CN112418905A - Online advertisement accurate delivery method based on machine learning - Google Patents

Online advertisement accurate delivery method based on machine learning Download PDF

Info

Publication number
CN112418905A
CN112418905A CN202011128887.4A CN202011128887A CN112418905A CN 112418905 A CN112418905 A CN 112418905A CN 202011128887 A CN202011128887 A CN 202011128887A CN 112418905 A CN112418905 A CN 112418905A
Authority
CN
China
Prior art keywords
user
data
online
learning
tgcw
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.)
Pending
Application number
CN202011128887.4A
Other languages
Chinese (zh)
Inventor
任程威
胡冀
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.)
Hangzhou Dianzi University
Original Assignee
Hangzhou Dianzi 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 Hangzhou Dianzi University filed Critical Hangzhou Dianzi University
Priority to CN202011128887.4A priority Critical patent/CN112418905A/en
Publication of CN112418905A publication Critical patent/CN112418905A/en
Pending legal-status Critical Current

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
    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • 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
    • G06N20/00—Machine learning

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Finance (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Development Economics (AREA)
  • Artificial Intelligence (AREA)
  • Accounting & Taxation (AREA)
  • Software Systems (AREA)
  • Strategic Management (AREA)
  • Evolutionary Biology (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Medical Informatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Game Theory and Decision Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention discloses an online advertisement accurate delivery method based on machine learning, which comprises the following steps: s10, data acquisition; s20, extracting feature labels; s30, carrying out TGCW online classification learning; s40, generating categories; and S50, advertisement putting. The method and the system aim at the portrayal of the user in the big data accurate marketing, analyze the flow data generated by the browsing behavior of the user so as to classify the user, and thus realize the accurate advertisement delivery aiming at various types of users.

Description

Online advertisement accurate delivery method based on machine learning
Technical Field
The invention belongs to the field of machine learning, and relates to an online advertisement accurate delivery method based on machine learning.
Background
With the rapid development of information technology, especially the wide application of the internet industry, the real-time processing requirement for massive data arriving at high speed appears in more and more fields. In most cases, service data generated by the e-commerce industry can be regarded as dynamically arriving streaming data, and compared with traditional data, the data has the characteristics of dynamics, disorder, limitless, imbalance, large volume and the like. However, people are usually simulated and peered by things in modern digital advertisement delivery systems, so that how to analyze user figures in real time from massive stream data generated by browsing behaviors of users needs to be considered in order to accurately classify each sample in order to accurately deliver advertisements to each user when describing user figures in accurate marketing of big data.
However, if the browsing behavior of the user is analyzed by using the traditional batch-processing learning method, on one hand, the problems of long learning time and low learning efficiency exist, and the learning classification precision is influenced to a certain extent because the imbalance of data is not considered; on the other hand, it is difficult to efficiently update the model for incremental data, i.e., most of the previously learned models cannot be memorized, resulting in difficulty in efficiently adapting the model to the problems of concept migration and concept evolution occurring in the data. The traditional machine learning algorithm in the batch processing mode becomes more and more compelling under the current big data environment, and the TGCW online learning method is used for directly calculating data in real time in a memory through a flow type calculation frame, so that the method is beneficial to learning user behavior flow data in the E-commerce industry and provides a powerful tool for accurate advertisement delivery.
Therefore, in view of the fact that the existing method is difficult to meet the requirement of online classification of unbalanced stream data, and the accuracy of classification delivery still has room for improvement. The prior art has at least the following disadvantages or problems:
1. the traditional batch learning method is difficult to memorize previously learned models and perform efficient and rapid learning on incremental data, so that the learning algorithms are difficult to apply to accurate and rapid advertisement delivery systems.
2. Existing online learning algorithms do not take into account the imbalance of data when used for online classification learning for user advertisement placement. In reality, a plurality of categories have imbalance problems, which are common and reasonable and meet the expectations of people, so that the problem that the data imbalance is not considered in the practical application of advertisement putting of the conventional online learning classification algorithm is urgently solved.
3. In the past, the classification precision of the online classification learning algorithm still has room for improvement, so the advertisement putting precision can still be improved.
Disclosure of Invention
The invention accurately selects the online characteristics of the online user stream data through the TGCW online learning method, classifies the online user stream data according to the characteristics of user behaviors, and finally can accurately and quickly deliver the corresponding advertisements to the corresponding users. The method solves the problem of unbalanced stream data classification which is not solved by the traditional online classification method, and improves the classification performance to realize the improvement of the advertisement putting accuracy.
The invention provides a classification method for identifying transformer substation equipment types based on three-dimensional point cloud data, which comprises the following steps:
s10, data acquisition;
s20, extracting feature labels;
s30, carrying out TGCW online classification learning;
s40, generating categories;
and S50, advertisement putting.
Preferably, the data acquisition is to collect the user data into the data acquisition module through a cloud.
Preferably, the user data includes the gender, age, location of residence, platform behavior characteristics of the user, browsing duration of the user on the content, and a user tag, wherein the platform behavior characteristics of the user include browsing content of the user, a purchase record of the user, and a rating of the user on the content; the user tags include user interests, user requirements, and user preferences.
Preferably, the feature tag extraction is to extract features and tags by counting data in the data acquisition module.
Preferably, the TGCW online classification learning performs incremental learning to generate a dynamic classification learning model by inputting the extracted features and labels into a TGCW online learning algorithm, and includes the following steps:
s31, setting parameters;
s32, initializing the model;
s33, inputting the user characteristic value corresponding to the characteristic into the TGCW online learning algorithm;
s34, setting weights for different types of data by considering the imbalance of the data and different cost parameters, then carrying out online feature selection on online data by using a confidence weighting method, selecting features which can realize the minimum cost sensitive loss by a truncation gradient algorithm, and finally updating an online learning model.
Preferably, the class generation is to generate a class corresponding to the user by classifying the user corresponding to the feature value by a learned model after inputting the feature and unlabeled stream data into the TGCW online learning algorithm.
Preferably, the advertisement delivery is to push the advertisement corresponding to the user category to the user terminal.
The method and the device set weights for different types of data by considering the imbalance of the user behavior data through the TGCW algorithm and using different cost parameters to perform online feature selection so as to achieve the purpose of accurately classifying the unbalanced data online and finally realize accurate real-time advertisement delivery of different types of users. The method solves the problem that the traditional online learning algorithm does not consider the data imbalance by using the TGCW algorithm, and improves the precision of user classification and advertisement putting. The method has the following specific beneficial effects:
1. the invention replaces the prior conventional online learning algorithm in the advertisement accurate delivery scheme with the TGCW algorithm,
the algorithm adds a cost sensitivity strategy for processing unbalanced stream data, well improves the problem that the conventional algorithm does not consider the unbalance of the data, improves the classification precision, and can more efficiently and accurately analyze the browsing behavior of the user and put corresponding advertisements.
2. The TGCW online classification learning module can simultaneously carry out model learning and model prediction. If the input user feature label data has a label, updating the learning model; if no label exists, the user category of the user can be predicted. And finally, pushing the advertisements corresponding to the categories to the user according to the classified categories.
Drawings
FIG. 1 is a flowchart illustrating steps of a method for accurately delivering online advertisements based on machine learning according to an embodiment of the present invention;
fig. 2 is a flowchart illustrating specific steps of S30 of the method for accurately delivering online advertisements based on machine learning according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
On the contrary, the invention is intended to cover alternatives, modifications, equivalents and alternatives which may be included within the spirit and scope of the invention as defined by the appended claims. Furthermore, in the following detailed description of the present invention, certain specific details are set forth in order to provide a better understanding of the present invention. It will be apparent to one skilled in the art that the present invention may be practiced without these specific details.
In the invention, TGCW is an abbreviation of Truncated Gradient Confidence Weighted, and Chinese is as follows: the gradient confidence weights are truncated.
Referring to fig. 1, a technical solution of the present invention, which is an embodiment of the present invention, is a flow chart of steps of an online advertisement accurate delivery method based on machine learning, and includes the following steps:
s10, data acquisition;
s20, extracting feature labels;
s30, carrying out TGCW online classification learning;
s40, generating categories;
and S50, advertisement putting.
And S10, data acquisition, namely, the user data is collected into the data acquisition module through the cloud.
The user data comprises the gender, the age group, the resident place, the platform behavior characteristics of the user, the browsing duration of the user to the content and a user tag, wherein the platform behavior characteristics of the user comprise the browsing content of the user, the purchasing record of the user and the grade of the user to the content; the user tags include user interests, user requirements, and user preferences.
And S20, extracting feature labels, namely extracting features and labels by counting the data in the data acquisition module.
Referring to fig. 2, S30, TGCW online classification learning, which is to input the extracted features and labels to a TGCW online learning algorithm for incremental learning to generate a dynamic classification learning model, includes the following steps:
s31, setting parameters;
s32, initializing the model;
s33, inputting the user characteristic value corresponding to the characteristic into the TGCW online learning algorithm;
s34, setting weights for different types of data by considering the imbalance of the data and different cost parameters, then carrying out online feature selection on online data by using a confidence weighting method, selecting features which can realize the minimum cost sensitive loss by a truncation gradient algorithm, and finally updating an online learning model.
And S40, category generation, namely after inputting the characteristic and unlabeled stream data into the TGCW online learning algorithm, classifying the user corresponding to the characteristic value by the learned model, and generating a classification category corresponding to the user.
And S50, advertisement putting, namely pushing the advertisement corresponding to the user category to the user terminal.
In a specific embodiment, the user data acquisition module includes the following registration features: gender, age bracket, resident place, platform behavior characteristics of the user: browsing content of the user, purchasing records of the user, grading of the user on the content, browsing duration of the user on the content and a user tag: user interest, user demand and user preference, inputting the data into a user feature tag extraction module, extracting and counting the collected data into the gender of the user
Figure BDA0002734445910000051
(
Figure BDA0002734445910000052
Figure BDA0002734445910000053
Indicating that the user is a female and that,
Figure BDA0002734445910000054
male) age group
Figure BDA0002734445910000055
(
Figure BDA0002734445910000056
Figure BDA0002734445910000057
Indicating that the user is 15-20 years old,
Figure BDA0002734445910000058
is shown as being between 20 and 30 years old,
Figure BDA0002734445910000059
is shown as being 30-40 years old,
Figure BDA00027344459100000510
indicating that the subject is between the ages of 40 and 50,
Figure BDA00027344459100000511
indicating greater than 50 years old), resident locations
Figure BDA00027344459100000512
(
Figure BDA00027344459100000513
Wherein
Figure BDA00027344459100000514
Indicating that the user's resident location is zone 0,
Figure BDA00027344459100000515
the number of the regions is represented as a region 1,
Figure BDA00027344459100000516
is a region 2, and the number of the regions,
Figure BDA0002734445910000061
denoted as zone 3), the platform behavior characteristics of the user: browsing contents of user
Figure BDA0002734445910000062
(
Figure BDA0002734445910000063
Figure BDA0002734445910000064
Figure BDA0002734445910000065
Indicating that the browsing content of the user is a basketball wrister,
Figure BDA0002734445910000066
indicated as a view through the soccer shoe,
Figure BDA0002734445910000067
representing browsing of badminton shoes.), user's scoring of content
Figure BDA0002734445910000068
Figure BDA0002734445910000069
User's browsing duration of content
Figure BDA00027344459100000610
(
Figure BDA00027344459100000611
And a user tag: user interest in ball games Yi(Yi∈{0，1，2，3})，Yi>The label value of 0 indicates that the user interest category corresponding to the record is YiWherein Y isi1 indicates that the user is interested in basketball, Yi2 indicates that the user is interested in football, Yi3 indicates that the user is interested in badminton; if a record has no label, the label is set as Yi0. Setting the parameters of the TGCW online learning algorithm as a gravity parameter g being 0.1, K being 10, Gamma being 0,01, θ is 0.01, C is 0.3, and η is 0.9. Then the obtained statistics are compared
Figure BDA00027344459100000613
Inputting into TGCW algorithm, if the input XiCorresponding Yi>0, updating X according to the steps of the TGCW online learning algorithmiUpdating the online learning model according to the corresponding trust weight parameter omega, importing the updated new model into a user category generating module, generating the user category of the record and inputting the user category into an advertisement putting module; if Y isiIf the user interest category Y is 0, the record is input to a user category generation module, and the predicted user interest category Y is output according to a model in the current modulei', order Yi＝Yi', and will YiAnd inputting an advertisement putting module. Finally, the advertisement putting module puts the ball according to the interest category Y of the useri(YiE {1,2,3}) and 3 ball advertisements corresponding to the interest categories of 3 balls, for example, if the advertisement delivery module receives YiAnd (3) advertising the badminton racket to the user corresponding to the record.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (7)

1. An online advertisement accurate delivery method based on machine learning is characterized by comprising the following steps:
s10, data acquisition;
s20, extracting feature labels;
s30, carrying out TGCW online classification learning;
s40, generating categories;
and S50, advertisement putting.
2. The method of claim 1, wherein the data collection is performed by receiving user data into a data collection module via a cloud.
3. The method of claim 2, wherein the user data comprises gender, age, location of residence of the user, platform behavior characteristics of the user, browsing duration of the user for the content, and user tags, wherein the platform behavior characteristics of the user comprise browsing content of the user, purchasing records of the user, and scores of the user for the content; the user tags include user interests, user requirements, and user preferences.
4. The method of claim 1, wherein the feature tag extraction is to extract features and tags by counting data in a data acquisition module.
5. The method of claim 1, wherein the TGCW online classification learning performs incremental learning to generate a dynamic classification learning model for inputting the extracted features and labels to a TGCW online learning algorithm, comprising the steps of:
s31, setting parameters;
s32, initializing the model;
s33, inputting the user characteristic value corresponding to the characteristic into the TGCW online learning algorithm;
s34, setting weights for different types of data by considering the imbalance of the data and different cost parameters, then carrying out online feature selection on online data by using a confidence weighting method, selecting features which can realize the minimum cost sensitive loss by a truncation gradient algorithm, and finally updating an online learning model.
6. The method of claim 5, wherein the class generation is to generate a class corresponding to the user by classifying the user corresponding to the feature value by a learned model after inputting the feature and unlabeled stream data into the TGCW online learning algorithm.
7. The method of claim 6, wherein the advertisement delivery is to push the advertisement corresponding to the user category to the user terminal.
CN202011128887.4A 2020-10-20 2020-10-20 Online advertisement accurate delivery method based on machine learning Pending CN112418905A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011128887.4A CN112418905A (en) 2020-10-20 2020-10-20 Online advertisement accurate delivery method based on machine learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011128887.4A CN112418905A (en) 2020-10-20 2020-10-20 Online advertisement accurate delivery method based on machine learning

Publications (1)

Publication Number Publication Date
CN112418905A true CN112418905A (en) 2021-02-26

Family

ID=74841559

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011128887.4A Pending CN112418905A (en) 2020-10-20 2020-10-20 Online advertisement accurate delivery method based on machine learning

Country Status (1)

Country Link
CN (1) CN112418905A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114422835A (en) * 2021-12-29 2022-04-29 上海数即数据科技有限公司 Advertisement directional promotion platform based on big data analysis
CN115409553A (en) * 2022-08-30 2022-11-29 南京智慧橙网络科技有限公司 Advertisement delivery system and method based on big data and position information

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110111151A (en) * 2019-05-10 2019-08-09 四川合锐信息技术有限公司 A kind of accurate advertisement analysis method and system based on crowd's label data
CN110310148A (en) * 2019-06-05 2019-10-08 上海易点时空网络有限公司 Method for delivering advertisement accurately based on big data and machine learning

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110111151A (en) * 2019-05-10 2019-08-09 四川合锐信息技术有限公司 A kind of accurate advertisement analysis method and system based on crowd's label data
CN110310148A (en) * 2019-06-05 2019-10-08 上海易点时空网络有限公司 Method for delivering advertisement accurately based on big data and machine learning

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
JI HU 等: "Truncated Gradient Confidence-Weighted Based Online Learning for Imbalance Streaming Data", 《2019 IEEE INTERNATIONAL CONFERENCE ON MULTIMEDIA AND EXPO》, pages 133 - 138 *
高兰兰: "基于行为定向的精准广告投放***的研究与实现", 《中国优秀硕士学位论文全文数据库信息科技辑》, no. 8, pages 11 - 12 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114422835A (en) * 2021-12-29 2022-04-29 上海数即数据科技有限公司 Advertisement directional promotion platform based on big data analysis
CN115409553A (en) * 2022-08-30 2022-11-29 南京智慧橙网络科技有限公司 Advertisement delivery system and method based on big data and position information
CN115409553B (en) * 2022-08-30 2024-05-10 南京智慧橙网络科技有限公司 Advertisement putting system and method based on big data and position information

Similar Documents

Publication Publication Date Title
CN110163647B (en) Data processing method and device
US11734725B2 (en) Information sending method, apparatus and system, and computer-readable storage medium
CN106201871B (en) Based on the Software Defects Predict Methods that cost-sensitive is semi-supervised
CN102708131B (en) By consumer's automatic classification in fine point
CN107346496B (en) Target user orientation method and device
CN109582875A (en) A kind of personalized recommendation method and system of online medical education resource
CN111597446B (en) Content pushing method and device based on artificial intelligence, server and storage medium
CN112418905A (en) Online advertisement accurate delivery method based on machine learning
CN108415913A (en) Crowd's orientation method based on uncertain neighbours
CN103617235A (en) Method and system for network navy account number identification based on particle swarm optimization
CN110597987A (en) Search recommendation method and device
CN110175851A (en) A kind of cheating detection method and device
CN111984873A (en) Service recommendation system and method
CN108810577B (en) User portrait construction method and device and electronic equipment
CN108647789A (en) A kind of intelligent body deep value function learning method based on the sampling of state distributed awareness
CN112287976A (en) Object recommendation method and device, electronic equipment and storage medium
CN107291722B (en) Descriptor classification method and device
CN113407849B (en) Coarse ordering method and device
CN115456676A (en) Game advertisement visual delivery data analysis management method and system
CN115168700A (en) Information flow recommendation method, system and medium based on pre-training algorithm
CN114742599A (en) User value layering method and device based on self-training, storage medium and equipment
CN114757695A (en) Method and device for constructing user purchasing behavior prediction model and computer equipment
CN111353803B (en) Advertiser classification method and device and computing equipment
CN112507214A (en) Data processing method, device, equipment and medium based on user name
CN112613775A (en) Resource quality evaluation method and device, electronic device and storage 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