CN113487367B - Advertisement position prediction system based on cloud computing - Google Patents
Advertisement position prediction system based on cloud computing Download PDFInfo
- Publication number
- CN113487367B CN113487367B CN202110857627.9A CN202110857627A CN113487367B CN 113487367 B CN113487367 B CN 113487367B CN 202110857627 A CN202110857627 A CN 202110857627A CN 113487367 B CN113487367 B CN 113487367B
- Authority
- CN
- China
- Prior art keywords
- price
- matrix
- advertisement
- historical
- data
- 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
Links
- 239000011159 matrix material Substances 0.000 claims abstract description 146
- 238000007405 data analysis Methods 0.000 claims abstract description 67
- 238000007726 management method Methods 0.000 claims abstract description 35
- 238000004891 communication Methods 0.000 claims abstract description 11
- 238000000926 separation method Methods 0.000 claims description 60
- 239000013598 vector Substances 0.000 claims description 37
- 238000012795 verification Methods 0.000 claims description 27
- 238000012545 processing Methods 0.000 claims description 11
- 230000001174 ascending effect Effects 0.000 claims description 8
- 238000000354 decomposition reaction Methods 0.000 claims description 8
- 238000013524 data verification Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 description 12
- 238000000034 method Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0242—Determining effectiveness of advertisements
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0206—Price or cost determination based on market factors
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0273—Determination of fees for advertising
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0277—Online advertisement
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to an advertisement space prediction system based on cloud computing, which comprises: the advertisement management system comprises an advertiser client and an advertisement delivery management platform, wherein the advertiser client and the advertisement delivery management platform are in communication connection; the advertisement delivery management platform comprises a data receiving module, a data analysis module, a price prediction module and a database, wherein the modules are in communication connection. The data receiving module receives an advertisement management request sent by an advertiser client to acquire a historical price sequence. The data analysis module acquires a first price disturbance matrix according to the historical price sequence to acquire a regression model of the first price disturbance matrix and the historical price sequence. The price prediction module establishes an advertisement price prediction function according to the regression model of the first price disturbance matrix and the historical price sequence, obtains predicted price distribution data of the target advertisement position according to the advertisement price prediction function, and then sends the predicted price distribution data to the advertiser client.
Description
Technical Field
The invention relates to the field of cloud computing and digital advertising, in particular to an advertising space prediction system based on cloud computing.
Background
Cloud computing is one type of distributed computing, in which a huge data computing process is broken down into numerous applets by a network "cloud", and then the applets are processed and analyzed by a system of multiple servers to obtain results and returned to the user. The early stage of cloud computing is simple distributed computing, the task distribution is solved, and the computing results are combined. Thus, cloud computing is also known as grid computing. Processing of tens of thousands of data can be completed in a short time by cloud computing, thereby achieving powerful network services.
Existing advertising systems typically analyze based on real-time ad spot prices and cannot predict the prices of ad spots for a period of time in the future in combination with historical ad spot price data. Thus, unnecessary funds may be wasted in determining the corresponding advertising program, such as without regard to the price of the ad spot for a future period of time.
Disclosure of Invention
In view of this, the present invention provides a cloud computing-based advertisement space prediction system, which includes:
the advertisement management system comprises an advertiser client and an advertisement delivery management platform, wherein the advertiser client and the advertisement delivery management platform are in communication connection; the advertisement delivery management platform comprises a data receiving module, a data analysis module, a price prediction module and a database, wherein the modules are in communication connection;
the data receiving module receives an advertisement management request sent by an advertiser client, acquires historical price data of a target advertisement position from a database according to a target identifier, and then performs data processing on the historical price data to obtain a historical price sequence; the advertisement management request includes: a target identifier, a target price threshold, a target prediction period, and a price perturbation factor;
the data analysis module acquires a first price disturbance matrix according to the historical price sequence, constructs a correlation coefficient matrix of the first price disturbance matrix, and then carries out characteristic decomposition on the correlation coefficient matrix to obtain all characteristic values of the correlation coefficient matrix and characteristic vectors corresponding to each characteristic value;
the data analysis module performs feature decomposition on the correlation coefficient matrix, which comprises the following steps: the data analysis module takes the correlation coefficient matrix as a first correlation matrix, carries out matrix similarity transformation on the first correlation matrix to obtain a second correlation matrix, and takes the element with the largest absolute value in non-main diagonal elements in the second correlation matrix as a core element of the second correlation matrix;
the data analysis module acquires an element with the largest median value in a row of the core element in the second correlation matrix as a first element of the second correlation matrix, and acquires an element with the largest median value in a column of the core element in the second correlation matrix as a second element of the second correlation matrix;
the data analysis module obtains a rotation angle according to the core element, the first element and the second element of the second correlation matrix, rotates the second correlation matrix according to the rotation angle, and obtains all eigenvalues of a correlation coefficient matrix of the first price disturbance matrix and eigenvectors corresponding to the eigenvalues according to the rotated second correlation matrix;
the data analysis module obtains a regression model of the first price disturbance matrix and the historical price sequence according to all the characteristic values of the correlation coefficient matrix and the characteristic vector corresponding to each characteristic value;
and the price prediction module establishes an advertisement price prediction function according to the regression model, predicts the advertisement position price of the target advertisement position in the target prediction period according to the advertisement price prediction function so as to obtain predicted price distribution data of the target advertisement position, and sends the predicted price distribution data to the advertiser client.
