CN104484372A - Detecting method and device of business object sending information - Google Patents

Abstract

The invention provides a detecting method and device of business object sending information. The detecting method includes generating a detecting model of business object sending information on the basis of attribute information of several senders historically uploading the business object sending information; receiving detecting requests carrying sender identifications; detecting the business object sending information uploaded by the senders corresponding to the sender identifications. The detecting model of the business object sensing information is generated by analyzing a great deal of historical information, thereby being capable of more objectively and accurately detecting the business object sending information uploaded by the senders. Further, extra analysis and statistics of the sending effect of the senders are omitted during detection of the detecting model, detecting is simpler and more convenient, and detecting efficiency is improved.

Description

A kind of detection method of business object impression information and device

Technical field

The present invention relates to networking technology area, be specifically related to a kind of detection method and device of business object impression information.

Background technology

Along with the development of internet, the network user gets more and more, therefore business object is rendered on Internet service object release platform by putting person more and more, so that business object is recommended these network users, makes user can recognize up-to-date information in real time.

Putting person is by buying keyword and bidding to participate in the competition with covering flow as much as possible, increase click volume, the demand such as structure, region, time according to business formulates business object impression information, and this business object impression information is uploaded to business object release platform, business object release platform carries out the input of business object according to this business object impression information.

The good and bad situation of business object impression information can have direct impact to the covering flow, click volume etc. of putting person, if business object impression information is second-rate, then putting person will be caused to throw in the effect of business object poor, affect income.Particularly for the putting person of some lack of experience, because it is lacked experience to formulation business object impression information and the keyword aspect such as to bid, bid ranking situation rearward or too high bidding therefore usually is caused to upset the situation of bid ranking environment.Therefore, if the business object impression information can uploaded putting person detects, then there is very large meaning to reminding putting person's business object impression information Problems existing and formulating the business object impression information more optimized.

The method that the business object impression information uploaded putting person at present detects, mainly the independent input of the history to the business object impression information that this putting person uploads effect is analyzed, and generate corresponding analysis report, as analyzed the history input effect of different geographical, different interest, different keyword respectively, thus generate region report, interest report, keyword report etc.

But adopt above-mentioned separately to the business object impression information detection mode that the history input effect of the business object impression information that this putting person uploads is analyzed, testing result is inaccurate, and detection efficiency is lower, cannot meet the demand of user.

Summary of the invention

In view of the above problems, the present invention is proposed to provide a kind of overcoming the problems referred to above or the detection method of business object impression information solved the problem at least in part and the pick-up unit of corresponding business object impression information.

According to one aspect of the present invention, provide a kind of detection method of business object impression information, comprising:

Upload the attribute information of the putting person of business object impression information in advance based on multiple history, generate the detection model of business object impression information;

Receive the detection request carrying putting person's mark;

According to the detection model of described business object impression information, the business object impression information that corresponding putting person uploads is identified to described putting person and detects.

According to a further aspect in the invention, provide a kind of pick-up unit of business object impression information, comprising:

Generation module, is suitable for the attribute information of the putting person uploading business object impression information in advance based on multiple history, generates the detection model of business object impression information;

Receiver module, is suitable for receiving the detection request carrying putting person's mark;

Detection module, is suitable for the detection model according to described business object impression information, identifies the business object impression information that corresponding putting person uploads detect described putting person.

According to the detection scheme of business object impression information of the present invention, the attribute information of the putting person of business object impression information is uploaded in advance based on multiple history, generate the detection model of business object impression information, follow-up after receiving the detection request carrying putting person's mark, can detect the business object impression information that the putting person that putting person identifies correspondence uploads according to the detection model of the business object impression information generated.The detection model of above-mentioned business object impression information generates in conjunction with a large amount of historical information analyses, therefore, it is possible to more objective, exactly the business object impression information that putting person uploads is detected, and without the need to carrying out analytic statistics to the input effect of the putting person detected again when detecting according to this detection model, testing process is easier, improves detection efficiency.

Above-mentioned explanation is only the general introduction of technical solution of the present invention, in order to technological means of the present invention can be better understood, and can be implemented according to the content of instructions, and can become apparent, below especially exemplified by the specific embodiment of the present invention to allow above and other objects of the present invention, feature and advantage.

Accompanying drawing explanation

By reading hereafter detailed description of the preferred embodiment, various other advantage and benefit will become cheer and bright for those of ordinary skill in the art.Accompanying drawing only for illustrating the object of preferred implementation, and does not think limitation of the present invention.And in whole accompanying drawing, represent identical parts by identical reference symbol.In the accompanying drawings:

Fig. 1 shows the flow chart of steps of the detection method of a kind of business object impression information in the embodiment of the present invention one;

Fig. 2 shows the flow chart of steps of the detection method of a kind of business object impression information in the embodiment of the present invention two;

Fig. 3 shows the structured flowchart of the pick-up unit of a kind of business object impression information in the embodiment of the present invention three;

Fig. 4 shows the structured flowchart of the pick-up unit of a kind of business object impression information in the embodiment of the present invention four.

