CN106528777A - Cross-screen user identification normalizing method and system - Google Patents
Cross-screen user identification normalizing method and system Download PDFInfo
- Publication number
- CN106528777A CN106528777A CN201610977040.0A CN201610977040A CN106528777A CN 106528777 A CN106528777 A CN 106528777A CN 201610977040 A CN201610977040 A CN 201610977040A CN 106528777 A CN106528777 A CN 106528777A
- Authority
- CN
- China
- Prior art keywords
- terminal unit
- data
- information
- terminal devices
- predicted
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
Abstract
The invention discloses a cross-screen user identification normalizing method and system. The method comprises the steps of for a plurality of terminal devices, obtaining behavior attribute information corresponding to each terminal device from a pre-established database according to identification information of each terminal device; determining a plurality of candidate sets according to behavior attribute information of the terminal device, wherein each candidate set comprises a plurality of related terminal devices; extracting feature parameters of the plurality of candidate sets, taking the feature parameters as training data and establishing a classification model; and determining classification information of to-be-predicted data according to the classification model and generating uniform identification information for the terminal devices of the same class. Through application of the method and the system, the related terminal devices can be determined effectively.
Description
Technical field
The present invention relates to Internet technical field, more particularly to a kind of method and its system across screen ID normalizing.
Background technology
In internet, applications field, user behavior analysis are referred in statistics and analysis user access network service process
(include accessing and browse webpage, interact formula operation, using APP etc.) real-time that produces and historical user behavior letter
Breath.In the middle of the action process of user access network service, comprising a large amount of valuable information.Wherein, user behavior information
Including but not limited to herein below：The access times of network service, visiting frequency, access the time of staying, operation active time,
User input key word, user clicks on links, user interactive (such as pay attention in, cancel concern, give a mark, save as bookmark, add
Enter shopping cart, take out shopping cart, formed order, cancel an order, pay the bill, reimbursement etc.).
By the research to user behavior information, the rule that user is showed when access network is serviced can be therefrom found
Rule distribution, and to be lifted, Consumer's Experience, efficient information are pushed and promotion target marketing provides science, accurate objective basis.
Big data technology is with the total data resource of any system as object and therefrom finds the phase showed between data
The information processing technology of sexual intercourse is closed, process optimization, targeted message and the advertisement for having been widely used for the Internet at present is pushed away
Give, user individual service with the aspect such as improve, become the powerful background support in network service behind.
Carry out in the middle of the prior art of big data statistics, analysis with application in user oriented behavioural information, be by each
User behavior and the content as object of action all serve only as an isolated data point, so by collecting whole user behaviors
And its mass data point formed by object, therefrom find Statistical Distribution.The defect of this pattern is can not to find and profit
With relatedness mutual between different user behaviors and its object of action.
The content of the invention
Present invention is primarily targeted at a kind of method and its system across screen ID normalizing is provided, it is existing to solve
The problems referred to above of technology, wherein：
A kind of method across screen ID normalizing is proposed according to embodiments of the present invention, and which includes：
For multiple terminal devices, obtained from the data base for pre-building according to the identification information of each terminal unit respectively
Take the corresponding behavior property information of each terminal unit；
Multiple candidate collections are determined according to the behavior property information of terminal unit, wherein each Candidate Set includes associated
Multiple terminal devices；
The characteristic parameter of the plurality of candidate collection is extracted as training data and disaggregated model is built；
The classification information of data to be predicted is determined according to the disaggregated model, and the terminal unit of same classification is generated into system
One identification information.
Wherein, the associated plurality of terminal unit includes：The terminal unit logged in using same IP address.
Wherein, the characteristic parameter includes one below or its combination：Terminal equipment type, OS Type, application
Program information, the information of terminal unit access website, the co-occurrence behavior characteristicss of multiple terminal devices, the co-occurrence of multiple terminal devices
The potential interest matching degree feature of scene characteristic, multiple terminal devices.
Wherein, the structure disaggregated model includes：
Extract positive sample data and negative sample data from the characteristic parameter respectively, trained according to sample data
Data and data to be predicted.
Wherein, determine that the classification information of data to be predicted includes according to the disaggregated model：
Determine the associated probability of the corresponding multiple terminal devices of data to be predicted.
A kind of system across screen ID normalizing is proposed according to embodiments of the present invention also, which includes：
Acquisition module, for for multiple terminal devices, being built from advance according to the identification information of each terminal unit respectively
The corresponding behavior property information of each terminal unit is obtained in vertical data base；
Data extraction module, determines multiple candidate collections for the behavior property information according to terminal unit, wherein each
Candidate Set includes associated plurality of terminal unit；
Disaggregated model builds module, for extracting the characteristic parameter of the plurality of candidate collection as training data and building
Disaggregated model；
User's sort module, for the classification information of data to be predicted is determined according to the disaggregated model, and by same point
The terminal unit of class generates unified identification information.
Wherein, the associated plurality of terminal unit includes：The terminal unit logged in using same IP address.
Wherein, the characteristic parameter includes one below or its combination：Terminal equipment type, OS Type, application
Program information, the information of terminal unit access website, the co-occurrence behavior characteristicss of multiple terminal devices, the co-occurrence of multiple terminal devices
The potential interest matching degree feature of scene characteristic, multiple terminal devices.
Wherein, the disaggregated model builds module and is additionally operable to, and extracts positive sample data from the characteristic parameter respectively
With negative sample data, training data and data to be predicted are obtained according to sample data.
Wherein, user's sort module is additionally operable to, and determines the associated of the corresponding multiple terminal devices of data to be predicted
Probability.
Technology according to the present invention scheme, the user logged in using distinct device by analysis user behavior recognition, and really
Fixed associated terminal unit such that it is able to by unified angle analysis and the behavioural information of digging user.
Description of the drawings
Accompanying drawing described herein is used for providing a further understanding of the present invention, constitutes the part of the application, this
Bright schematic description and description does not constitute inappropriate limitation of the present invention for explaining the present invention.In the accompanying drawings：
Fig. 1 is the flow chart of the method across screen ID normalizing according to embodiments of the present invention；
Fig. 2 is the structured flowchart of the system across screen ID normalizing according to embodiments of the present invention.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with drawings and the specific embodiments, to this
Invention is described in further detail.
A kind of embodiments in accordance with the present invention, there is provided method across screen ID normalizing.
Fig. 1 is the flow chart of the method across screen ID normalizing according to embodiments of the present invention, as shown in figure 1, described
Method at least includes (step S102-S108)：
Step S102, for multiple terminal devices, respectively according to the identification information of each terminal unit from pre-building
The corresponding behavior property information of each terminal unit is obtained in data base.
In the present embodiment, the initial data to being obtained by data base such as is filtered, extracted and is changed at the associative operation,
Effective ID and its corresponding effective attribute and behavioural information are obtained, extraction is characterized and data basis is provided.Data are pre-
Process is mainly divided into following two steps：Relation extraction and object extraction.
Relation extraction mainly gets rid of useless field information from initial data, retains to user's classification (id draws logical)
Helpful field, and the filtration on some bases is done, production Methods table.Effective ID is mainly stored in relation table, is set
Standby attribute, ip address informations, behavioural information (such as timestamp, url of access etc.).Field in relation table has done basic testing
Card and filter operation (such as length checking, character set checking, pattern match etc.), can also enter according to the specification of field collection in addition
Row logic verify.
Object extraction mainly carries out secondary filter to various marks in relation table, and to its event number, go tuple etc. to enter
Row statistics.Secondary filter is to prevent some object occurrence number exceptions, filters out statistical noise data.Such as by statistics
It can be found that some ip belong to public ip, to the logical effect of follow-up drawing not substantially, such that it is able to filter out；Can be sent out by statistics
Existing certain user's behavior number of times is only once, then it can be assumed that this ID is an invalid ID, so as to
Can filter out.
Step S104, determines multiple candidate collections according to the behavior property information of terminal unit, wherein in each Candidate Set
Including associated plurality of terminal unit.
Latent structure and extraction launch around the scene of two ID and co-occurrence.Such as at (D1) on the same day, two
Individual internet user identification U1 and U2, had behavior in same IP address (IP1), then two the fact that all similar
The set that individual ID is constituted, can be used as the logical candidate collection of drawing.For each sample in Candidate Set, can be multiple
Dimension carries out the construction of feature and extraction.Specifically, characteristic parameter can include following several classes：
1st, the static attribute feature of unique user mark；
(1) user uses device type, level of resolution, OS Type；
Device type is such as：Mobile phone, flat board, PC etc..Operating system such as iPhone, Android, Windows,
Linux etc..
(2) user uses software category, version number；
Different browsers and its version number that such as computer or surfing Internet with cell phone are used, the app used on mobile device
Title etc..
2nd, the dynamic attribute feature of unique user mark；
(1) user changes the frequency of ip；
(2) user's browsing on each website/app/use time distribution；
(3) user browses most articles, classification, brand etc.；
(4) time period of user Jing often online.
3rd, the co-occurrence behavior characteristicss of ID pair；
Ip number of (1) two ID co-occurrence；
When there is ip co-occurrences in (2) two ID, the Mean Time Between Replacement of user behavior；
The number of the domain name that (3) two ID co-browses are crossed；
Article that (4) two ID co-browses are crossed, classification, the number of brand.
4th, the co-occurrence scene characteristic of ID pair；
(1) on the ip of two ID co-occurrences, the ID total quantity for occurring；
(2) on the ip of two ID co-occurrences, the total event number for occurring；
The time attribute (time period on the same day, what day etc.) of (3) two ID co-occurrences.
5th, the potential interest matching degree feature of ID pair.
Based on the browsed domain name of ID, article, classification, branding data, using the method for deep learning, used
Matching degree feature in the potential interest dimension at family.
Step S106, extract the plurality of candidate collection characteristic parameter is as training data and builds disaggregated model.
If there is similarity or relatedness between the behavior that a user is produced under multiple ID, then then
Select the general character between different user mark or sexual behaviour etc. is associated as the feature of input, and obtained using GBDT Algorithm for Training
To model, it is eventually used for predicting that two ID draw logical probability.
As behavior of the same user in different user mark has general character or relatedness, with IP co-occurrences (i.e. not
Occur under same IP address with ID) as the basis of training data.Based on why selecting this feature it is
There is general character between these ID as different user mark in same IP address occurs both reflecting, and due to IP
Address distributed by operator and non-user can be controlled thus as an objective reality it for drawing logical object one have by force
The effect of contraction of power.Such as two ID traditionally may be closely similar in behavior：Same type website in preference, accesses
Cross same commodity etc..But due to their IP address, possible one is located at one, Beijing positioned at Shanghai, then be not considered as that they are
Same person.
In actual applications, using the data of IP co-occurrences and the positive sample that can accurately draw the common factor of logical data as training set
This, and negative sample is extracted according to certain ratio, ultimately form the input of model training.The power of each feature can be obtained after training
Weight values, can be predicted accordingly and be calculated the logical probability of two or more ID drawings.
Step S108, determines the classification information of data to be predicted according to the disaggregated model, and by the terminal of same classification
Equipment generates unified identification information.
ID is the mark in electronic digit aspect, on the one hand, the result of model prediction is two ID
The logical result of drawing, in fact, the ID that possesses of real user may many more than two, need to collect one by clustering
All ID of individual real user；On the other hand, there is an ID while occurring in multiple equipment, an equipment
There is the practical situation of multiple ID, cause to draw logical result that the feelings that the multipair Duola of similar community relations figure is led to occur
Condition so that draw logical result divergingization；The all ID for truly collecting a real user to greatest extent are clusters
Target.
In the application scenarios across screen ID normalizing, ID one by one is like personal in community；
Relation between ID has pooled " the community relations figure " of an ID, all user's marks of a real user
Knowledge is this " community relations figure " " social circle ", and this is the theoretical basiss that cluster result is adjusted with community discovery algorithm.Mutually
Community relations figure under networking background describes graph of a relation between Internet user, or perhaps a kind of model of social networkies
Statement.How social circle is described with real social circle one by one in excavating in community relations figure.
Community discovery algorithm is the general designation of the class algorithm that real social circle how is excavated from community relations figure, figure
Segmentation is the core content of community discovery algorithm, however, segmentation necessarily brings the loss on side, while how to weigh what segmentation brought
Loss and the effect weighed after segmentation are difficult points.The application uses the thought of modularity clusters (segmentation).
The modularity degree of measurement figure segmentation is set so as to obtain two points of results of optimum of a figure by structure, is then compared
Compared with the modularity of artwork and two points of results so as to differentiate whether figure needs to continue to divide, optimal point of a figure is finally obtained
Cut.
Modularity degree Q：
Wherein：N represents the nodes of figure, and m represents the side number of figure, and Aij represents the side number between node i and node j, ki tables
Show the side number (out-degree) of node i, si represents 1 or -1 (1 and -1 represents two points of results respectively),
Q can be write as following form：
Wherein：
So as to a figure segmentation problem is switched to mathematical problem (singular value decomposition) successfully：
The critical point of figure segmentation, the Q after segmentation is without more excellent：
So far, " the community relations figure " of a connection is divided into several " social circles " by optimum.Each " social circle "
Interior ID is mutually " to draw logical ", that is, be considered as same user.
Embodiments in accordance with the present invention, additionally provide a kind of system across screen ID normalizing.
Fig. 2 is the structured flowchart of the system across screen ID normalizing according to embodiments of the present invention, as shown in Fig. 2 should
System includes：
Acquisition module 21, for for multiple terminal devices, respectively according to the identification information of each terminal unit from advance
The corresponding behavior property information of each terminal unit is obtained in the data base of foundation；
Data extraction module 22, determines multiple candidate collections for the behavior property information according to terminal unit, wherein often
Individual Candidate Set includes associated plurality of terminal unit；
Disaggregated model builds module 23, for extracting the characteristic parameter of the plurality of candidate collection as training data structure
Build disaggregated model；
User's sort module 24, for the classification information of data to be predicted is determined according to the disaggregated model, and will be same
The terminal unit of classification generates unified identification information.
Wherein, the associated plurality of terminal unit includes：The terminal unit logged in using same IP address.
Wherein, the characteristic parameter includes one below or its combination：Terminal equipment type, OS Type, application
Program information, the information of terminal unit access website, the co-occurrence behavior characteristicss of multiple terminal devices, the co-occurrence of multiple terminal devices
The potential interest matching degree feature of scene characteristic, multiple terminal devices.
Further, the disaggregated model builds module 23 and is additionally operable to, and extracts positive sample from the characteristic parameter respectively
Notebook data and negative sample data, obtain training data and data to be predicted according to sample data.
Further, user's sort module 24 is additionally operable to, and determines the corresponding multiple terminal devices of data to be predicted
Associated probability.
The operating procedure of the method for the present invention is corresponding with the architectural feature of system, no longer can be repeated one by one with cross-referenced.
In sum, technology according to the present invention scheme, by analyzing what user behavior recognition was logged in using distinct device
User, and determine associated terminal unit such that it is able to by unified angle analysis and the behavioural information of digging user.
Embodiments of the invention are the foregoing is only, the present invention is not limited to, for those skilled in the art
For member, the present invention can have various modifications and variations.All any modifications within the spirit and principles in the present invention, made,
Equivalent, improvement etc., should be included within scope of the presently claimed invention.
Claims (10)
1. it is a kind of across the method for shielding ID normalizing, it is characterised in that to include：
For multiple terminal devices, obtained from the data base for pre-building often according to the identification information of each terminal unit respectively
The corresponding behavior property information of individual terminal unit；
Multiple candidate collections are determined according to the behavior property information of terminal unit, wherein each Candidate Set includes associated many
Individual terminal unit；
The characteristic parameter of the plurality of candidate collection is extracted as training data and disaggregated model is built；
The classification information of data to be predicted is determined according to the disaggregated model, and the terminal unit of same classification is generated into unification
Identification information.
2. method according to claim 1, it is characterised in that the associated plurality of terminal unit includes：Using same
The terminal unit that one IP address is logged in.
3. method according to claim 1, it is characterised in that the characteristic parameter includes one below or its combination：Eventually
End equipment type, OS Type, application information, terminal unit access the information of website, multiple terminal devices and are total to
It is existing be characterized, the potential interest matching degree feature of the co-occurrence scene characteristic of multiple terminal devices, multiple terminal devices.
4. method according to claim 1, it is characterised in that the structure disaggregated model includes：
Extract positive sample data and negative sample data from the characteristic parameter respectively, training data is obtained according to sample data
With data to be predicted.
5. method according to claim 1, it is characterised in that the classification of data to be predicted is determined according to the disaggregated model
Information includes：
Determine the associated probability of the corresponding multiple terminal devices of data to be predicted.
6. it is a kind of across the system for shielding ID normalizing, it is characterised in that to include：
Acquisition module, for for multiple terminal devices, respectively according to the identification information of each terminal unit from pre-building
The corresponding behavior property information of each terminal unit is obtained in data base；
Data extraction module, determines multiple candidate collections for the behavior property information according to terminal unit, wherein each candidate
Concentration includes associated plurality of terminal unit；
Disaggregated model builds module, for extracting the characteristic parameter of the plurality of candidate collection as training data and building classification
Model；
User's sort module, for determining the classification information of data to be predicted according to the disaggregated model, and by same classification
Terminal unit generates unified identification information.
7. system according to claim 6, it is characterised in that the associated plurality of terminal unit includes：Using same
The terminal unit that one IP address is logged in.
8. system according to claim 6, it is characterised in that the characteristic parameter includes one below or its combination：Eventually
End equipment type, OS Type, application information, terminal unit access the information of website, multiple terminal devices and are total to
It is existing be characterized, the potential interest matching degree feature of the co-occurrence scene characteristic of multiple terminal devices, multiple terminal devices.
9. system according to claim 6, it is characterised in that the disaggregated model builds module and is additionally operable to, from the spy
Extract positive sample data and negative sample data in levying parameter respectively, training data and number to be predicted are obtained according to sample data
According to.
10. system according to claim 6, it is characterised in that user's sort module is additionally operable to, and determines number to be predicted
According to the associated probability of corresponding multiple terminal devices.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610977040.0A CN106528777A (en) | 2016-10-27 | 2016-10-27 | Cross-screen user identification normalizing method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610977040.0A CN106528777A (en) | 2016-10-27 | 2016-10-27 | Cross-screen user identification normalizing method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106528777A true CN106528777A (en) | 2017-03-22 |
Family
ID=58350326
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610977040.0A Pending CN106528777A (en) | 2016-10-27 | 2016-10-27 | Cross-screen user identification normalizing method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106528777A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105373614A (en) * | 2015-11-24 | 2016-03-02 | 中国科学院深圳先进技术研究院 | Sub-user identification method and system based on user account |
| CN107330091A (en) * | 2017-07-04 | 2017-11-07 | 百度在线网络技术（北京）有限公司 | Information processing method and device |
| CN107515915A (en) * | 2017-08-18 | 2017-12-26 | 晶赞广告（上海）有限公司 | User based on user behavior data identifies correlating method |
| CN107741899A (en) * | 2017-10-16 | 2018-02-27 | 北京小米移动软件有限公司 | The method, apparatus and system of processing terminal data |
| CN108122271A (en) * | 2017-12-15 | 2018-06-05 | 南京变量信息科技有限公司 | A kind of description photo automatic generation method and device |
| CN108830052A (en) * | 2018-05-25 | 2018-11-16 | 恒安嘉新（北京）科技股份公司 | A kind of striding equipment Internet user's recognition methods based on AI |
| CN111080349A (en) * | 2019-12-04 | 2020-04-28 | 北京悠易网际科技发展有限公司 | Method, apparatus, server and medium for identifying multiple devices of same user |
| CN111340062A (en) * | 2020-02-04 | 2020-06-26 | 恩亿科(北京)数据科技有限公司 | Mapping relation determining method and device |
| CN112601215A (en) * | 2020-12-01 | 2021-04-02 | 深圳市和讯华谷信息技术有限公司 | Method and device for unifying equipment identifications |
| CN113744030A (en) * | 2021-09-08 | 2021-12-03 | 未鲲(上海)科技服务有限公司 | Recommendation method, device, server and medium based on AI user portrait |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103634275A (en) * | 2012-08-21 | 2014-03-12 | 联想(北京)有限公司 | Multi-device management and task migration methods and management server |
| CN105608179A (en) * | 2015-12-22 | 2016-05-25 | 百度在线网络技术（北京）有限公司 | Method and device for determining relevance of user identification |
| CN105677844A (en) * | 2016-01-06 | 2016-06-15 | 北京摩比万思科技有限公司 | Mobile advertisement big data directional pushing and user cross-screen recognition method |
-
2016
- 2016-10-27 CN CN201610977040.0A patent/CN106528777A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103634275A (en) * | 2012-08-21 | 2014-03-12 | 联想(北京)有限公司 | Multi-device management and task migration methods and management server |
| CN105608179A (en) * | 2015-12-22 | 2016-05-25 | 百度在线网络技术（北京）有限公司 | Method and device for determining relevance of user identification |
| CN105677844A (en) * | 2016-01-06 | 2016-06-15 | 北京摩比万思科技有限公司 | Mobile advertisement big data directional pushing and user cross-screen recognition method |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105373614B (en) * | 2015-11-24 | 2018-09-28 | 中国科学院深圳先进技术研究院 | A kind of child user recognition methods and system based on user account |
| CN105373614A (en) * | 2015-11-24 | 2016-03-02 | 中国科学院深圳先进技术研究院 | Sub-user identification method and system based on user account |
| CN107330091A (en) * | 2017-07-04 | 2017-11-07 | 百度在线网络技术（北京）有限公司 | Information processing method and device |
| US11244153B2 (en) | 2017-07-04 | 2022-02-08 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for processing information |
| CN107515915A (en) * | 2017-08-18 | 2017-12-26 | 晶赞广告（上海）有限公司 | User based on user behavior data identifies correlating method |
| CN107515915B (en) * | 2017-08-18 | 2020-02-18 | 晶赞广告（上海）有限公司 | User identification association method based on user behavior data |
| CN107741899B (en) * | 2017-10-16 | 2021-07-27 | 北京小米移动软件有限公司 | Method, device and system for processing terminal data |
| CN107741899A (en) * | 2017-10-16 | 2018-02-27 | 北京小米移动软件有限公司 | The method, apparatus and system of processing terminal data |
| CN108122271A (en) * | 2017-12-15 | 2018-06-05 | 南京变量信息科技有限公司 | A kind of description photo automatic generation method and device |
| CN108830052A (en) * | 2018-05-25 | 2018-11-16 | 恒安嘉新（北京）科技股份公司 | A kind of striding equipment Internet user's recognition methods based on AI |
| CN111080349A (en) * | 2019-12-04 | 2020-04-28 | 北京悠易网际科技发展有限公司 | Method, apparatus, server and medium for identifying multiple devices of same user |
| CN111080349B (en) * | 2019-12-04 | 2023-04-21 | 北京悠易网际科技发展有限公司 | Method, device, server and medium for identifying multiple devices of same user |
| CN111340062A (en) * | 2020-02-04 | 2020-06-26 | 恩亿科(北京)数据科技有限公司 | Mapping relation determining method and device |
| CN112601215A (en) * | 2020-12-01 | 2021-04-02 | 深圳市和讯华谷信息技术有限公司 | Method and device for unifying equipment identifications |
| CN113744030A (en) * | 2021-09-08 | 2021-12-03 | 未鲲(上海)科技服务有限公司 | Recommendation method, device, server and medium based on AI user portrait |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106528777A (en) | Cross-screen user identification normalizing method and system | |
| CN103218431B (en) | A kind ofly can identify the system that info web gathers automatically | |
| Li et al. | Community detection using hierarchical clustering based on edge-weighted similarity in cloud environment | |
| CN107862022B (en) | Culture resource recommendation system | |
| CN104899508B (en) | A kind of multistage detection method for phishing site and system | |
| CN106383887A (en) | Environment-friendly news data acquisition and recommendation display method and system | |
| CN103248677B (en) | The Internet behavioural analysis system and method for work thereof | |
| WO2017084205A1 (en) | Network user identity authentication method and system | |
| CN103838754A (en) | Information searching device and method | |
| Ha et al. | An analysis on information diffusion through BlogCast in a blogosphere | |
| CN106446124A (en) | Website classification method based on network relation graph | |
| CN111259220A (en) | Data acquisition method and system based on big data | |
| CN111460803B (en) | Equipment identification method based on Web management page of industrial Internet of things equipment | |
| CN106844588A (en) | A kind of analysis method and system of the user behavior data based on web crawlers | |
| CN105159898B (en) | A kind of method and apparatus of search | |
| CN111984787A (en) | Public opinion hotspot obtaining method and system based on internet data | |
| Xu | Cultural communication in double-layer coupling social network based on association rules in big data | |
| CN110717089A (en) | User behavior analysis system and method based on weblog | |
| CN104991920A (en) | Label generation method and apparatus | |
| CN115757963A (en) | User behavior image drawing method based on distributed log analysis | |
| CN115051863A (en) | Abnormal flow detection method and device, electronic equipment and readable storage medium | |
| CN104077366B (en) | A kind of method and apparatus for being used to determine characteristic information in the network device | |
| CN113688346A (en) | Illegal website identification method, device, equipment and storage medium | |
| Li et al. | High-accuracy model recognition method of mobile device based on weighted feature similarity | |
| CN106294292A (en) | Chapters and sections catalogue screening technique and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170322 |
|
| RJ01 | Rejection of invention patent application after publication |