According to a preferred embodiment, the price perturbation factor is a factor affecting the ad spot price of the targeted ad spot, including time, popularity, audience rating, and traffic; the target identifier is an advertisement position identifier of a target advertisement position, and the advertisement position identifier is used for uniquely identifying the advertisement position; the target prediction period is the prediction period of the advertisement space price; the target price threshold is the highest ad spot price that the advertiser can accept.
According to a preferred embodiment, an advertiser client is a device having a computing function, a storage function, and a communication function for use by an advertiser, comprising: smart phones, desktop computers, notebook computers, smart watches, and smart wearable devices; the advertiser publishes advertising information for marketing goods, providing services, or promoting concepts.
According to a preferred embodiment, the data receiving module performs data processing on the historical price data to obtain a historical price sequence includes:
the data receiving module extracts advertisement space prices of each historical time period in the historical price data;
the data receiving module is used for arranging the advertisement space prices of each historical time period in an ascending order according to time so as to obtain an initial historical price sequence.
According to a preferred embodiment, the data receiving module performs data processing on the historical price data to obtain a historical price sequence includes:
the data receiving module performs primary parameter separation on the initial historical price sequence to obtain a first separation component and a first separation residue;
the data receiving module performs second parameter separation on the first separation residue to obtain a second separation component and a second separation residue;
the data receiving module performs third parameter separation on the second separation residue to obtain a third separation component and a third separation residue; performing iterative operation on the steps until the separation residues cannot be separated continuously;
the data receiving module carries out linear summation on the separation component obtained by each parameter separation and the separation residue obtained by the last parameter separation so as to obtain a historical price sequence.
According to a preferred embodiment, the data receiving module linearly sums the separated component of each parameter separation and the separated residue of the last parameter separation to obtain a historical price sequence comprises:
wherein, ad price (n) is a historical price sequence, c is the number of separated components, l is the index of the separated components, h l (n) is the first separation component, and g (n) is the separation residue.
According to a preferred embodiment, the data analysis module constructs a correlation coefficient matrix of the first price perturbation matrix comprising:
the data analysis module acquires all price disturbance factors of the target advertisement position, and performs data analysis on the historical price sequence to acquire a correlation coefficient of each price disturbance factor and the advertisement position price of each historical time period;
and the data analysis module is used for carrying out ascending arrangement on all the correlation coefficients of each price disturbance factor according to the numerical value to obtain a price disturbance vector of each price disturbance factor, and generating a first price disturbance matrix according to the price disturbance vectors of all the price disturbance factors.
According to a preferred embodiment, the data analysis module obtains a regression model of the first price disturbance matrix and the historical price sequence according to all the eigenvalues of the correlation coefficient matrix and the eigenvector corresponding to each eigenvalue, including:
the data analysis module performs data verification on all feature vectors according to all feature values of the correlation coefficient matrix to obtain a plurality of target feature vectors, generates a target feature matrix according to all the target feature vectors, and then generates a second price disturbance matrix according to the target feature matrix;
the data analysis module establishes a regression model of the second price disturbance matrix and the historical price sequence according to the second price disturbance matrix, and transforms the regression model of the second price disturbance matrix and the historical price sequence to obtain a regression model of the first price disturbance matrix and the historical price sequence.
According to a preferred embodiment, the data analysis module performs data verification on all feature vectors according to all feature values of the correlation coefficient matrix to obtain a plurality of target feature vectors, including:
the data analysis module performs descending order sequencing on all the characteristic values of the correlation coefficient matrix according to the numerical value;
the data analysis module calculates the sum of all the characteristic values to obtain a characteristic sum, takes the ratio of each characteristic value to the characteristic sum as the characteristic occupation ratio of each characteristic value, and then sets the characteristic verification value to be zero;
the data analysis module traverses all the characteristic values according to the arrangement sequence of the characteristic values, and takes the traversed characteristic values as target characteristic values; adding the feature verification value to the feature occupation ratio of the target feature value to update the feature verification value, and comparing the updated feature verification value with a feature verification threshold;
when the updated feature verification value is smaller than or equal to the feature verification threshold, the data analysis module takes the feature vector corresponding to the target feature value as a target feature vector, and traverses the next feature value according to the arrangement sequence of the feature values;
and stopping traversing the characteristic value by the data analysis module when the updated characteristic verification value is larger than the characteristic verification threshold.