Embodiment

Below with reference to accompanying drawings exemplary embodiment of the present disclosure is described in more detail.Although show exemplary embodiment of the present disclosure in accompanying drawing, however should be appreciated that can realize the disclosure in a variety of manners and not should limit by the embodiment set forth here.On the contrary, provide these embodiments to be in order to more thoroughly the disclosure can be understood, and complete for the scope of the present disclosure can be conveyed to those skilled in the art.

Embodiment one:

With reference to Fig. 1, show the flow chart of steps of the detection method of a kind of business object impression information in the embodiment of the present invention one.The method can comprise the following steps:

Step 100, uploads the attribute information of the putting person of business object impression information in advance based on multiple history, generate the detection model of business object impression information.

In the embodiment of the present invention, first the attribute information of the putting person uploading business object impression information based on multiple history can be obtained, then generate the detection model of business object impression information based on these attribute informations, follow-uply can to detect the business object impression information that putting person uploads according to the detection model of this business object impression information.

Step 102, receives the detection request carrying putting person's mark.

When certain putting person's request detection its upload business object impression information time, can corresponding operating be performed, these operations can trigger generate detect request, can comprise in this detection request the putting person of request detection putting person mark.

Step 104, according to the detection model of described business object impression information, identifies to described putting person the business object impression information that corresponding putting person uploads and detects.

Receive above-mentioned carry putting person mark detection request after, can according to the detection model of the business object impression information generated in advance, identify to this putting person the business object impression information that corresponding putting person uploads to detect, thus obtaining the score value of this business object impression information, the business object impression information more optimized for positioning service object impression information Problems existing and formulation provides foundation.

In the embodiment of the present invention, the detection model of business object impression information generates in conjunction with a large amount of historical information analyses, therefore, it is possible to more objective, exactly the business object impression information that putting person uploads is detected, and without the need to carrying out analytic statistics to the input effect of the putting person detected again when detecting according to this detection model, testing process is easier, improves detection efficiency.

Embodiment two:

With reference to Fig. 2, show the flow chart of steps of the detection method of a kind of business object impression information in the embodiment of the present invention two.The method can comprise the following steps:

Step 200, uploads the attribute information of the putting person of business object impression information in advance based on multiple history, generate the detection model of business object impression information.

Putting person is when needs recommend business object to the network user, first formulate popularization plan, divide popularization group, buy keyword, intention etc. is set, then business object impression information is generated according to these factors, last registering service object release platform, is uploaded to this business object impression information in business object release platform.Follow-up business object release platform can carry out the input of business object according to this business object impression information.

In the embodiment of the present invention, the detection model of the attribute information generation business object impression information of the putting person of business object impression information can be uploaded based on multiple history, therefore this detection model combines the feature of the attribute information of a large amount of putting person, can detect business object impression information more objectively.

In one preferred embodiment of the invention, this step 200 can comprise following sub-step a1 ~ sub-step a3:

Sub-step a1, many of gathering in advance in browser show daily record and many click logs;

When user browses webpage in a browser, after input keyword is searched for, search engine sorts with reference to the business object that different business object is corresponding to this keyword to the information such as to bid of this keyword, and determine whether push away a left side, then these business objects are illustrated in search result web page.During displayed web page, in browser, daily record is shown in generation, show that daily record describes the behavior of displayed web page.Can comprise in this displaying daily record: whether the mark of the putting person that the mark of the business object of displaying, the business object of displaying belong to, the business object of displaying be by the business object relevant information pushing away the information on a left side, the massfraction (being calculated by search engine) etc. of the business object of displaying is shown, other information of displaying can also be comprised, relevant information of the relevant information of the keyword such as shown, the picture of displaying etc. in certain displaying daily record.

When the business object of showing in user's webpage clicking, will generate click logs in browser, click logs describes the behavior of click-to-call service object.Can comprise in this click logs: whether the business object of the mark of the putting person that the mark of the business object of click, the business object of described click belong to, described click is by the business object relevant information pushing away the information on a left side, the consumption figures etc. of the business object of described click is clicked.

Sub-step a2, adds up according to described displaying daily record and described click logs the attribute information that each history uploads the putting person of business object impression information respectively;

Wherein, the attribute information of putting person can comprise: pageview, and/or left side pageview, and/or click volume, and/or left side click volume, and/or massfraction, and/or consumption figures, and/or clicking rate, namely can to comprise in pageview, left side pageview, click volume, left side click volume, massfraction, consumption figures, clicking rate any one or more for the attribute information of putting person.

