CN108804430A - A kind of SEM launches data sorting system and its sorting technique - Google Patents
A kind of SEM launches data sorting system and its sorting technique Download PDFInfo
- Publication number
- CN108804430A CN108804430A CN201710281438.5A CN201710281438A CN108804430A CN 108804430 A CN108804430 A CN 108804430A CN 201710281438 A CN201710281438 A CN 201710281438A CN 108804430 A CN108804430 A CN 108804430A
- Authority
- CN
- China
- Prior art keywords
- data
- feature
- module
- launches
- sem
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of SEM to launch data sorting system, including database module, and data information is launched for storing；Data launch analog module, for analogue data launch process and collect dispensing feedback information；Characteristic extracting module, for carrying out feature extraction in the feedback information of collection；Feature processing block, for being associated property of the feature processing to extraction；Sort module, for feature to classify to data according to treated；Mark module establishes classification map relationship for sorted data to be marked.The present invention can improve the deficiencies in the prior art, improve the accuracy rate and speed of data classification.
Description
Technical field
The present invention relates to SEM to launch analysis technical field, and especially a kind of SEM launches data sorting system and its classification side
Method.
Background technology
Search engine refers to a kind of information query system based on Internet, including Information Access, information management and
Information retrieval.Search engine marketing is exactly the mode that search engine is used according to consumer, passes through a whole set of technology and strategy
Marketing message is passed to target group by system as far as possible using the chance of user search information.Search engine marketing (SEM) is made
For a kind of Network Marketing Mode, it is therefore intended that promote website, enhance the reputation, obtained more by the result of search engine return
Good sale or channels has more to moulding Network brand, website promotion, online sales and product promotion etc.
Apparent effect.In order to improve the effect of SEM impression informations, need to classify to launching data, existing SEM launches data
Categorizing system is directly compared both for some (or multiple) feature of data, is then classified, this mode classification
It is not only computationally intensive, but also the relevance between data different characteristic can not be excavated, cause classification accuracy low.
Invention content
The technical problem to be solved in the present invention is to provide a kind of SEM to launch data sorting system and its sorting technique, can
The deficiencies in the prior art are solved, the accuracy rate and speed of data classification are improved.
In order to solve the above technical problems, the technical solution used in the present invention is as follows.
A kind of SEM dispensings data sorting system, including,
Database module launches data information for storing；
Data launch analog module, for analogue data launch process and collect dispensing feedback information；
Characteristic extracting module, for carrying out feature extraction in the feedback information of collection；
Feature processing block, for being associated property of the feature processing to extraction；
Sort module, for feature to classify to data according to treated；
Mark module establishes classification map relationship for sorted data to be marked.
A kind of above-mentioned SEM launches the sorting technique of data sorting system, includes the following steps：
A, it will be stored in database module for the data of dispensing, and launches analog module by data and carries out dispensing simulation；
B, characteristic extracting module extracts the feature in feedback information；
C, feature processing block handles the being associated property of feature of extraction；
D, according to treated, feature classifies to data to sort module；
E, classification map relationship is established, and deposit as a result, sorted data are marked in mark module according to classification
Storage is in database module.
Preferably, in the step A, dispensing simulation is carried out using several keywords at random, to the key simulated
Word is recorded in data launch analog module, and is deleted and recorded keyword relevance and wait for that simulation is crucial more than threshold value
Word；For the degree of correlation of keyword on the basis of keyword fields, the degree of correlation is directly proportional to fields degree of overlapping.
Preferably, in step B, the feature of extraction includes showing cost, the amount of showing, click volume, conversion ratio, launching area
Domain, release time, frequency.
Preferably, in step B, the inhomogeneity characteristic value for belonging to same feedback information is extracted, is formed based on anti-
The fisrt feature matrix of feedforward information；The same class characteristic value belonged in different feedback informations is extracted, feature based is formed
The second characteristic matrix of value.
Preferably, in step C, being associated property of the feature processing of extraction is included the following steps,
C1, the transposed matrix for seeking second characteristic matrix successively, by second characteristic matrix transposed matrix institute corresponding with its
The linear space of representative is compared, and the weighted value of this corresponding category feature value of similarity of two linear space is at just
Than；
C2, corresponding characteristic value in fisrt feature matrix is modified using the weighted value of each category feature value；
C3, the feature vector for seeking revised fisrt feature matrix, using non-linear transform function to feature vector group
At vector space carry out nonlinear transformation processing the eigenfunction for the dimension that disappears is returned with reducing vector space dimension
One changes, and is merged into remaining feature vector.
Preferably, in step D, classification is carried out to data and is included the following steps,
D1, projection of the vector space after the corresponding dimensionality reduction of data in each dimension is sought；
D2, classified to data according to the similarity in same dimension upslide shadow.
It is using advantageous effect caused by above-mentioned technical proposal：The present invention can reduce the operation of data assorting process
Amount increases the excavation of the degree of correlation pair different characteristic between, improves the accuracy of data classification, improvement SEM impression informations
Effect.
Description of the drawings
Fig. 1 is the hardware elementary diagram of the present invention.
In figure：1, database module；2, data launch analog module；3, characteristic extracting module；4, feature processing block；5,
Sort module；6, mark module.
Specific implementation mode
Referring to Fig.1, a kind of SEM launches data sorting system, it is characterised in that：Including,
Database module 1 launches data information for storing；
Data launch analog module 2, for analogue data launch process and collect dispensing feedback information；
Characteristic extracting module 3, for carrying out feature extraction in the feedback information of collection；
Feature processing block 4, for being associated property of the feature processing to extraction；
Sort module 5, for feature to classify to data according to treated；
Mark module 6 establishes classification map relationship for sorted data to be marked.
A kind of above-mentioned SEM launches the sorting technique of data sorting system, includes the following steps：
A, it will be stored in database module 1 for the data of dispensing, and launches analog module 2 by data and carries out dispensing mould
It is quasi-；
B, characteristic extracting module 3 extracts the feature in feedback information；
C, feature processing block 4 handles the being associated property of feature of extraction；
D, according to treated, feature classifies to data to sort module 5；
E, classification map relationship is established as a result, sorted data are marked in mark module 6 according to classification, and
It is stored in database module 1.
In step A, dispensing simulation is carried out using several keywords at random, mould is launched in data to the keyword simulated
It is recorded in quasi- module 2, and deletes and recorded the keyword to be simulated that keyword relevance is more than threshold value；The phase of keyword
For Guan Du on the basis of keyword fields, the degree of correlation is directly proportional to fields degree of overlapping.Threshold value is averaged according to keyword
The degree of correlation and dispensing require to be specifically chosen.
In step B, the feature of extraction includes when showing cost, the amount of showing, click volume, conversion ratio, dispensing region, dispensing
Between, frequency.
In step B, the inhomogeneity characteristic value for belonging to same feedback information is extracted, forms the based on feedback information
One eigenmatrix；The same class characteristic value belonged in different feedback informations is extracted, form feature based value second is special
Levy matrix.
In step C, being associated property of the feature processing of extraction is included the following steps,
C1, the transposed matrix for seeking second characteristic matrix successively, by second characteristic matrix transposed matrix institute corresponding with its
The linear space of representative is compared, and the weighted value of this corresponding category feature value of similarity of two linear space is at just
Than；
C2, corresponding characteristic value in fisrt feature matrix is modified using the weighted value of each category feature value；
C3, the feature vector for seeking revised fisrt feature matrix, using non-linear transform function to feature vector group
At vector space carry out nonlinear transformation processing the eigenfunction for the dimension that disappears is returned with reducing vector space dimension
One changes, and is merged into remaining feature vector.
In step D, classification is carried out to data and is included the following steps,
D1, projection of the vector space after the corresponding dimensionality reduction of data in each dimension is sought；
D2, classified to data according to the similarity in same dimension upslide shadow.
Wherein, in step D1, first choice is iterated processing to vector space,
T '=TK
T is vector space, and K is Iterative Matrix, and K is non-non-singular matrix.By iteration, can be effectively increased in vector space
The difference degree of characteristic information and non-characteristic information, convenient for distinguishing.
In step E, secondary classification is carried out to the classification map relationship of foundation, the path counted between different mappings intersects
Point, the classification map relationship that crosspoint quantity is more than to threshold value are classified as one kind.Threshold value is according to the average in crosspoint between mapping
Amount and dispensing require specific determine.By the secondary classification to classification map relationship, can improve for launching data after classification
Retrieval rate.
The present invention is suitable for the classification of the SEM dispensing data of big data quantity, and classification speed is fast, and it is good to launch effect.
In the description of the present invention, it is to be understood that, term " longitudinal direction ", " transverse direction ", "upper", "lower", "front", "rear",
The orientation or positional relationship of the instructions such as "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outside" is based on attached drawing institute
The orientation or positional relationship shown is merely for convenience of the description present invention, does not indicate or imply the indicated device or element must
There must be specific orientation, with specific azimuth configuration and operation, therefore be not considered as limiting the invention.
The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention.The technology of the industry
Personnel are it should be appreciated that the present invention is not limited to the above embodiments, and the above embodiments and description only describe this
The principle of invention, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes
Change and improvement all fall within the protetion scope of the claimed invention.The claimed scope of the invention by appended claims and its
Equivalent thereof.
Claims (7)
1. a kind of SEM launches data sorting system, it is characterised in that：Including,
Database module (1) launches data information for storing；
Data launch analog module (2), for analogue data launch process and collect dispensing feedback information；
Characteristic extracting module (3), for carrying out feature extraction in the feedback information of collection；
Feature processing block (4), for being associated property of the feature processing to extraction；
Sort module (5), for feature to classify to data according to treated；
Mark module (6) establishes classification map relationship for sorted data to be marked.
2. a kind of SEM described in claim 1 launches the sorting technique of data sorting system, it is characterised in that including following step
Suddenly：
A, dispensing mould will be carried out for the data of dispensing deposit database module (1), and by data dispensing analog module (2)
It is quasi-；
B, characteristic extracting module (3) extracts the feature in feedback information；
C, feature processing block (4) handles the being associated property of feature of extraction；
D, according to treated, feature classifies to data to sort module (5)；
E, classification map relationship is established, and deposit as a result, sorted data are marked in mark module (6) according to classification
Storage is in database module (1).
3. SEM according to claim 2 launches the sorting technique of data sorting system, it is characterised in that：In step A, with
Machine carries out dispensing simulation using several keywords, remembers in data launch analog module (2) to the keyword simulated
Record, and delete and recorded the keyword to be simulated that keyword relevance is more than threshold value；The degree of correlation of keyword is with keyword institute
On the basis of category field, the degree of correlation is directly proportional to fields degree of overlapping.
4. SEM according to claim 2 launches the sorting technique of data sorting system, it is characterised in that：In step B, carry
The feature taken includes showing cost, the amount of showing, click volume, conversion ratio, launching region, release time, frequency.
5. SEM according to claim 4 launches the sorting technique of data sorting system, it is characterised in that：It, will in step B
The inhomogeneity characteristic value for belonging to same feedback information extracts, and forms the fisrt feature matrix based on feedback information；It will belong to
Same class characteristic value in different feedback informations extracts, and forms the second characteristic matrix of feature based value.
6. SEM according to claim 5 launches the sorting technique of data sorting system, it is characterised in that：It is right in step C
The being associated property of feature of extraction, which is handled, to be included the following steps,
C1, the transposed matrix for seeking second characteristic matrix successively, representated by second characteristic matrix transposed matrix corresponding with its
Linear space compared, the weighted value of this corresponding category feature value of similarity of two linear space is directly proportional；
C2, corresponding characteristic value in fisrt feature matrix is modified using the weighted value of each category feature value；
C3, the feature vector for seeking revised fisrt feature matrix, form feature vector using non-linear transform function
Vector space carries out nonlinear transformation processing and the eigenfunction for the dimension that disappears is normalized with reducing vector space dimension,
It is merged into remaining feature vector.
7. SEM according to claim 6 launches the sorting technique of data sorting system, it is characterised in that：It is right in step D
Data carry out classification and include the following steps,
D1, projection of the vector space after the corresponding dimensionality reduction of data in each dimension is sought；
D2, classified to data according to the similarity in same dimension upslide shadow.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710281438.5A CN108804430A (en) | 2017-04-26 | 2017-04-26 | A kind of SEM launches data sorting system and its sorting technique |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710281438.5A CN108804430A (en) | 2017-04-26 | 2017-04-26 | A kind of SEM launches data sorting system and its sorting technique |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108804430A true CN108804430A (en) | 2018-11-13 |
Family
ID=64068897
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710281438.5A Pending CN108804430A (en) | 2017-04-26 | 2017-04-26 | A kind of SEM launches data sorting system and its sorting technique |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108804430A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113342804A (en) * | 2021-03-06 | 2021-09-03 | 广东信通通信有限公司 | Big data-based dissociative data tagged reutilization method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100055677A1 (en) * | 2007-01-04 | 2010-03-04 | The Regents Of The University Of California | Method for genetic identification of unknown organisms |
| CN102508859A (en) * | 2011-09-29 | 2012-06-20 | 北京亿赞普网络技术有限公司 | Advertisement classification method and device based on webpage characteristic |
| CN103295150A (en) * | 2013-05-20 | 2013-09-11 | 厦门告之告信息技术有限公司 | Advertising release system and advertising release method capable of accurately quantizing and counting release effects |
| CN103365900A (en) * | 2012-04-01 | 2013-10-23 | 阿里巴巴集团控股有限公司 | Method and device for throwing on-line material |
| CN103377249A (en) * | 2012-04-27 | 2013-10-30 | 纽海信息技术（上海）有限公司 | Keyword putting method and system |
| CN105431875A (en) * | 2013-08-05 | 2016-03-23 | 雅虎公司 | Search engine marketing optimizer |
-
2017
- 2017-04-26 CN CN201710281438.5A patent/CN108804430A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100055677A1 (en) * | 2007-01-04 | 2010-03-04 | The Regents Of The University Of California | Method for genetic identification of unknown organisms |
| CN102508859A (en) * | 2011-09-29 | 2012-06-20 | 北京亿赞普网络技术有限公司 | Advertisement classification method and device based on webpage characteristic |
| CN103365900A (en) * | 2012-04-01 | 2013-10-23 | 阿里巴巴集团控股有限公司 | Method and device for throwing on-line material |
| CN103377249A (en) * | 2012-04-27 | 2013-10-30 | 纽海信息技术（上海）有限公司 | Keyword putting method and system |
| CN103295150A (en) * | 2013-05-20 | 2013-09-11 | 厦门告之告信息技术有限公司 | Advertising release system and advertising release method capable of accurately quantizing and counting release effects |
| CN105431875A (en) * | 2013-08-05 | 2016-03-23 | 雅虎公司 | Search engine marketing optimizer |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113342804A (en) * | 2021-03-06 | 2021-09-03 | 广东信通通信有限公司 | Big data-based dissociative data tagged reutilization method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107577688B (en) | Original article influence analysis system based on media information acquisition | |
| Karim et al. | Decision tree and naive bayes algorithm for classification and generation of actionable knowledge for direct marketing | |
| CN106649455A (en) | Big data development standardized systematic classification and command set system | |
| CN104573130B (en) | The entity resolution method and device calculated based on colony | |
| CN110909164A (en) | Text enhancement semantic classification method and system based on convolutional neural network | |
| CN107609052A (en) | A kind of generation method and device of the domain knowledge collection of illustrative plates based on semantic triangle | |
| CN105719191B (en) | The social groups that behavior semanteme is not known under multiscale space find method | |
| CN112463971B (en) | E-commerce commodity classification method and system based on hierarchical combination model | |
| CN109635118A (en) | A kind of user's searching and matching method based on big data | |
| CN105824959A (en) | Public opinion monitoring method and system | |
| CN104834693A (en) | Depth-search-based visual image searching method and system thereof | |
| CN110569904B (en) | Method for constructing machine learning model and computer-readable storage medium | |
| CN110188346A (en) | A kind of network security bill part intelligence analysis method based on information extraction | |
| CN110008306A (en) | A kind of data relationship analysis method, device and data service system | |
| CN103034656B (en) | Chapters and sections content layered approach and device, article content layered approach and device | |
| CN107169106A (en) | Video retrieval method, device, storage medium and processor | |
| CN106446124B (en) | A kind of Website classification method based on cyberrelationship figure | |
| CN108763496A (en) | A kind of sound state data fusion client segmentation algorithm based on grid and density | |
| Islam et al. | Review analysis of ride-sharing applications using machine learning approaches: Bangladesh perspective | |
| CN103761286B (en) | A kind of Service Source search method based on user interest | |
| López-Cifuentes et al. | Attention-based knowledge distillation in scene recognition: the impact of a dct-driven loss | |
| Bitarafan et al. | Spgd_hin: Spammer group detection based on heterogeneous information network | |
| CN108804430A (en) | A kind of SEM launches data sorting system and its sorting technique | |
| CN110674265B (en) | Unstructured information oriented feature discrimination and information recommendation system | |
| CN109739840A (en) | Data processing empty value method, apparatus and terminal device |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181113 |
|
| WD01 | Invention patent application deemed withdrawn after publication |