According to a preferred embodiment, the price prediction module obtains advertisement space prices of the target advertisement space in each time period in the target prediction period according to the predicted price distribution data of the target advertisement space, and compares the advertisement space prices of each time period with a target price threshold value respectively;
the price prediction module marks all advertisement space prices which are smaller than or equal to the target price threshold value in time, and the advertisement space prices which are smaller than or equal to the target price threshold value are sorted in ascending order according to the price to obtain an advertisement space price analysis table, and then the advertisement space price analysis table is sent to an advertiser client.
According to a preferred embodiment, the price prediction module establishes the advertisement price prediction function according to a regression model of the first price perturbation matrix and the historical price sequence comprises:
P(R(τ)|ad price (n)) is an advertisement price prediction function, R (tau) is a regression model of a first price disturbance matrix and a historical price sequence, ad price (n) is a historical price sequence, τ is price accuracy.
The invention has the following beneficial effects: according to the invention, the price of the target advertisement position in a future period is predicted through the historical price data of the target advertisement position, the advertiser makes a corresponding advertisement delivery plan according to the predicted price of the target advertisement position in the future period, the cost performance of advertisement delivery is improved, a better advertisement effect is obtained with less funds, and the advertisement delivery strategy is optimized.
Drawings
Fig. 1 is a block diagram of an ad spot prediction system based on cloud computing according to an exemplary embodiment.
Detailed Description
In order to make the embodiments, technical solutions and advantages of the present invention more obvious, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the embodiments are some, but not all embodiments of the present invention. It should be understood by those skilled in the art that these embodiments are merely for explaining the technical principles of the present invention, and are not intended to limit the scope of the present invention.
Online advertising can maximize user interest in advertisements, maximize media revenue, and maximize advertiser ROI (return on investment). Advertisement patterns in online advertising are varied and common include Search based on Search engines (Sponsored Search) and Contextual based on web page content (Contextual Ads) or content matching advertisements (Content Match Ads). The advertiser's purchase of ad traffic may specify contractual impressions of media to ad impressions (GD, guaranteed Delivery) by contracting or Non-contractual impressions of Non-fixed impressions traffic (NGD, non-Guaranteed Delivery) limited only by the budget. According to different advertisement delivery platforms and advertisement forms, the pricing methods of advertisement delivery include CPM (Cost-Per-Mile) based on advertisement exposure times, CPC (Cost-Per-Click) based on Click times, CPA (Cost-Per-Action) based on user actions and other pricing methods, wherein CPC/CPA is the pricing method based on advertisement delivery effects, and additionally, the pricing method also comprises a mixed charging method combining CPM and effects.
According to the invention, the historical advertisement price of the target advertisement position is analyzed and processed, so that the advertisement position price of the target advertisement position in a future period of time is predicted, and an advertiser formulates a corresponding advertisement delivery plan according to the predicted advertisement position price. The advertiser can put advertisements in a time period with relatively low advertisement space price, so that the advertisement putting cost and the putting cost can be greatly reduced, and the function of optimizing the advertisement putting strategy of the advertiser is realized.
Referring to fig. 1, in one embodiment, a cloud computing-based ad spot prediction system may include: the advertisement management system comprises an advertiser client and an advertisement delivery management platform, wherein the advertiser client and the advertisement delivery management platform are in communication connection; the advertisement delivery management platform comprises a data receiving module, a data analysis module, a price prediction module and a database, wherein the modules are in communication connection;
the data receiving module receives an advertisement management request sent by an advertiser client, acquires historical price data of a target advertisement position from a database according to a target identifier, and then performs data processing on the historical price data to obtain a historical price sequence; the advertisement management request includes: a target identifier, a target price threshold, a target prediction period, and a price perturbation factor.
The data analysis module acquires a first price disturbance matrix according to the historical price sequence, constructs a correlation coefficient matrix of the first price disturbance matrix, and then carries out characteristic decomposition on the correlation coefficient matrix to obtain all characteristic values of the correlation coefficient matrix and characteristic vectors corresponding to each characteristic value;
the data analysis module performs feature decomposition on the correlation coefficient matrix, which comprises the following steps: the data analysis module takes the correlation coefficient matrix as a first correlation matrix, carries out matrix similarity transformation on the first correlation matrix to obtain a second correlation matrix, and takes the element with the largest absolute value in non-main diagonal elements in the second correlation matrix as a core element of the second correlation matrix;
the data analysis module acquires an element with the largest median value in a row of the core element in the second correlation matrix as a first element of the second correlation matrix, and acquires an element with the largest median value in a column of the core element in the second correlation matrix as a second element of the second correlation matrix;
the data analysis module obtains a rotation angle according to the core element, the first element and the second element of the second correlation matrix, rotates the second correlation matrix according to the rotation angle, and obtains all eigenvalues of a correlation coefficient matrix of the first price disturbance matrix and eigenvectors corresponding to the eigenvalues according to the rotated second correlation matrix;
the data analysis module obtains a regression model of the first price disturbance matrix and the historical price sequence according to all the characteristic values of the correlation coefficient matrix and the characteristic vector corresponding to each characteristic value;
and the price prediction module establishes an advertisement price prediction function according to the regression model, predicts the advertisement position price of the target advertisement position in the target prediction period according to the advertisement price prediction function so as to obtain predicted price distribution data of the target advertisement position, and sends the predicted price distribution data to the advertiser client.
The following is a detailed description of the method and principles of operation of the present invention for ease of understanding.
Specifically, in one embodiment, the workflow executed by the cloud computing-based ad spot prediction system specifically includes the following steps:
s1, a data receiving module of the advertisement putting management platform receives an advertisement management request sent by an advertiser client, acquires historical price data of a target advertisement position from a database according to a target identifier, and then performs data processing on the historical price data to obtain a historical price sequence.
The advertisement management request includes a target identifier, a target price threshold, a target prediction period, and a price perturbation factor. The price disturbance factor is a factor affecting the advertisement space price of the target advertisement space, and comprises time, popularity, audience rating and flow; the target identifier is an advertisement position identifier of the target advertisement position, and the advertisement position identifier is used for carrying out unique identification on the advertisement position; the target prediction period is the prediction period of the advertisement space price; the target price threshold is the highest advertisement space price acceptable by the advertiser; advertisers are market bodies that promote goods, provide services, or promote concepts to release advertising information; the advertisement space price is the advertisement fee that needs to be paid when each advertiser performs advertisement delivery.
When the advertiser needs to put advertisements, an advertisement management request is sent to the advertisement putting management platform, and a target identifier for indicating a target advertisement position, a target price threshold for indicating the highest price of the target advertisement position acceptable by the advertiser and a target prediction period for indicating a prediction period are carried in the advertisement management request.
The advertisement delivery management platform determines a target advertisement position of the advertisement main analysis according to the target identifier, and determines a future time period of the advertisement main prediction according to the target prediction time.
The advertisement delivery management platform responds to the advertisement management request sent by the advertiser client, predicts the price change of the corresponding advertisement position in a future period for the advertiser according to the advertisement delivery demand of the advertiser to obtain an advertisement position price analysis table, and the advertiser makes an advertisement delivery plan according to the advertisement position price analysis table.
In practical situations, the advertisement space prices in different time periods can change, and the fluctuation range of the advertisement space prices is influenced by time, heat, audience rating and flow. Therefore, the invention considers the influence of the price disturbance factor on the advertisement price when predicting the advertisement price, and is beneficial to improving the accuracy of the advertisement delivery management platform for predicting the advertisement price.
An advertiser client is a device with computing, storage, and communication functions for use by an advertiser, comprising: smart phones, desktop computers, notebook computers, smart watches, and smart wearable devices.
When an advertiser performs advertisement delivery, if the advertisement delivery cost is not analyzed and a corresponding advertisement delivery strategy is formulated, high advertisement cost becomes the burden of enterprises when the income brought by advertisement delivery is poor. For example, the advertiser does not consider the influence of the advertisement delivery price on the advertisement delivery profit when the advertiser determines the corresponding advertisement delivery strategy, and performs advertisement delivery at the highest advertisement price point, and when the income caused by advertisement delivery is not ideal, the advertiser also needs to bear high advertisement cost.
To avoid a lack of analysis of ad spot prices by advertisers in determining corresponding ad placement strategies, the advertisers may send corresponding ad management requests to the ad placement management platform for prediction and analysis of ad spot prices to enable the advertisers to further optimize the corresponding ad placement strategies.
The historical price data for the targeted ad spot includes historical prices for each time period of the targeted ad spot history. The historical price data of a targeted ad spot is the advertising costs paid by different advertisers when historically delivering advertisements on the targeted ad spot.
When the historical price data of the target advertisement position is obtained from the database according to the target identifier, all the historical prices of the target advertisement position, namely the historical prices which are longer than the current time point, are not required to be obtained, the reference value is smaller, and the historical prices which are closer to the current time point are more valuable. Thus, the historical price data contains the historical price of the targeted ad spot for the last consecutive period of time.
The historical price sequence is an ordered arrangement of advertisement space prices of the target advertisement space in each historical time period, and comprises the advertisement space prices of the target advertisement space in each historical time period; the advertisement putting management platform analyzes and predicts the advertisement price sequence to obtain the advertisement putting price of the target advertisement position in a future period.
In one embodiment, the data receiving module performs data processing on the historical price data to obtain a historical price sequence includes:
the data receiving module extracts advertisement space prices of each historical time period in the historical price data;
the data receiving module is used for carrying out ascending arrangement on the advertisement space prices of each historical time period according to the time sequence so as to obtain an initial historical price sequence.
In one embodiment, the data receiving module performs data processing on the historical price data to obtain a historical price sequence includes:
the data receiving module performs primary parameter separation on the initial historical price sequence to obtain a first separation component and a first separation residue;
the data receiving module performs second parameter separation on the first separation residue to obtain a second separation component and a second separation residue;
the data receiving module performs third parameter separation on the second separation residue to obtain a third separation component and a third separation residue; performing iterative operation on the steps until the separation residues cannot be separated continuously;
the data receiving module carries out linear summation on the separation component obtained by each parameter separation and the separation residue obtained by the last parameter separation so as to obtain a historical price sequence.
In one embodiment, the data receiving module linearly sums the separated component of each parameter separation and the separated residue of the last parameter separation to obtain the historical price sequence comprises:
wherein, ad price (n) is a historical price sequence, c is the number of separated components, l is the index of the separated components, h l (n) is the first separation component, and g (n) is the separation residue.
According to the invention, the influence of the price disturbance factor on the advertisement space price is considered in the process of predicting the advertisement space price of the target advertisement space, so that the obtained advertisement price prediction function is more accurate, and the accuracy of the advertisement space price prediction result is improved.
S2, a data analysis module acquires all price disturbance factors of the target advertisement space, and performs data analysis on the historical price sequence to acquire a correlation coefficient of each price disturbance factor and the advertisement space price of each historical time period; and the data analysis module is used for carrying out ascending arrangement on all the correlation coefficients of each price disturbance factor according to the values of the correlation coefficients to obtain price disturbance vectors of each price disturbance factor, and generating a first price disturbance matrix according to the price disturbance vectors of all the price disturbance factors.
The correlation coefficient is the correlation degree of the price disturbance factor and the advertisement space price.
S3, the data analysis module constructs a correlation coefficient matrix of the first price disturbance matrix, and performs feature decomposition on the correlation coefficient matrix of the first price disturbance matrix to obtain all feature values of the correlation coefficient matrix and feature vectors corresponding to each feature value.
In one embodiment, the data analysis module performs feature decomposition on the correlation coefficient matrix of the first price disturbance matrix to obtain all feature values of the correlation coefficient matrix of the first price disturbance matrix and feature vectors corresponding to each feature value, where the feature vectors include:
the data analysis module takes a correlation coefficient matrix of the first price disturbance matrix as a first correlation matrix, and carries out matrix similarity transformation on the first correlation matrix to obtain a second correlation matrix;
the data analysis module takes the element with the largest absolute value in the non-main diagonal elements in the second correlation matrix as the core element of the second correlation matrix;
the data analysis module acquires the element with the largest line median of the core elements in the second correlation matrix and takes the element as the first element of the second correlation matrix;
the data analysis module acquires an element with the largest median value in the column of the core element in the second correlation matrix and takes the element as a second element of the second correlation matrix;
the data analysis module obtains a rotation angle according to the core element, the first element and the second element of the second correlation matrix, rotates the second correlation matrix according to the rotation angle, and then obtains all characteristic values of the correlation coefficient matrix and characteristic vectors corresponding to each characteristic value according to the rotated second correlation matrix.
And S4, the data analysis module performs data verification on all the feature vectors according to all the feature values of the correlation coefficient matrix to obtain a plurality of target feature vectors, generates a target feature matrix according to all the target feature vectors, and then generates a second price disturbance matrix according to the target feature matrix.
In one embodiment, the data analysis module performs data verification on all feature vectors according to all feature values of the correlation coefficient matrix to obtain a plurality of target feature vectors, including:
the data analysis module performs descending order sequencing on all the characteristic values of the correlation coefficient matrix according to the numerical values of the characteristic values;
the data analysis module calculates the sum of all the characteristic values to obtain a characteristic sum, takes the ratio of each characteristic value to the characteristic sum as the characteristic occupation ratio of each characteristic value, and then sets the characteristic verification value to be zero;
the data analysis module traverses all the characteristic values according to the arrangement sequence of the characteristic values, takes the traversed characteristic values as target characteristic values, adds the characteristic verification value with the characteristic occupation ratio of the target characteristic values to update the characteristic verification value, and compares the updated characteristic verification value with a characteristic verification threshold;
when the updated feature verification value is smaller than or equal to the feature verification threshold, the data analysis module takes the feature vector corresponding to the target feature value as a target feature vector, and traverses the next feature value according to the arrangement sequence of the feature values;
and stopping traversing the characteristic value by the data analysis module when the updated characteristic verification value is larger than the characteristic verification threshold.
S5, the data analysis module establishes a regression model of the second price disturbance matrix and the historical price sequence according to the second price disturbance matrix, and transforms the regression model of the second price disturbance matrix and the historical price sequence to obtain a regression model of the first price disturbance matrix and the historical price sequence.
And S6, the price prediction module establishes an advertisement price prediction function according to the regression model of the first price disturbance matrix and the historical price sequence, predicts the advertisement position price of the target advertisement position in the target prediction period according to the advertisement price prediction function so as to obtain predicted price distribution data of the target advertisement position, and sends the predicted price distribution data to the advertiser client.
In one embodiment, the price prediction module establishing the advertisement price prediction function based on the regression model of the first price perturbation matrix and the historical price sequence comprises:
P(R(τ)|ad price (n)) is an advertisement price prediction function, R (tau) is a regression model of a first price disturbance matrix and a historical price sequence, ad price (n) is a historical price sequence, τ is price accuracy.
In one embodiment, the price prediction module generates an ad slot price analysis table from the predicted price distribution data for the targeted ad slot;
the price prediction module sends the advertisement space price analysis table to an advertiser client;
and the advertiser makes an advertisement putting plan according to the advertisement space price analysis table.
Specifically, the price prediction module generating the advertisement space price analysis table according to the predicted price distribution data of the target advertisement space includes:
the price prediction module acquires advertisement space prices of the target advertisement space in each time period in the target prediction period according to the predicted price distribution data of the target advertisement space;
the price prediction module is used for comparing the advertisement space price of each time period with a target price threshold value respectively;
the price prediction module marks the time of all the advertisement space prices which are smaller than or equal to the target price threshold value, and the advertisement space prices which are smaller than or equal to the target price threshold value are sorted in ascending order according to the price so as to obtain an advertisement space price analysis table.
The target prediction period is preset by an advertiser according to the advertisement putting requirement of the advertiser. The target prediction period may be a continuous period of time or a specific period of time in the future, i.e., the advertiser may set the prediction period encompassed by the target prediction period to one or more.
The predicted price distribution data comprises advertisement space prices of the target advertisement spaces corresponding to the time periods in the target prediction period, and the predicted advertisement space prices are arranged according to time sequence.
According to the invention, the price of the target advertisement position in a future period is predicted through the historical price data of the target advertisement position, the advertiser makes a corresponding advertisement delivery plan according to the predicted price of the target advertisement position in the future period, the cost performance of advertisement delivery is improved, a better advertisement effect is obtained with less funds, and the advertisement delivery strategy is optimized.
Thus far, the technical solution of the present invention has been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of protection of the present invention is not limited to these specific embodiments. Equivalent modifications and substitutions for related technical features may be made by those skilled in the art without departing from the principles of the present invention, and such modifications and substitutions will be within the scope of the present invention.
Claims (8)
1. The advertisement space prediction system based on cloud computing is characterized by comprising an advertiser client and an advertisement putting management platform, wherein the advertiser client and the advertisement putting management platform are in communication connection; the advertisement delivery management platform comprises a data receiving module, a data analysis module, a price prediction module and a database, wherein the modules are in communication connection;
the data receiving module receives an advertisement management request sent by an advertiser client, acquires historical price data of a target advertisement position from a database according to a target identifier, and then performs data processing on the historical price data to obtain a historical price sequence; the advertisement management request includes: a target identifier, a target price threshold, a target prediction period, and a price perturbation factor;
the data analysis module acquires a first price disturbance matrix according to the historical price sequence, constructs a correlation coefficient matrix of the first price disturbance matrix, and then carries out characteristic decomposition on the correlation coefficient matrix to obtain all characteristic values of the correlation coefficient matrix and characteristic vectors corresponding to each characteristic value;
the data analysis module performs feature decomposition on the correlation coefficient matrix, which comprises the following steps: the data analysis module takes the correlation coefficient matrix as a first correlation matrix, carries out matrix similarity transformation on the first correlation matrix to obtain a second correlation matrix, and takes the element with the largest absolute value in non-main diagonal elements in the second correlation matrix as a core element of the second correlation matrix;
the data analysis module acquires an element with the largest median value in a row of the core element in the second correlation matrix as a first element of the second correlation matrix, and acquires an element with the largest median value in a column of the core element in the second correlation matrix as a second element of the second correlation matrix;
the data analysis module obtains a rotation angle according to the core element, the first element and the second element of the second correlation matrix, rotates the second correlation matrix according to the rotation angle, and obtains all eigenvalues of a correlation coefficient matrix of the first price disturbance matrix and eigenvectors corresponding to the eigenvalues according to the rotated second correlation matrix;
the data analysis module obtains a regression model of the first price disturbance matrix and the historical price sequence according to all the characteristic values of the correlation coefficient matrix and the characteristic vector corresponding to each characteristic value;
and the price prediction module establishes an advertisement price prediction function according to the regression model, predicts the advertisement position price of the target advertisement position in the target prediction period according to the advertisement price prediction function so as to obtain predicted price distribution data of the target advertisement position, and sends the predicted price distribution data to the advertiser client.
2. The system of claim 1, wherein the advertiser client is a device having a computing function, a storage function, and a communication function for use by an advertiser, comprising: smart phones, desktop computers, notebook computers, smart watches, and smart wearable devices.
3. The system of claim 2, wherein the data receiving module performing data processing on the historical price data to obtain the historical price sequence comprises:
the data receiving module extracts advertisement space prices of each historical time period in the historical price data;
the data receiving module is used for arranging the advertisement space prices of each historical time period in an ascending order according to time so as to obtain an initial historical price sequence.
4. A system according to claim 3, wherein the data receiving module performing data processing on the historical price data to obtain the historical price sequence comprises:
the data receiving module performs primary parameter separation on the initial historical price sequence to obtain a first separation component and a first separation residue;
the data receiving module performs second parameter separation on the first separation residue to obtain a second separation component and a second separation residue;
the data receiving module performs third parameter separation on the second separation residue to obtain a third separation component and a third separation residue; performing iterative operation on the steps until the separation residues cannot be separated continuously;
the data receiving module carries out linear summation on the separation component obtained by each parameter separation and the separation residue obtained by the last parameter separation so as to obtain a historical price sequence.
5. The system of claim 4, wherein the data receiving module linearly sums the separated component of each parameter separation and the separated residue of the last parameter separation to obtain the historical price sequence comprises:
wherein, ad price (n) is a historical price sequence, c is the number of separated components, l is the index of the separated components, h l (n) is the first separation component, and g (n) is the separation residue.
6. The system of claim 5, wherein the data analysis module constructing a correlation coefficient matrix for the first price perturbation matrix comprises:
the data analysis module acquires all price disturbance factors of the target advertisement position, and performs data analysis on the historical price sequence to acquire a correlation coefficient of each price disturbance factor and the advertisement position price of each historical time period;
and the data analysis module is used for carrying out ascending arrangement on all the correlation coefficients of each price disturbance factor according to the numerical value to obtain a price disturbance vector of each price disturbance factor, and generating a first price disturbance matrix according to the price disturbance vectors of all the price disturbance factors.
7. The system of claim 6, wherein the data analysis module obtains a regression model of the first price perturbation matrix and the historical price sequence based on all eigenvalues of the correlation coefficient matrix and the eigenvector corresponding to each eigenvalue comprises:
the data analysis module performs data verification on all feature vectors according to all feature values of the correlation coefficient matrix to obtain a plurality of target feature vectors, generates a target feature matrix according to all the target feature vectors, and then generates a second price disturbance matrix according to the target feature matrix;
the data analysis module establishes a regression model of the second price disturbance matrix and the historical price sequence according to the second price disturbance matrix, and transforms the regression model of the second price disturbance matrix and the historical price sequence to obtain a regression model of the first price disturbance matrix and the historical price sequence.
8. The system of claim 7, wherein the data analysis module performing data verification on all feature vectors according to all feature values of the correlation coefficient matrix to obtain a plurality of target feature vectors comprises:
the data analysis module performs descending order sequencing on all the characteristic values of the correlation coefficient matrix according to the numerical value;
the data analysis module calculates the sum of all the characteristic values to obtain a characteristic sum, takes the ratio of each characteristic value to the characteristic sum as the characteristic occupation ratio of each characteristic value, and then sets the characteristic verification value to be zero;
the data analysis module traverses all the characteristic values according to the arrangement sequence of the characteristic values, and takes the traversed characteristic values as target characteristic values; adding the feature verification value to the feature occupation ratio of the target feature value to update the feature verification value, and comparing the updated feature verification value with a feature verification threshold;
when the updated feature verification value is smaller than or equal to the feature verification threshold, the data analysis module takes the feature vector corresponding to the target feature value as a target feature vector, and traverses the next feature value according to the arrangement sequence of the feature values;
and stopping traversing the characteristic value by the data analysis module when the updated characteristic verification value is larger than the characteristic verification threshold.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110857627.9A CN113487367B (en) | 2021-07-28 | 2021-07-28 | Advertisement position prediction system based on cloud computing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110857627.9A CN113487367B (en) | 2021-07-28 | 2021-07-28 | Advertisement position prediction system based on cloud computing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113487367A CN113487367A (en) | 2021-10-08 |
| CN113487367B true CN113487367B (en) | 2023-12-26 |
Family
ID=77944254
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110857627.9A Active CN113487367B (en) | 2021-07-28 | 2021-07-28 | Advertisement position prediction system based on cloud computing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113487367B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102116626A (en) * | 2009-12-31 | 2011-07-06 | 北京控制工程研究所 | Prediction and correction method of node of star point track image |
| CN102663519A (en) * | 2012-04-01 | 2012-09-12 | 浙江盘石信息技术有限公司 | Optimization system of media selection in network advertisement delivery and method thereof |
| CN112215643A (en) * | 2020-10-12 | 2021-01-12 | 上海酷量信息技术有限公司 | Preloading system and method based on historical advertising price |
| US10938674B1 (en) * | 2016-07-01 | 2021-03-02 | EMC IP Holding Company LLC | Managing utilization of cloud computing resources |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150112768A1 (en) * | 2013-10-22 | 2015-04-23 | Craig Klooster | Information gathering and price forecasting for heavy equipment sales |
-
2021
- 2021-07-28 CN CN202110857627.9A patent/CN113487367B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102116626A (en) * | 2009-12-31 | 2011-07-06 | 北京控制工程研究所 | Prediction and correction method of node of star point track image |
| CN102663519A (en) * | 2012-04-01 | 2012-09-12 | 浙江盘石信息技术有限公司 | Optimization system of media selection in network advertisement delivery and method thereof |
| US10938674B1 (en) * | 2016-07-01 | 2021-03-02 | EMC IP Holding Company LLC | Managing utilization of cloud computing resources |
| CN112215643A (en) * | 2020-10-12 | 2021-01-12 | 上海酷量信息技术有限公司 | Preloading system and method based on historical advertising price |
Non-Patent Citations (4)
| Title |
|---|
| 刘永平 ; 郭小波 ; 胡坤伦 ; .采用云计算对粮食价格进行分析和预测.现代食品.2015,(21),全文. * |
| 彭琳 ; 林明 ; .基于NARX神经网络的农产品价格时间序列预测方法研究.农机化研究.2013,(11),全文. * |
| 田红丽 ; 李成群 ; 闫会强 ; 陈昕 ; .基于相关系数和灰色模型的振荡序列预测.计算机工程与设计.2020,(02),全文. * |
| 程毛林 ; 韩云 ; 易雅馨 ; .经济周期波动的组合预测模型与应用.统计与决策.2013,(19),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113487367A (en) | 2021-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yuan et al. | A survey on real time bidding advertising | |
| US8489460B2 (en) | Method and apparatus for advertising bidding | |
| US20090083098A1 (en) | System and method for an online auction with optimal reserve price | |
| US11216850B2 (en) | Predictive platform for determining incremental lift | |
| US20080065479A1 (en) | System and method for optimizing online advertisement auctions by applying linear programming using special ordered sets | |
| US20080027802A1 (en) | System and method for scheduling online keyword subject to budget constraints | |
| US20080275775A1 (en) | System and method for using sampling for scheduling advertisements in an online auction | |
| US20090112690A1 (en) | System and method for online advertising optimized by user segmentation | |
| US20100106556A1 (en) | Time-weighted and scaling optimization of allocation of online advertisement inventory | |
| Najafi-Asadolahi et al. | Cost-per-click pricing for display advertising | |
| US20100106605A1 (en) | Inventory allocation with tradeoff between fairness and maximal value of remaining inventory | |
| WO2007064503A1 (en) | Ad campaign optimization | |
| US20090112691A1 (en) | System and method for scheduling online keyword auctions over multiple time periods subject to budget and query volume constraints | |
| Jiao et al. | Profit maximization auction and data management in big data markets | |
| US8719096B2 (en) | System and method for generating a maximum utility slate of advertisements for online advertisement auctions | |
| US20080027803A1 (en) | System and method for optimizing throttle rates of bidders in online keyword auctions subject to budget constraints | |
| Zhang et al. | Statistical arbitrage mining for display advertising | |
| WO2009158094A2 (en) | Systems and methods for creating an index to measure a performance of digital ads as defined by an advertiser | |
| Bhatnagar et al. | Providing a window of opportunity for converting eStore visitors | |
| CN107886241A (en) | Resource analysis method, apparatus, medium and electronic equipment | |
| US20090254397A1 (en) | System and method for optimizing online keyword auctions subject to budget and estimated query volume constraints | |
| US20160342699A1 (en) | Systems, methods, and devices for profiling audience populations of websites | |
| CN113487367B (en) | Advertisement position prediction system based on cloud computing | |
| Dasgupta et al. | Multi-attribute dynamic pricing for online markets using intelligent agents | |
| CN111401943B (en) | Multi-source advertisement bidding system and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20231116 Address after: 518000, 2nd Floor, Shenzhen Bay Venture Capital Building, No. 25 Haitian Second Road, Binhai Community, Yuehai Street, Nanshan District, Shenzhen City, Guangdong Province Applicant after: Shenzhen youyou Internet Technology Co.,Ltd. Address before: 610000 Chengdu hi tech Incubation Park, north section of Tianfu Avenue, high tech Zone, Wuhou District, Chengdu, Sichuan Applicant before: Chen Qian |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |