured to: converting the static information of the current crowdsourcing service supply and demand time sequence data into a static graph;
a recursive graph conversion module configured to: converting dynamic information of the current crowdsourcing service supply and demand time sequence data into a recursive graph;
an output module configured to: and inputting the static graph, the recursive graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap.
It should be noted here that the above-mentioned obtaining module, static graph converting module, recursive graph converting and outputting module correspond to steps S101 to S104 in the first embodiment, and the above-mentioned modules are the same as the examples and application scenarios realized by the corresponding steps, but are not limited to the contents disclosed in the first embodiment. It should be noted that the modules described above as part of a system may be implemented in a computer system such as a set of computer-executable instructions.
In the foregoing embodiments, the descriptions of the embodiments have different emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
The proposed system can be implemented in other ways. For example, the above-described system embodiments are merely illustrative, and for example, the division of the above-described modules is merely a logical division, and in actual implementation, there may be other divisions, for example, multiple modules may be combined or integrated into another system, or some features may be omitted, or not executed.
EXAMPLE III
The present embodiment also provides an electronic device, including: one or more processors, one or more memories, and one or more computer programs; wherein, a processor is connected with the memory, the one or more computer programs are stored in the memory, and when the electronic device runs, the processor executes the one or more computer programs stored in the memory, so as to make the electronic device execute the method according to the first embodiment.
It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general purpose processors, digital signal processors DSP, application specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and so on. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory may include both read-only memory and random access memory, and may provide instructions and data to the processor, and a portion of the memory may also include non-volatile random access memory. For example, the memory may also store device type information.
In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in a processor or instructions in the form of software.
The method in the first embodiment may be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor. The software modules may be located in ram, flash, rom, prom, or eprom, registers, etc. as is well known in the art. The storage medium is located in a memory, and a processor reads information in the memory and completes the steps of the method in combination with hardware of the processor. To avoid repetition, it is not described in detail here.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the technical solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.
Example four
The present embodiments also provide a computer-readable storage medium for storing computer instructions, which when executed by a processor, perform the method of the first embodiment.
The above description is only a preferred embodiment of the present disclosure and is not intended to limit the present disclosure, and various modifications and changes may be made to the present disclosure by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present disclosure should be included in the protection scope of the present disclosure.

Claims (9)

1. The method for predicting the crowdsourcing service supply and demand gap based on the time sequence is characterized by comprising the following steps of:
acquiring time sequence data of supply and demand of the current crowdsourcing service and text data of the current crowdsourcing service; the current crowdsourcing service supply and demand time sequence data specifically comprises a series of continuous supply and demand difference values of a certain to-be-researched area from the current moment; the current crowdsourced service text data comprises weather and temperature data from weather forecasts about the current time of the area to be researched and data that the current time belongs to a weekday or holiday type;
converting the static information of the current crowdsourcing service supply and demand time sequence data into a static graph;
converting dynamic information of the current crowdsourcing service supply and demand time sequence data into a recursive graph;
inputting the static graph, the recursion graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap;
inputting the static graph, the recursive graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap specifically comprises the following steps:
wherein, the network structure of the convolutional neural network specifically comprises:
four branches in parallel: the first branch circuit, the second branch circuit, the third branch circuit and the fourth branch circuit;
wherein, first branch road includes: a convolution layer a1, a residual block a2 and a residual block a3 connected in sequence; wherein, the input end of the convolution layer a1 is used for inputting a GASF graph;
wherein, the second branch road includes: a convolution layer b1, a residual block b2 and a residual block b3 connected in sequence; wherein, the input end of the convolution layer b1 is used for inputting a GADF graph;
wherein, the third branch road includes: a convolution layer c1, a residual block c2 and a residual block c3 connected in sequence;
wherein, the fourth branch road includes: the embedded layer d1 and the full connection layer d2 are connected in sequence;
the output end of the residual block a3, the output end of the residual block b3 and the output end of the residual block c3 are connected with the input end of the first connecting layer; the first connection layer fuses the output characteristics of the residual block a3, the residual block b3 and the residual block c3 together to obtain a first fusion characteristic matrix;
the output end of the first connecting layer and the output end of the full connecting layer d2 are both connected with the input end of the second connecting layer; the second connection layer fuses the first fusion characteristic matrix output by the first connection layer and the output value of the full connection layer d2 together to obtain a second fusion characteristic matrix;
the output end of the second connecting layer is connected with the input end of the full connecting layer, and the full connecting layer realizes the classification of the second fusion characteristic matrix; the full connection layer is connected with the output layer.
2. The method as claimed in claim 1, wherein the static information of the current time series data of the crowdsourcing service supply and demand is converted into a static map; the method specifically comprises the following steps:
assuming that the current time is t time, segmenting the crowdsourcing service supply and demand time sequence data within the range from t-N time to t-1 time into time sequence data blocks with equal size, wherein a set overlapping proportion exists between the time sequence data blocks; n is a positive integer greater than 1;
scaling the numerical value in each time sequence data block after segmentation into a [0,1] interval;
converting the scaled numerical value and the corresponding time into an angle and a radius under a polar coordinate;
drawing a polar coordinate graph according to the converted polar coordinates;
generating two gram angle field matrixes which are a gram angle matrix, a field GASF matrix and a gram angle difference field GADF matrix according to the polar coordinate graph;
and converting the GASF matrix and the GADF matrix into a GASF graph and a GADF graph.
3. The method as claimed in claim 1, wherein the dynamic information of the current time series data of the crowdsourcing service supply and demand is converted into a recursive graph; the method specifically comprises the following steps:
assuming that the current time is t time, segmenting the crowdsourcing service supply and demand time sequence data within the range from t-N time to t-1 time into time sequence data blocks with equal size, wherein a set overlapping proportion exists between the time sequence data blocks; n is a positive integer greater than 1;
calculating the norm of the difference between the time series data blocks according to the segmented time series data blocks;
and obtaining a two-dimensional matrix with the element of 1 or 0 according to the step function and the norm, representing 0 and 1 by using different colors, and further obtaining a colored graph with the color representing the element value of the two-dimensional matrix, wherein the colored graph is a recursive graph.
4. The method as claimed in claim 1, wherein the static graph, the recursive graph and the current crowdsourcing service text data are input into the trained convolutional neural network, and the current crowdsourcing service supply and demand gap prediction result is output; wherein, the training step of the trained convolutional neural network comprises:
constructing a training set; the training set includes: historical crowdsourcing service supply and demand time sequence data and historical crowdsourcing service text data of known supply and demand gap labels;
converting static information of the historical crowdsourcing service supply and demand time sequence data of the training set into a static graph;
converting dynamic information of historical crowdsourcing service supply and demand time sequence data of a training set into a recursion graph;
and inputting the static graph, the recursive graph, the historical crowdsourcing service text data and the known supply and demand gap labels into a convolutional neural network, training the convolutional neural network, and stopping training when the loss function of the convolutional neural network obtains a minimum value to obtain the trained convolutional neural network.
5. The method as claimed in claim 1, wherein the static graph, the recursive graph and the current crowdsourcing service text data are input into the trained convolutional neural network, and the current crowdsourcing service supply and demand gap prediction result is output; the working principle comprises the following steps:
performing feature extraction on the GASF graph by a first branch of the trained convolutional neural network to obtain a first feature matrix;
performing feature extraction on the GADF image by a second branch of the trained convolutional neural network to obtain a second feature matrix;
performing feature extraction on the recursive graph by a third branch of the trained convolutional neural network to obtain a third feature matrix;
performing feature extraction on the current crowdsourcing service text data by a fourth branch of the trained convolutional neural network to obtain a fourth feature matrix;
the trained convolutional neural network performs characteristic fusion on the first characteristic matrix, the second characteristic matrix and the third characteristic matrix to obtain a first fusion characteristic matrix;
the trained convolutional neural network performs characteristic fusion on the first fusion characteristic matrix and the fourth characteristic matrix to obtain a fifth characteristic matrix; and classifying the fifth characteristic matrix by the trained convolutional neural network to obtain a supply and demand gap prediction result.
6. The method as claimed in claim 5, wherein the trained fourth branch of the convolutional neural network performs feature extraction on the text data of the crowdsourcing service to obtain a fourth feature matrix; the method specifically comprises the following steps:
inputting the content of the text data into an embedding layer, and extracting a characteristic matrix;
and inputting the features in the feature matrix output by the embedding layer into the full-connection layer for further classification, and outputting by the full-connection layer to obtain a fourth feature matrix.
7. Time series-based crowdsourcing service supply and demand gap prediction system is characterized by comprising:
an acquisition module configured to: acquiring time sequence data of supply and demand of the current crowdsourcing service and text data of the current crowdsourcing service; the current crowdsourcing service supply and demand time sequence data specifically comprises a series of continuous supply and demand difference values of a certain to-be-researched area from the current moment; the current crowdsourced service text data comprises weather and temperature data from weather forecasts about the current time of the area to be researched and data that the current time belongs to a weekday or holiday type;
a static graph conversion module configured to: converting the static information of the current crowdsourcing service supply and demand time sequence data into a static graph;
a recursive graph conversion module configured to: converting dynamic information of the current crowdsourcing service supply and demand time sequence data into a recursive graph;
an output module configured to: inputting the static graph, the recursion graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the prediction result of the current crowdsourcing service supply and demand gap;
inputting the static graph, the recursive graph and the current crowdsourcing service text data into the trained convolutional neural network, and outputting the current crowdsourcing service supply and demand gap prediction result specifically as follows:
wherein, the network structure of the convolutional neural network specifically includes:
four branches in parallel: the first branch, the second branch, the third branch and the fourth branch;
wherein, first branch road includes: a convolution layer a1, a residual block a2 and a residual block a3 connected in sequence; wherein, the input end of the convolution layer a1 is used for inputting a GASF graph;
wherein, the second branch road includes: a convolution layer b1, a residual block b2 and a residual block b3 connected in sequence; wherein, the input end of the convolution layer b1 is used for inputting a GADF graph;
wherein, the third branch road includes: a convolution layer c1, a residual block c2 and a residual block c3 connected in sequence;
wherein, the fourth branch road includes: the embedded layer d1 and the full connection layer d2 are connected in sequence;
the output end of the residual block a3, the output end of the residual block b3 and the output end of the residual block c3 are all connected with the input end of the first connecting layer; the first connection layer fuses the output characteristics of the residual block a3, the residual block b3 and the residual block c3 together to obtain a first fusion characteristic matrix;
the output end of the first connecting layer and the output end of the full connecting layer d2 are both connected with the input end of the second connecting layer; the second connection layer fuses the first fusion characteristic matrix output by the first connection layer and the output value of the full connection layer d2 together to obtain a second fusion characteristic matrix;
the output end of the second connection layer is connected with the input end of the full connection layer, and the full connection layer realizes classification of the second fusion characteristic matrix; the full connection layer is connected with the output layer.
8. An electronic device, comprising:
a memory for non-transitory storage of computer readable instructions; and
a processor for executing the computer readable instructions,
wherein the computer readable instructions, when executed by the processor, perform the method of any of claims 1-6.
9. A storage medium storing non-transitory computer-readable instructions, wherein the instructions of the method of any one of claims 1-6 are performed when the non-transitory storage computer-readable instructions are executed by a computer.
CN202110774605.6A 2021-07-08 2021-07-08 Crowdsourcing service supply and demand gap prediction method and system based on time sequence Active CN113393056B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110774605.6A CN113393056B (en) 2021-07-08 2021-07-08 Crowdsourcing service supply and demand gap prediction method and system based on time sequence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110774605.6A CN113393056B (en) 2021-07-08 2021-07-08 Crowdsourcing service supply and demand gap prediction method and system based on time sequence

Publications (2)

Publication Number Publication Date
CN113393056A CN113393056A (en) 2021-09-14
CN113393056B true CN113393056B (en) 2022-11-25

Family

ID=77625675

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110774605.6A Active CN113393056B (en) 2021-07-08 2021-07-08 Crowdsourcing service supply and demand gap prediction method and system based on time sequence

Country Status (1)

Country Link
CN (1) CN113393056B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105046375A (en) * 2015-09-01 2015-11-11 景德金 Seamless steel pipe production key equipment internet of things maintenance prediction system
CN111915024A (en) * 2020-09-25 2020-11-10 点内（上海）生物科技有限公司 Sequence prediction model training method, prediction system, prediction method and medium
CN112508256A (en) * 2020-12-01 2021-03-16 安徽大学 User demand active prediction method and system based on crowdsourcing
CN112584791A (en) * 2018-06-19 2021-03-30 托尼尔公司 Neural network for diagnosing shoulder disorders

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111323228B (en) * 2020-03-20 2022-04-29 广东技术师范大学 Fault diagnosis method based on time series mapping and convolutional neural network
CN111541570B (en) * 2020-04-22 2021-05-07 北京交通大学 Cloud service QoS prediction method based on multi-source feature learning
CN111639969B (en) * 2020-05-28 2023-05-26 浙江大学 Dynamic incentive calculation method, system, equipment and medium for crowdsourcing system
CN112073298B (en) * 2020-08-26 2021-08-17 重庆理工大学 Social network link abnormity prediction system integrating stacked generalization and cost sensitive learning
CN112179654B (en) * 2020-09-28 2022-02-01 西南交通大学 Rolling bearing fault identification method based on GAF-CNN-BiGRU network
CN112396092B (en) * 2020-10-26 2023-09-29 北京航空航天大学 Crowdsourcing developer recommendation method and device
CN112328914B (en) * 2020-11-06 2024-06-21 辽宁工程技术大学 Task allocation method based on space-time crowdsourcing worker behavior prediction
CN112487799A (en) * 2020-12-14 2021-03-12 成都易书桥科技有限公司 Crowdsourcing task recommendation algorithm using extrinsic product attention
CN112712063B (en) * 2021-01-18 2022-04-26 贵州大学 Tool wear value monitoring method, electronic device and storage medium

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105046375A (en) * 2015-09-01 2015-11-11 景德金 Seamless steel pipe production key equipment internet of things maintenance prediction system
CN112584791A (en) * 2018-06-19 2021-03-30 托尼尔公司 Neural network for diagnosing shoulder disorders
CN111915024A (en) * 2020-09-25 2020-11-10 点内（上海）生物科技有限公司 Sequence prediction model training method, prediction system, prediction method and medium
CN112508256A (en) * 2020-12-01 2021-03-16 安徽大学 User demand active prediction method and system based on crowdsourcing

Also Published As

Publication number Publication date
CN113393056A (en) 2021-09-14

Similar Documents

Publication Publication Date Title
Tekouabou et al. Improving parking availability prediction in smart cities with IoT and ensemble-based model
CN112435462B (en) Method, system, electronic device and storage medium for short-time traffic flow prediction
CN109886330B (en) Text detection method and device, computer readable storage medium and computer equipment
CN109936525A (en) A kind of abnormal account preventing control method, device and equipment based on graph structure model
Dragović et al. Simulation modelling of ship-berth link with priority service
CN113095346A (en) Data labeling method and data labeling device
CN111815946A (en) Method and device for determining abnormal road section, storage medium and electronic equipment
CN114549369B (en) Data restoration method and device, computer and readable storage medium
CN111311908A (en) Method and device for identifying and processing repeated traffic information
CN114663655A (en) Image segmentation model training method, image semantic segmentation device and related equipment
CN116582449A (en) Network performance prediction model training method, device, equipment and storage medium
Meng et al. A mobilenet-SSD model with FPN for waste detection
CN109214326A (en) A kind of information processing method, device and system
CN111950791A (en) Flight delay prediction method, device, server and storage medium
CN114332509B (en) Image processing method, model training method, electronic device and automatic driving vehicle
CN113393056B (en) Crowdsourcing service supply and demand gap prediction method and system based on time sequence
CN112163019B (en) Trusted electronic batch record processing method based on block chain and block chain service platform
CN115906988A (en) Neural network inference architecture creation method, neural network inference method and device
CN116561240A (en) Electronic map processing method, related device and medium
CN113192315B (en) Traffic flow distribution prediction method, prediction device and terminal equipment
Bravos et al. A Capability–Driven modelling approach applied in smart transportation & management systems for large scale events
Zhou et al. Deep spatio-temporal convolutional neural network for city traffic flow prediction
CN110070371B (en) Data prediction model establishing method and equipment, storage medium and server thereof
CN110309848A (en) The method that off-line data and stream data real time fusion calculate
CN116778534B (en) Image processing method, device, equipment and medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant