CN110827094B - Anti-cheating method and system for advertisement delivery - Google Patents
Anti-cheating method and system for advertisement delivery Download PDFInfo
- Publication number
- CN110827094B CN110827094B CN201911116609.4A CN201911116609A CN110827094B CN 110827094 B CN110827094 B CN 110827094B CN 201911116609 A CN201911116609 A CN 201911116609A CN 110827094 B CN110827094 B CN 110827094B
- Authority
- CN
- China
- Prior art keywords
- data
- advertisement
- cheating
- flow
- filtering
- 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
- 238000000034 method Methods 0.000 title claims abstract description 17
- 238000010801 machine learning Methods 0.000 claims abstract description 19
- 230000002159 abnormal effect Effects 0.000 claims abstract description 18
- 238000001914 filtration Methods 0.000 claims description 23
- 238000012549 training Methods 0.000 claims description 15
- 238000012360 testing method Methods 0.000 claims description 12
- 238000002372 labelling Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 230000001105 regulatory effect Effects 0.000 claims description 3
- 238000011282 treatment Methods 0.000 claims description 2
- 230000008901 benefit Effects 0.000 abstract description 5
- 230000006399 behavior Effects 0.000 description 8
- 238000012417 linear regression Methods 0.000 description 6
- 238000006243 chemical reaction Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 230000001680 brushing effect Effects 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
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
- G06Q30/0248—Avoiding fraud
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Transfer Between Computers (AREA)
Abstract
The invention discloses an anti-cheating method and system for advertisement delivery, which are characterized in that a plurality of obvious abnormal users are found through business rules, and then a plurality of abnormal users which can be found through multidimensional analysis are found through a machine learning algorithm, so that the advantages of the two are well combined, the identification accuracy of anti-cheating can be effectively improved, and the risk of false killing is reduced.
Description
Technical Field
The invention relates to the field of advertisement delivery, in particular to an anti-cheating method and system for advertisement delivery.
Background
The mobile advertisement brings great flow benefit to advertisers when entering the mobile Internet era, and potential threat of the gray industrial chain related to advertisement cheating also sounds the alarm to industry personnel. It is clear that identifying and intercepting cheating users has significant implications for improving the quality of advertisements and reducing the data variance with third party monitoring companies.
The current common cheating behaviors are divided into machine behaviors and manual behaviors, wherein the machine behaviors comprise IP repeated brushing amount, different IP repeated brushing amounts, intelligent cheating of the machine, flow hijacking and the like. Artificial actions include real water army cheating, etc.
The current optimal strategy for anti-cheating is to lead the cheating cost to increase dramatically, and lead the profit of the cheating behavior to be reduced greatly, thereby reducing the proportion of the cheating behavior in normal business behavior as much as possible. At present, the industry mainly identifies and intercepts cheating users through a plurality of strong rules and operation experience, and common rules comprise equipment numbers and IP duplication elimination, exposure and click frequency in the validity period, abnormal data blacklists, SDK encryption protection and the like.
The machine learning algorithm is divided into a supervised learning algorithm and an unsupervised learning algorithm, and the Linear Regression (LR) belongs to one of the supervised learning algorithms, and has the characteristics of simplicity, high calculation speed and strong interpretability, and is widely applied to various industries.
The current anti-cheating in the industry has the following defects:
1. the method is excessively dependent on the business rule, and the business rule has the characteristics of simplicity and strong interpretability, but the threshold setting is empirically and is easy to crack.
2. The machine learning algorithm LR has not been applied in advertisement anti-cheating, has poor interpretability, and relies on negative examples.
3. The data of the users are changed in real time, and newly added cheating users on some days cannot be accurately found by relying on offline data.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides the anti-cheating method and the system for advertisement delivery, which improve the accuracy of anti-cheating identification and the flow quality.
In order to solve the technical problems, the invention adopts the following technical scheme: an anti-cheating method for advertisement delivery, comprising the following steps:
1) Collecting exposure and click data of advertisements and exposure and click data of users;
2) Preprocessing the data in the step 1) to remove abnormal data;
3) Extracting the characteristics of the data from which the abnormal data are removed, extracting the characteristics and converting the characteristics, and vectorizing the characteristics;
4) Labeling negative samples in the advertisement request by using the vectorized characteristics, and according to positive samples: negative sample = K:1, generating sample data according to the proportion of M: the proportion of N is randomly divided into a training set and a testing set, the training set is input into a machine learning model, and the super parameters of the machine learning model are adjusted according to the testing set to obtain a prediction model;
5) Inputting real-time data and offline data of the advertisement flow into the prediction model to obtain the probability of whether the advertisement flow is the cheating flow, if the probability value is greater than a threshold value, judging the advertisement flow as the cheating flow, filtering the advertisement and returning a filtering identifier; otherwise, the advertisement is returned directly.
Before step 5), the following treatments are also carried out:
a) Requesting an advertisement from an advertisement delivery engine, wherein the advertisement delivery engine writes the log data of the server into a message middleware kafka;
b) Processing the data in kafka, and counting real-time data;
c) Configuring anti-cheating business rules, if the real-time data violates the business rules, judging the requested advertisement as cheating flow, and writing the equipment number and IP of the accessed equipment into a storage system;
d) Reading blacklist data in a storage system, filtering the equipment number and the IP address of the request, directly filtering advertisements if the equipment number or the IP address is in a blacklist, and returning a filtering identification, wherein the blacklist data comprises blacklist equipment and the IP address; otherwise, step 5) is entered.
Correspondingly, the invention also provides an anti-cheating system for advertisement delivery, which comprises an offline system, an online system and a prediction unit; wherein the offline system comprises:
the data acquisition unit is used for acquiring exposure and click data of advertisements and exposure and click data of users;
the preprocessing unit is used for preprocessing the data acquired by the data acquisition unit, removing abnormal data, extracting the characteristics of the data from which the abnormal data is removed, extracting the characteristics and converting the characteristics, and vectorizing the characteristics;
the training unit is used for labeling negative samples in the advertisement request by utilizing the vectorized characteristics, and according to positive samples: negative sample = K:1, generating sample data according to the proportion of M: the proportion of N is randomly divided into a training set and a testing set, the training set is input into a machine learning model, the super parameters of the machine learning model are regulated according to the testing set, a prediction model is obtained, and the operation of a prediction unit is executed;
the online system includes:
a calculation engine for processing the data in kafka and counting real-time data;
the rule engine configures anti-cheating business rules, if the real-time data violates the business rules, the requested advertisement is judged to be cheating flow, and the equipment number and the IP of the accessed equipment are written into the storage system;
advertisement delivery engine: the method comprises the steps that after an advertisement request is received, log data of a server side are written into a message middleware kafka; reading blacklist data in a storage system, filtering the equipment number and the IP address of the request, directly filtering advertisements if the equipment number or the IP address is in a blacklist, and returning a filtering identification, wherein the blacklist data comprises blacklist equipment and the IP address; otherwise, executing the operation of the prediction unit;
the prediction unit is used for inputting real-time data and offline data of advertisement flow of the offline system and the online system into the prediction model to obtain the probability of whether the advertisement flow is the cheating flow, if the probability value is larger than a threshold value, judging the advertisement flow as the cheating flow, filtering the advertisement and returning a filtering identifier; otherwise, the advertisement is returned directly.
The business rule comprises: whether the number of times a certain device is accessed for a certain period of time exceeds a set threshold; whether a certain device has a plurality of IP addresses; whether the exposure times of a certain device in the same time period are larger than a set threshold value or not; whether the number of clicks of a certain device in the same time period is larger than a set threshold. Whether the requested advertisement is cheating flow or not is judged according to the rule, and the method is simple and easy to implement and easy to realize.
In the invention, the real-time data can be the number of times the same device accesses the interface or the number of times the same IP accesses the interface.
The storage system is redis, which is a high-performance key-value database. The redis can play a good role in supplementing the relational database in partial occasions, and is convenient to use.
Compared with the prior art, the invention has the following beneficial effects: the invention combines the advantages and disadvantages of the business rule and the machine learning algorithm, combines the business rule and the machine learning algorithm, improves the anti-cheating identification accuracy of advertisement delivery, improves the flow quality, and reduces the difference between the data monitored by an advertiser and the data monitored by a third-party advertisement platform.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
Anti-cheating is divided into identification of anti-cheating and interception of anti-cheating traffic. Anti-cheating recognition is divided into two parts, namely business rule recognition and recognition based on a machine learning algorithm. Recognition based on machine learning algorithms is divided into model training based on offline data and real-time prediction based on real-time interface data.
The business rule identification is based on experience, and finds that the frequency of accessing advertisements in unit time exceeds the normal range by one user or one IP according to common anomalies in some businesses, and the CTR accessed by one user exceeds the normal range by a certain period, so that the business rule application rule engine can be formulated into a rule for configuration.
The machine learning algorithm identification depends on offline data, can be cleaned according to exposure and click data reported by a client in the last period and request data of a server, is extracted in characteristics, is marked by combining with cheating users (negative samples) found at ordinary times, and is trained to obtain a prediction model of which the users (equipment plus IP) are abnormal users. Namely, the equipment number and the IP are input, the probability of whether the user is an abnormal user is output, and the user can be divided into three types of normal users, suspicious users and high-risk users according to the probability value. The high-risk user traffic is directly intercepted, advertisements are not returned directly, suspicious traffic is brought into an observation list, and normal traffic is returned directly to the advertisements. The method has the advantages that some obvious abnormal users are found through the business rules, then some abnormal users which can be found through multidimensional analysis are found through the machine learning algorithm, the advantages of the abnormal users and the machine learning algorithm are well combined, the identification accuracy of anti-cheating can be effectively improved, and the risk of false killing is reduced.
The invention provides a device for controlling whether to return advertisements according to data of users in a real-time advertisement delivery system.
The flow of the offline system is as follows:
s1, in an offline system, an advertisement SDK (software development kit) reports exposure and click data of advertisements through a data collection service, wherein the data collection service stores exposure and click data of users, and the data specifically comprises data such as equipment numbers, IP, machine types, operating system versions, access time, exposed advertisement IDs and the like of the users.
S2, removing abnormal data (such as abnormal values, messy codes and the like with unpaired formats) through data cleaning.
S3, extracting features, such as a label system of a person is established, performing feature extraction and feature conversion, and vectorizing the features (the data labeling mainly comprises manual labeling, such as frequent conversion of an IP address or region within 1 hour of a certain device, the IP feature of the device is marked as suspicious device by the user, the normal device feature extraction mainly comprises rule extraction when the IP address is not frequently converted, such as high-frequency users when the IP address of the certain device is defined to be changed for more than 30 times within 1 day, and low-frequency users when the IP address is 10-30 times and the feature conversion method is lower than 10 times, wherein the feature conversion method generally comprises methods such as onehot coding, word2vec and the like, and the common features are converted into vectors).
S4, labeling negative sample data of the cheating flow (the negative sample data of the cheating flow comprises vectorized features and labeled classifications obtained in the step S3, such as vectorized features extracted from the step S3 of a certain cheating device and classifications of the cheating device) and subjecting positive and negative samples to K:1, randomly generating data (such as 20% of negative samples, 80% of positive samples are put together and then the order is disturbed, wherein the positive samples are normal advertisement requests, the negative samples are cheating advertisement requests), randomly dividing the data into a training set and a test set according to the proportion of M to N, inputting the training set into an LR model of sparkMlib or a Widedeep model, adjusting super parameters according to the AUC of the test set, and storing an LR offline model for real-time prediction.
In the present invention, K is a constant, and may be set according to actual use, for example, K may be set to 4; m: n may be 8:2 (i.e. 80% data as training set and 20% data as test set), or 7:3 (70% data as training set, 30% data as test set), M: n can be set according to actual requirements.
The online system comprises two parts of real-time data calculation and real-time prediction, and the flow of the online system is as follows:
s1, an advertisement SDK requests an advertisement from an advertisement delivery engine, and the delivery engine writes log data of a server into a message middleware kafka. The server data comprises the accessed equipment ID, machine type, IP, advertisement ID and other data.
S2, the real-time streaming computing engine Apache Flink or spark streaming consumes kafka in real time, counts real-time data, and transmits the data to the rule engine like the times of accessing an interface by one device or the times of accessing the interface by the same IP.
S3, the rule engine configures some anti-cheating business rules, if the number of times of access of the same device in the same time period exceeds a set threshold, the anti-cheating flow is judged, the device number and the IP are written into redis (an open-source log-type Key-Value database which is written and supported by ANSIC language, can be based on memory and can be persistent, and an API. Redis providing multiple languages can be defined as a Key-Value storage system).
S4, the advertisement putting engine reads blacklist data in the redis to filter (the blacklist data is divided into equipment numbers and IP), if the equipment numbers or the IP are in the blacklist, advertisements are directly filtered, and a filtering result is returned to the advertisement client APP (if no advertisement exists, the client is informed that no advertisement exists, the client does not display the advertisement, the advertisement exists, and the client displays advertisement data). Otherwise, carrying out real-time prediction according to the user data.
S5, transmitting real-time data of the user to a cached LR model by combining with some offline data (such as user behavior data (such as active condition in a last period, online time length and last access time) of the user to obtain probability of whether the traffic (namely, the request of the advertisement SDK in S1 to the advertisement delivery engine) is cheating traffic, if the probability value is larger than a threshold value (generally 0.5 and can be regulated according to service requirements), judging the traffic as cheating traffic, filtering advertisements, returning a filtering identifier, and otherwise, directly returning the advertisements.
In step S3, the business rule includes:
1. the access times of the same device in a certain time period exceed a certain threshold (can be set according to actual requirements);
2. the same equipment has a plurality of IP addresses (which can be set according to actual requirements);
3. the same equipment exposes a large amount of light at the same time (can be set according to actual requirements);
4. the same device clicks in a large amount at the same time (which can be set according to actual requirements).
Claims (7)
1. An anti-cheating method for advertisement delivery, which is characterized by comprising the following steps:
1) Collecting exposure and click data of advertisements and exposure and click data of users;
2) Preprocessing the data in the step 1) to remove abnormal data;
3) Extracting the characteristics of the data from which the abnormal data are removed, extracting the characteristics and converting the characteristics, and vectorizing the characteristics;
4) Labeling negative samples in the advertisement request by using the vectorized characteristics, and according to positive samples: negative sample = K:1, generating sample data according to the proportion of M: the proportion of N is randomly divided into a training set and a testing set, the training set is input into a machine learning model, and the super parameters of the machine learning model are adjusted according to the testing set to obtain a prediction model;
5) Inputting real-time data and offline data of the advertisement flow into the prediction model to obtain the probability of whether the advertisement flow is the cheating flow, if the probability value is greater than a threshold value, judging the advertisement flow as the cheating flow, filtering the advertisement and returning a filtering identifier; otherwise, directly returning the advertisement;
before step 5), the following treatments are also carried out:
a) Requesting an advertisement from an advertisement delivery engine, wherein the advertisement delivery engine writes the log data of the server into a message middleware kafka;
b) Processing the data in kafka, and counting real-time data;
c) Configuring anti-cheating business rules, if the real-time data violates the business rules, judging the requested advertisement as cheating flow, and writing the equipment number and IP of the accessed equipment into a storage system;
d) Reading blacklist data in a storage system, filtering the equipment number and the IP address of the request, directly filtering advertisements if the equipment number or the IP address is in a blacklist, and returning a filtering identification, wherein the blacklist data comprises blacklist equipment and the IP address; otherwise, step 5) is entered.
2. The method of claim 1, wherein the machine learning model is an LR model or a WideDeep model.
3. An anti-cheating system for advertisement delivery comprises an offline system, an online system and a prediction unit; wherein the offline system comprises:
the data acquisition unit is used for acquiring exposure and click data of advertisements and exposure and click data of users;
the preprocessing unit is used for preprocessing the data acquired by the data acquisition unit, removing abnormal data, extracting the characteristics of the data from which the abnormal data is removed, extracting the characteristics and converting the characteristics, and vectorizing the characteristics;
the training unit is used for labeling negative samples in the advertisement request by utilizing the vectorized characteristics, and according to positive samples: negative sample = K:1, generating sample data according to the proportion of M: the proportion of N is randomly divided into a training set and a testing set, the training set is input into a machine learning model, the super parameters of the machine learning model are regulated according to the testing set, a prediction model is obtained, and the operation of a prediction unit is executed;
the online system includes:
a calculation engine for processing the data in kafka and counting real-time data;
the rule engine configures anti-cheating business rules, if the real-time data violates the business rules, the requested advertisement is judged to be cheating flow, and the equipment number and the IP of the accessed equipment are written into the storage system;
advertisement delivery engine: the method comprises the steps that after an advertisement request is received, log data of a server side are written into a message middleware kafka; reading blacklist data in a storage system, filtering the equipment number and the IP address of the request, directly filtering advertisements if the equipment number or the IP address is in a blacklist, and returning a filtering identification, wherein the blacklist data comprises blacklist equipment and the IP address; otherwise, executing the operation of the prediction unit;
the prediction unit is used for inputting real-time data and offline data of advertisement flow of the offline system and the online system into the prediction model to obtain the probability of whether the advertisement flow is the cheating flow, if the probability value is larger than a threshold value, judging the advertisement flow as the cheating flow, filtering the advertisement and returning a filtering identifier; otherwise, the advertisement is returned directly.
4. The anti-cheating system of advertising according to claim 3, wherein the log data of the server includes a device ID, model, IP, advertisement ID, access time of access.
5. The anti-cheating system of advertising according to claim 3, wherein the business rules comprise: whether the number of times a certain device is accessed for a certain period of time exceeds a set threshold; whether a certain device has a plurality of IP addresses; whether the exposure times of a certain device in the same time period are larger than a set threshold value or not; whether the number of clicks of a certain device in the same time period is larger than a set threshold.
6. The anti-cheating system of advertising according to claim 3, wherein the real-time data comprises a number of times the same device accesses an interface or a number of times the same IP accesses an interface.
7. The anti-cheating system for advertising according to claim 3, wherein the storage system is redis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911116609.4A CN110827094B (en) | 2019-11-15 | 2019-11-15 | Anti-cheating method and system for advertisement delivery |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911116609.4A CN110827094B (en) | 2019-11-15 | 2019-11-15 | Anti-cheating method and system for advertisement delivery |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110827094A CN110827094A (en) | 2020-02-21 |
| CN110827094B true CN110827094B (en) | 2023-05-23 |
Family
ID=69555404
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911116609.4A Active CN110827094B (en) | 2019-11-15 | 2019-11-15 | Anti-cheating method and system for advertisement delivery |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110827094B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111404835B (en) * | 2020-03-30 | 2023-05-30 | 京东科技信息技术有限公司 | Flow control method, device, equipment and storage medium |
| CN111640027B (en) * | 2020-05-28 | 2023-09-12 | 北京金山云网络技术有限公司 | Service data processing method, service processing device and electronic equipment |
| CN111930783A (en) * | 2020-07-09 | 2020-11-13 | 北京皮尔布莱尼软件有限公司 | Monitoring method, monitoring system and computing device |
| CN111882367B (en) * | 2020-08-11 | 2023-11-21 | 北京云和互动信息技术有限公司 | Method for monitoring and tracking online advertisements through analysis of user surfing behavior |
| CN114078016B (en) * | 2020-08-12 | 2023-10-10 | 腾讯科技（深圳）有限公司 | Anti-cheating behavior identification method and device, electronic equipment and storage medium |
| CN112101993B (en) * | 2020-09-11 | 2022-12-23 | 厦门美图之家科技有限公司 | Offline anti-cheating method and device, electronic equipment and readable storage medium |
| CN112188291B (en) * | 2020-09-24 | 2022-11-29 | 北京明略昭辉科技有限公司 | Method and device for identifying advertisement position abnormity |
| CN113032234A (en) * | 2021-03-25 | 2021-06-25 | 北京明略昭辉科技有限公司 | Method and system for identifying abnormal codes of equipment and method for monitoring number of equipment |
| CN113657924B (en) * | 2021-07-21 | 2023-10-31 | 安徽赤兔马传媒科技有限公司 | Online intelligent screen advertisement anti-cheating system and alarm based on machine learning |
| CN114119037B (en) * | 2022-01-24 | 2022-05-17 | 深圳尚米网络技术有限公司 | Marketing anti-cheating system based on big data |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106022834A (en) * | 2016-05-24 | 2016-10-12 | 腾讯科技（深圳）有限公司 | Advertisement against cheating method and device |
| WO2016169193A1 (en) * | 2015-04-24 | 2016-10-27 | 百度在线网络技术（北京）有限公司 | Method and apparatus for detecting cheated clicks |
| CN106156093A (en) * | 2015-04-01 | 2016-11-23 | 阿里巴巴集团控股有限公司 | The recognition methods of ad content and device |
| CN106886915A (en) * | 2017-01-17 | 2017-06-23 | 华南理工大学 | A kind of ad click predictor method based on time decay sampling |
| CN107067274A (en) * | 2016-12-27 | 2017-08-18 | 北京掌阔移动传媒科技有限公司 | One DSP real time bid ad system based on blended learning model |
| CN107330737A (en) * | 2017-07-05 | 2017-11-07 | 北京奇虎科技有限公司 | The anti-performance optimization method and device practised fraud |
| CN108109011A (en) * | 2017-12-28 | 2018-06-01 | 北京皮尔布莱尼软件有限公司 | A kind of anti-cheat method of advertisement and computing device |
| CN109905738A (en) * | 2019-03-26 | 2019-06-18 | 湖南快乐阳光互动娱乐传媒有限公司 | Video advertisement abnormal display monitoring method and device, storage medium and electronic equipment |
| CN110097389A (en) * | 2018-01-31 | 2019-08-06 | 上海甚术网络科技有限公司 | A kind of anti-cheat method of ad traffic |
| CN110335086A (en) * | 2019-07-10 | 2019-10-15 | 西安点告网络科技有限公司 | The anti-method and apparatus practised fraud of advertisement |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150262256A1 (en) * | 2014-03-13 | 2015-09-17 | Microsoft Corporation | Collaborative insights for advertisement viewers |
-
2019
- 2019-11-15 CN CN201911116609.4A patent/CN110827094B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106156093A (en) * | 2015-04-01 | 2016-11-23 | 阿里巴巴集团控股有限公司 | The recognition methods of ad content and device |
| WO2016169193A1 (en) * | 2015-04-24 | 2016-10-27 | 百度在线网络技术（北京）有限公司 | Method and apparatus for detecting cheated clicks |
| CN106022834A (en) * | 2016-05-24 | 2016-10-12 | 腾讯科技（深圳）有限公司 | Advertisement against cheating method and device |
| CN107067274A (en) * | 2016-12-27 | 2017-08-18 | 北京掌阔移动传媒科技有限公司 | One DSP real time bid ad system based on blended learning model |
| CN106886915A (en) * | 2017-01-17 | 2017-06-23 | 华南理工大学 | A kind of ad click predictor method based on time decay sampling |
| CN107330737A (en) * | 2017-07-05 | 2017-11-07 | 北京奇虎科技有限公司 | The anti-performance optimization method and device practised fraud |
| CN108109011A (en) * | 2017-12-28 | 2018-06-01 | 北京皮尔布莱尼软件有限公司 | A kind of anti-cheat method of advertisement and computing device |
| CN110097389A (en) * | 2018-01-31 | 2019-08-06 | 上海甚术网络科技有限公司 | A kind of anti-cheat method of ad traffic |
| CN109905738A (en) * | 2019-03-26 | 2019-06-18 | 湖南快乐阳光互动娱乐传媒有限公司 | Video advertisement abnormal display monitoring method and device, storage medium and electronic equipment |
| CN110335086A (en) * | 2019-07-10 | 2019-10-15 | 西安点告网络科技有限公司 | The anti-method and apparatus practised fraud of advertisement |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110827094A (en) | 2020-02-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110827094B (en) | Anti-cheating method and system for advertisement delivery | |
| CN108881194B (en) | Method and device for detecting abnormal behaviors of users in enterprise | |
| Nguyen et al. | Automatic image filtering on social networks using deep learning and perceptual hashing during crises | |
| CN110263238B (en) | Big data-based public opinion listening system | |
| WO2015172685A1 (en) | Method and apparatus for identifying malicious account | |
| CN113378899B (en) | Abnormal account identification method, device, equipment and storage medium | |
| CN106657007A (en) | Method for recognizing abnormal batch ticket booking behavior based on DBSCAN model | |
| CN107707541A (en) | A kind of attack daily record real-time detection method based on machine learning of streaming | |
| CN109660557A (en) | Attack IP portrait generation method, attack IP portrait generating means and electronic equipment | |
| CN111754241A (en) | User behavior perception method, device, equipment and medium | |
| CN110648172B (en) | Identity recognition method and system integrating multiple mobile devices | |
| CN104156447A (en) | Intelligent social platform advertisement early warning and handling method | |
| CN110401636A (en) | A kind of big data air control method and apparatus for supervising abnormal access | |
| CN109561086A (en) | A method of anti-crawler is carried out using Praxeology | |
| CN111666978B (en) | Intelligent fault early warning system for IT system operation and maintenance big data | |
| CN111431883B (en) | Web attack detection method and device based on access parameters | |
| CN116865994A (en) | Network data security prediction method based on big data | |
| CN115296933A (en) | Industrial production data risk level assessment method and system | |
| CN107944293A (en) | Fictitious assets guard method, system, equipment and storage medium | |
| CN114385668A (en) | Cold data cleaning method, device, equipment and storage medium | |
| CN107871213B (en) | Transaction behavior evaluation method, device, server and storage medium | |
| CN116862243B (en) | Enterprise risk analysis prediction method, system and medium based on neural network | |
| CN111431884B (en) | Host computer defect detection method and device based on DNS analysis | |
| CN113746790A (en) | Abnormal flow management method, electronic device and storage medium | |
| CN114422168A (en) | Malicious machine traffic identification method and system |
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 |