Therefore, this sub-step a2 can comprise following sub-step:

Step a21, adds up in all displaying daily records the quantity of business object that belong to same putting person, that show, using the pageview of this quantity as described putting person; And/or,

Step a22, adds up in all displaying daily records and belongs to same putting person and pushed away the quantity of the business object of left displaying, using the left side pageview of this quantity as described putting person; And/or,

Step a23, adds up in all click logs the quantity of business object that belong to same putting person, that click, using the click volume of this quantity as described putting person; And/or,

Step a24, adds up in all click logs and belongs to same putting person and pushed away the quantity of the business object of left click, using the left side click volume of this quantity as described putting person; And/or,

Step a25, adds up the quantity of the business object of same displaying in all displaying daily records, using the pageview of this quantity as described business object; Add up the quantity of the business object of same click in all click logs, using the click volume of this quantity as described business object; Calculate the click volume of each business object and the quotient of pageview respectively, using the clicking rate of described quotient as described business object; Calculate the mean value belonging to the clicking rate of all business objects of same putting person, as the clicking rate of described putting person; And/or,

Step a26, calculates in all displaying daily records the mean value of the massfraction of business object that belong to same putting person, that show, using the massfraction of this mean value as described putting person; And/or,

Step a27, calculates in all click logs the summation of the consumption figures of business object that belong to same putting person, that click, using the consumption figures of this summation as described putting person.

According to the content that the attribute information of putting person comprises, from above-mentioned sub-step a21 ~ sub-step a27, corresponding sub-step is selected to perform.Such as, when the attribute information of putting person comprises pageview, left side pageview, click volume and left side click volume, then perform step a21 ~ sub-step a24, etc.When performing all or part of sub-step in above-mentioned steps a21 ~ sub-step a27 again, the execution sequence for each sub-steps is not limited, and can successively perform, and also can perform simultaneously.

After adding up each history respectively and uploading the attribute information of the putting person of business object impression information, by the attribute information write into Databasce of these putting persons, can also use for subsequent query.

Sub-step a3, the attribute information based on described multiple putting person generates the detection model of business object impression information.

After the attribute information getting each putting person above-mentioned, the detection model of business object impression information can be generated based on the attribute information of above-mentioned multiple putting person.

In one preferred embodiment of the invention, this sub-step a3 can comprise following sub-step a31 ~ sub-step a34:

Step a31, for each attribute information of each putting person, calculates the Rank scores value of current attribute information in the attribute information that all putting persons are identical with this attribute information respectively;

Add up the attribute information of multiple putting person, each putting person comprises again one or more attribute information, therefore for each attribute information of each putting person, the Rank scores value of current attribute information in the attribute information that all putting persons are identical with this attribute information is calculated respectively.Such as, for the pageview of one of them putting person, calculate the Rank scores value of pageview in the pageview of all putting persons of this putting person; For the click volume of one of them putting person, calculate the Rank scores value of click volume in the click volume of all putting persons of this putting person.After calculating, the Rank scores value of each attribute information of each putting person can be obtained.

Wherein, Rank scores value can be standardized centesimal system, can calculate this Rank scores value according to the ratio of the total number of putting person shared by the rank of current attribute information.Such as, the rank of current attribute information is 100, and total number of putting person is 1000, then shared by the rank of current attribute information, the ratio of the total number of putting person is 10%, and the Rank scores value of current attribute information is 90 points.

Step a32, using the Rank scores value of all properties information of a putting person as an object, carries out cluster to all objects;

Using the Rank scores value of all properties information of a putting person as an object, each object can a corresponding proper vector, and this proper vector is the proper vector be made up of the Rank scores value of all properties information of this putting person.

This sub-step a32 can comprise following sub-step a321 ~ sub-step a326:

Sub-step a321, carries out hierarchical clustering to all objects, determines the initial clustering of destination number;

This sub-step a321 can comprise following sub-step a3211 ~ sub-step a3215:

Sub-step a3211, using an object as an initial clustering, calculates the distance between every two initial clusterings respectively;

Calculate the distance between two initial clusterings, be the distance between calculating two initial clustering characteristic of correspondence vectors.

If an object is an initial clustering, then the distance between every two initial clusterings is the distance between these two object characteristic of correspondence vectors.In the embodiment of the present invention, the distance calculated between two proper vectors can be the Euclidean distance between calculating two proper vectors, manhatton distance, cosine similarity, Hamming distance, Ming Shi distance etc.Euclidean distance is derived from the range formula of point-to-point transmission in Euclidean space, two n-dimensional vector a (x ₁₁, x ₁₂..., x _1n) and b (x ₂₁, x ₂₂..., x _2n) between Euclidean distance be: also can by the form being expressed as vector operation: manhatton distance also referred to as city block distance, two n-dimensional vector a (x ₁₁, x ₁₂..., x _1n) and b (x ₂₁, x ₂₂..., x _2n) between manhatton distance be: for the computation process of all the other distances, the embodiment of the present invention is discussed no longer one by one at this.

Sub-step a3212, merges into an initial clustering by apart from minimum two initial clusterings;

Sub-step a3213, utilizes B (k) value that initial clustering described in following formulae discovery is corresponding:

$B\left(k\right)=\underset{1}{\overset{C_k^2}{\mathrm{\Σ}}}\mathrm{interDis}+\underset{1}{\overset{k}{\mathrm{\Σ}}}\mathrm{intraDis}$

Wherein, interDis is the distance between every two initial clusterings, and intraDis is the distance sum between inner every two objects of initial clustering, and k is the quantity of initial clustering;

B (k) value corresponding to each initial clustering can be calculated by this sub-step a3213.

Sub-step a3214, calculates the distance between the initial clustering after merging and other each initial clusterings, and returns and describedly will merge into the step of an initial clustering apart from minimum two initial clusterings, is till 1 until the number of initial clustering;

If an initial clustering comprises at least two objects, then when calculating the distance between this initial clustering and other initial clusterings, this initial clustering characteristic of correspondence vector is the central point of this initial clustering, is the mean value of all object characteristic of correspondence vectors that this initial clustering comprises.

After this sub-step a3214, can after each merging, the situation for current initial clustering obtains corresponding B (k) value.

Sub-step a3215, searches minimum B (k) value in all B (k) values, k corresponding for described minimum B (k) value initial clustering is defined as the initial clustering of destination number.

Sub-step a322, the barycenter of each initial clustering of random selecting;

For each initial clustering, the barycenter of this initial clustering of random selecting from the object that this initial clustering comprises, this barycenter is the object characteristic of correspondence vector of random selecting.

Sub-step a323, for each object, calculates the distance between existing object and each barycenter respectively, and is referred to by existing object in cluster corresponding to the barycenter minimum with the spacing of this object;

Distance between existing object and barycenter, is the distance between existing object characteristic of correspondence vector and barycenter.Such as, the initial clustering of the destination number determined is 6, then to there being 6 barycenter, calculate the distance between existing object and above-mentioned 6 barycenter respectively, be referred to by existing object in cluster corresponding to the barycenter minimum with the spacing of this object.

Sub-step a324, whether the cluster that each barycenter that judgement obtains is corresponding meets the condition of convergence; If not, then sub-step a325 is performed; If so, then sub-step a326 is performed;

This sub-step a324 can comprise following sub-step a3241 ~ sub-step a3243:

Sub-step a3241, utilizes the A value corresponding to cluster that each barycenter of obtaining described in following formulae discovery is corresponding:

$A=\min {\mathrm{\Σ}}_{i=1}^I{\mathrm{\Σ}}_{x_j\∈C_i}\mathrm{dist}{(\mathrm{center}\left(i\right),x_j)}^2$

Wherein, I is the quantity of cluster, C _ibe the combination of object in i-th cluster, x _jbe the jth object in i-th cluster, center (i) is the center of i-th cluster, and the center of i-th cluster is the mean value of all objects in i-th cluster;

Dist (center (i), x _j) be center center (i) of i-th cluster and the jth object x in i-th cluster _jbetween distance, be the distance between the jth object characteristic of correspondence vector in the mean value of all object characteristic of correspondence vectors in i-th cluster and i-th cluster.

Sub-step a3242, the A value that before obtaining, preset times calculates, and two adjacent A values every in the A value of this calculating and the A value that calculates of front preset times are compared;

Sub-step a3243, if the amplitude of variation of every two adjacent A values is all in preset range, then determines that the cluster that each barycenter of obtaining is corresponding meets the condition of convergence.

Wherein, for the concrete numerical value of preset times and preset range, the embodiment of the present invention is not limited.

Sub-step a325, recalculates the barycenter of each initial clustering, and returns sub-step a323;

If judge in sub-step a324 not meet the condition of convergence, then recalculate the barycenter of each initial clustering.The barycenter recalculating initial clustering herein can for calculating the mean value of all objects that this initial clustering comprises, the mean value of all object characteristic of correspondence vectors namely comprised.Then return and perform sub-step a323.

Sub-step a326, determines that the cluster that each barycenter of obtaining is corresponding is cluster result.

Sub-step a33, for each cluster, carries out linear regression analysis respectively, obtains the regression parameter that current cluster is corresponding;

Through above-mentioned sub-step a32, obtain the cluster that each barycenter is corresponding, then for each cluster, carry out linear regression analysis respectively, obtain the regression parameter that current cluster is corresponding.

This sub-step a33 can comprise following sub-step a331 ~ sub-step a332:

Sub-step a331, for each cluster, determines the following formula that each object in current cluster is corresponding respectively:

Y _n＝β ₀+β ₁X ₁+β ₂X ₂+…+β _mX _m+e

Wherein, β ₀~ β _mfor regression parameter, X ₁~ X _mbe respectively the Rank scores value of the attribute information of the n-th object, Y _nbe the Rank scores value of the consumption figures of the n-th object, m is the quantity of the attribute information of the n-th object, and e is stochastic error;

Sub-step a332, the system of equations that the formula corresponding according to each object in current cluster forms calculates regression parameter β corresponding to current cluster ₀~ β _mvalue.

Each object in current cluster is to there being an above-mentioned formula, and for current cluster, the system of equations formed according to these modes can calculate β wherein ₀~ β _mvalue, in computation process, stochastic error e can be deleted.

According to this sub-step a33, the regression parameter that each cluster is corresponding can be determined.

In the embodiment of the present invention, this sub-step a33 can adopt SPSS (Statistical Product andService Solutions, statistical product and service solution) software to carry out linear regression analysis for cluster result.Certainly, can also adopt other modes, the embodiment of the present invention is not limited this.

Step a34, for each cluster, utilizes regression parameter corresponding to current cluster to determine the detection model of the business object impression information that this cluster is corresponding respectively.

This sub-step a34 can comprise following sub-step a341 ~ sub-step a342:

Sub-step a341, for each cluster, utilizes the regression parameter β that current cluster is corresponding respectively ₀~ β _mvalue calculate following formula:

$\mathrm{Score}=\frac{{\mathrm{\β}}_0+{\mathrm{\β}}_1{X}_1+{\mathrm{\β}}_2{X}_2+...+{\mathrm{\β}}_m{X}_m}{{\mathrm{\β}}_0+{\mathrm{\β}}_1+{\mathrm{\β}}_2+...+{\mathrm{\β}}_m}$

Sub-step a342, is defined as the detection model of business object impression information corresponding to this cluster by the formula calculated.

Therefore, the detection model of business object impression information corresponding to each cluster can be obtained.

In one preferred embodiment of the invention, after executing above-mentioned sub-step a33, can also further for each cluster, the regression parameter corresponding to current cluster carries out significance test, and/or back substitution is checked.Namely only can carry out significance test, also only can carry out back substitution inspection, not only can also carry out significance test but also carry out back substitution inspection.

Significance test is exactly that the prior parameter to overall (stochastic variable) or population distribution form make a hypothesis, then utilize sample information to judge that whether rationally this hypothesis (standby then hypothesis), namely judges whether overall truth and null hypothesis have significant difference.For each cluster in above-mentioned sub-step a33, after carrying out linear regression analysis respectively, automatically can also obtain the coefficient of determination of the coefficient of determination and correction, the process of significance test is: judge whether the coefficient of determination of the coefficient of determination and correction is all greater than predetermined threshold value; If so, then determine to pass through significance test; If not, then determine not pass through significance test.

The process of back substitution inspection is: add up at least one attribute information uploading the putting person of business object impression information (can according to above-mentioned walk the mode of step a1 ~ sub-step a2 add up), and for each attribute information of each putting person, calculate the Rank scores value of current attribute information in the attribute information that all putting persons are identical with this attribute information respectively; The Rank scores value of the attribute information of at least one putting person is substituted into respectively formula (now, the β in above-mentioned sub-step a331 ₀~ β _mbe known, and e deletes), calculate corresponding Y value (the Rank scores value of consumption figures); Calculate the variance of the Rank scores value of the actual consumption value of above-mentioned Y value and this object; Judge described variance whether in preset range; If so, then determine to be checked by back substitution; If not, then determine not checked by back substitution.

If the regression parameter corresponding for certain cluster is not checked by significance test or back substitution, then can delete this cluster, or return to execution sub-step a32; If by significance test and back substitution inspection, then perform this sub-step a4.

Step 202, receives the detection request carrying putting person's mark.

In the embodiment of the present invention, one can be pre-set in a browser and detect control, when certain putting person needs the business object impression information uploaded self to detect, generation detection request can be triggered by clicking this detection control, putting person's mark of this putting person in this detection request, can be comprised.

Step 204, according to the detection model of described business object impression information, identifies to described putting person the business object impression information that corresponding putting person uploads and detects.

After receiving the detection request carrying putting person's mark, the detection model according to business object impression information detects the business object impression information that the putting person that this putting person identifies correspondence uploads.

In one preferred embodiment of the invention, this step 204 can comprise following sub-step b1 ~ sub-step b4:

Sub-step b1, obtains the attribute information that described putting person identifies corresponding putting person;

First, many of can gather in browser show daily record and many click logs; Then, add up according to described displaying daily record and click logs the attribute information that this putting person identifies corresponding putting person.For concrete process, with reference to the associated description of above-mentioned sub-step a1 ~ sub-step a2.

Sub-step b2, determines the cluster belonging to attribute information of the putting person that described mark is corresponding;

This step b2 can comprise following sub-step b21 ~ sub-step b23:

Sub-step b21, for each attribute information of the putting person of described mark correspondence, calculates the Rank scores value of current attribute information respectively;

In this sub-step b21, first, based on the displaying daily record obtained in above-mentioned sub-step b1 and click logs, the attribute information of each putting person can be added up respectively; Then, for each attribute information of the putting person of described mark correspondence, the Rank scores value of current attribute information is calculated respectively.For concrete process, with reference to the associated description of above-mentioned sub-step a2 ~ sub-step a3.

Sub-step b22, using the Rank scores value of the attribute information of described putting person as an object, calculates the distance between this object barycenter corresponding with each cluster respectively;

Sub-step b23, determines the cluster of cluster belonging to the described attribute information identifying corresponding putting person that the barycenter minimum with the spacing of described object is corresponding.

Sub-step b3, using the input of the Rank scores value of the attribute information of putting person corresponding for described mark as the detection model of business object impression information corresponding to the cluster determined;

Determine this mark correspondence putting person attribute information belonging to cluster after, the detection model of business object impression information corresponding to this acute class can be obtained, i.e. following formula:

$\mathrm{Score}=\frac{{\mathrm{\β}}_0+{\mathrm{\β}}_1{X}_1+{\mathrm{\β}}_2{X}_2+...+{\mathrm{\β}}_m{X}_m}{{\mathrm{\β}}_0+{\mathrm{\β}}_1+{\mathrm{\β}}_2+...+{\mathrm{\β}}_m}$

Then using the input of the Rank scores value of the attribute information of putting person corresponding for mark as the detection model of this business object impression information, i.e. X ₁~ X _m.

Sub-step b4, identifies the score value of the business object impression information that corresponding putting person uploads as described putting person using the output of the detection model of described business object impression information.

Through the calculating of the detection model of business object impression information, finally the output of the detection model of this business object impression information is the score value that putting person identifies the business object impression information that corresponding putting person uploads.

Step 206, shows that described putting person identifies the testing result of the business object impression information that corresponding putting person uploads.

After obtaining above-mentioned testing result, can show that this putting person identifies the testing result of the business object impression information that corresponding putting person uploads in a browser, the score value of the business object impression information namely uploaded.

Step 208, obtains the attribute information that described putting person identifies corresponding putting person, and shows described attribute information.

In one preferred embodiment of the invention, the attribute information that described putting person identifies corresponding putting person can also be obtained further, and show described attribute information.

For concrete exhibition method, the embodiment of the present invention is not limited, and such as, the mode loading prompting frame can be adopted to show, prompting frame comprises score value etc.

By the score value of business object impression information corresponding for this putting person and attribute information are showed putting person, make putting person clearly can understand the input effect of its business object, the business object impression information more optimized for positioning service object impression information Problems existing and formulation provides foundation.

Step 206 and step 208 are not limited to above-mentioned execution sequence, first can perform step 206, then perform step 208; Also first can perform step 208, then perform step 206; Can also perform step 206 and step 208, the embodiment of the present invention is not limited this simultaneously.

Business object in the embodiment of the present invention can be advertisement, and business object release platform is advertisement launching platform, and putting person is advertiser, and business object impression information is advertisement serving policy.

In the embodiment of the present invention, first, the objective health score assignings of making of index such as the attribute information of putting person are combined, for follow-up formulation Optimizing Suggestions provides Main Basis; Secondly, in conjunction with the input effect of a large amount of putting person, improve entirety and to bid the liveness of environment, be conducive to the integral benefit of business object release platform; Finally, use multiple linear regression analysis to build detection model, the relation of the healthy mark of putting person and each index can comparatively accurately be described.

Embodiment three:

With reference to Fig. 3, show the structured flowchart of the pick-up unit of a kind of business object impression information in the embodiment of the present invention three.

Generation module 300, is suitable for the attribute information of the putting person uploading business object impression information in advance based on multiple history, generates the detection model of business object impression information;

Receiver module 302, is suitable for receiving the detection request carrying putting person's mark;

Detection module 304, is suitable for the detection model according to described business object impression information, identifies the business object impression information that corresponding putting person uploads detect described putting person.

In the embodiment of the present invention, the attribute information of the putting person of business object impression information is uploaded in advance based on multiple history, generate the detection model of business object impression information, follow-up after receiving the detection request carrying putting person's mark, can detect the business object impression information that the putting person that putting person identifies correspondence uploads according to the detection model of the business object impression information generated.The detection model of above-mentioned business object impression information generates in conjunction with a large amount of historical information analyses, therefore, it is possible to more objective, exactly the business object impression information that putting person uploads is detected, and without the need to carrying out analytic statistics to the input effect of the putting person detected again when detecting according to this detection model, testing process is easier, improves detection efficiency.

Embodiment four:

With reference to Fig. 4, show the structured flowchart of the pick-up unit of a kind of business object impression information in the embodiment of the present invention four.

Generation module 400, is suitable for the attribute information of the putting person uploading business object impression information in advance based on multiple history, generates the detection model of business object impression information;

Receiver module 402, is suitable for receiving the detection request carrying putting person's mark;

Detection module 404, is suitable for the detection model according to described business object impression information, identifies the business object impression information that corresponding putting person uploads detect described putting person.

Result display module 406, is suitable for showing that described putting person identifies the testing result of the business object impression information that corresponding putting person uploads;

Attribute display module 408, is suitable for obtaining the attribute information that described putting person identifies corresponding putting person, and shows described attribute information.

In one preferred embodiment of the invention, generation module can comprise following submodule:

Log collection submodule, is suitable for many of gathering in advance in browser and shows daily record and many click logs;

Statistics of attributes submodule, is suitable for adding up according to described displaying daily record and described click logs the attribute information that each history uploads the putting person of business object impression information respectively;

Model generation submodule, is suitable for the detection model generating business object impression information based on the attribute information of described multiple putting person.

Wherein, described displaying daily record comprises: whether the business object of the mark of the putting person that the mark of the business object of displaying, the business object of described displaying belong to, described displaying is pushed away the massfraction of the information on a left side, the business object of described displaying; Described click logs comprises: whether the business object of the mark of the putting person that the mark of the business object of click, the business object of described click belong to, described click is pushed away the consumption figures of the information on a left side, the business object of described click; The attribute information of described putting person comprises: pageview, and/or left side pageview, and/or click volume, and/or left side click volume, and/or massfraction, and/or consumption figures, and/or clicking rate.

Statistics of attributes submodule can comprise with lower unit:

Pageview statistic unit, is suitable for adding up in all displaying daily records the quantity of business object that belong to same putting person, that show, using the pageview of this quantity as described putting person; And/or,

Left side pageview statistic unit, is suitable for adding up in all displaying daily records and belongs to same putting person and pushed away the quantity of the business object of left displaying, using the left side pageview of this quantity as described putting person; And/or,

Click volume statistic unit, is suitable for adding up in all click logs the quantity of business object that belong to same putting person, that click, using the click volume of this quantity as described putting person; And/or,

Left side click volume statistic unit, is suitable for adding up in all click logs and belongs to same putting person and pushed away the quantity of the business object of left click, using the left side click volume of this quantity as described putting person; And/or,

Clicking rate statistic unit, is suitable for the quantity of the business object of adding up same displaying in all displaying daily records, using the pageview of this quantity as described business object; Add up the quantity of the business object of same click in all click logs, using the click volume of this quantity as described business object; Calculate the click volume of each business object and the quotient of pageview respectively, using the clicking rate of described quotient as described business object; Calculate the mean value belonging to the clicking rate of all business objects of same putting person, as the clicking rate of described putting person; And/or,

Fractional statistics unit, is suitable for calculating in all displaying daily records the mean value of the massfraction of business object that belong to same putting person, that show, using the massfraction of this mean value as described putting person; And/or,

Consumption statistics unit, is suitable for calculating in all click logs the summation of the consumption figures of business object that belong to same putting person, that click, using the consumption figures of this summation as described putting person.

Model generation submodule can comprise with lower unit:

Computing unit, is suitable for each attribute information for each putting person, calculates the Rank scores value of current attribute information in the attribute information that all putting persons are identical with this attribute information respectively;

Cluster cell, is suitable for, using the Rank scores value of all properties information of a putting person as an object, carrying out cluster to all objects;

Analytic unit, is suitable for, for each cluster, carrying out linear regression analysis respectively, obtains the regression parameter that current cluster is corresponding;

Determining unit, is suitable for for each cluster, utilizes regression parameter corresponding to current cluster to determine the detection model of the business object impression information that this cluster is corresponding respectively.

Wherein, cluster cell can comprise unit quickly:

Hierarchical clustering subelement, is suitable for carrying out hierarchical clustering to all objects, determines the initial clustering of destination number;

Choose subelement, be suitable for the barycenter of each initial clustering of random selecting;

Sort out subelement, be suitable for for each object, calculate the distance between existing object and each barycenter respectively, and existing object is referred in cluster corresponding to the barycenter minimum with the spacing of this object;

Judgment sub-unit, whether the cluster that each barycenter being suitable for judging to obtain is corresponding meets the condition of convergence; If not, then recalculate the barycenter of each initial clustering, and call described classification subelement; If so, then determine that the cluster that each barycenter of obtaining is corresponding is cluster result.

Hierarchical clustering subelement, is specifically suitable for:

Using an object as an initial clustering, calculate the distance between every two initial clusterings respectively;

An initial clustering is merged into by apart from minimum two initial clusterings;

Utilize B (k) value that initial clustering described in following formulae discovery is corresponding:

$B\left(k\right)=\underset{1}{\overset{C_k^2}{\mathrm{\Σ}}}\mathrm{interDis}+\underset{1}{\overset{k}{\mathrm{\Σ}}}\mathrm{intraDis}$

Wherein, interDis is the distance between every two initial clusterings, and intraDis is the distance sum between inner every two objects of initial clustering, and k is the quantity of initial clustering;

Calculating the distance between the initial clustering after merging and other each initial clusterings, and return and describedly will merge into the step of an initial clustering apart from minimum two initial clusterings, is till 1 until the number of initial clustering;

Search minimum B (k) value in all B (k) values, k corresponding for described minimum B (k) value initial clustering is defined as the initial clustering of destination number.

Judgment sub-unit, is specifically suitable for:

Utilize the A value corresponding to cluster that each barycenter of obtaining described in following formulae discovery is corresponding:

$A=\min {\mathrm{\Σ}}_{i=1}^I{\mathrm{\Σ}}_{x_j\∈C_i}\mathrm{dist}{(\mathrm{center}\left(i\right),x_j)}^2$

Wherein, I is the quantity of cluster, C _ibe the combination of object in i-th cluster, x _jbe the jth object in i-th cluster, center (i) is the center of i-th cluster, and the center of i-th cluster is the mean value of all objects in i-th cluster;

The A value that before obtaining, preset times calculates, and two adjacent A values every in the A value of this calculating and the A value that calculates of front preset times are compared;

If the amplitude of variation of every two adjacent A values is all in preset range, then determine that the cluster that each barycenter of obtaining is corresponding meets the condition of convergence.

Analytic unit can comprise following subelement:

Formula determination subelement, is suitable for for each cluster, determines the following formula that each object in current cluster is corresponding respectively:

Y _n＝β ₀+β ₁X ₁+β ₂X ₂+…+β _mX _m+e

Wherein, β ₀~ β _mfor regression parameter, X ₁~ X _mbe respectively the Rank scores value of the attribute information of the n-th object, Y _nbe the Rank scores value of the consumption figures of the n-th object, m is the quantity of the attribute information of the n-th object, and e is stochastic error;

Parameter computation unit, is suitable for calculating regression parameter β corresponding to current cluster according to the system of equations of formula composition corresponding to each object in current cluster ₀~ β _mvalue.

Determining unit can comprise following subelement:

Formulae discovery subelement, is suitable for for each cluster, utilizes the regression parameter β that current cluster is corresponding respectively ₀~ β _mvalue calculate following formula:

$\mathrm{Score}=\frac{{\mathrm{\β}}_0+{\mathrm{\β}}_1{X}_1+{\mathrm{\β}}_2{X}_2+...+{\mathrm{\β}}_m{X}_m}{{\mathrm{\β}}_0+{\mathrm{\β}}_1+{\mathrm{\β}}_2+...+{\mathrm{\β}}_m}$

Clustering Model determination subelement, is suitable for the detection model formula calculated being defined as business object impression information corresponding to this cluster.

Detection module can comprise following submodule:

Attribute obtains submodule, is suitable for obtaining the attribute information that described putting person identifies corresponding putting person;

Cluster determination submodule, is suitable for the cluster belonging to attribute information determining the putting person that described mark is corresponding;

Information scoring submodule, is suitable for the input of the Rank scores value of the attribute information of putting person corresponding for described mark as the detection model of business object impression information corresponding to the cluster determined; The output of the detection model of described business object impression information is identified the score value of the business object impression information that corresponding putting person uploads as described putting person.

Cluster determination submodule can comprise with lower unit:

Score calculation unit, is suitable for, for each attribute information of putting person corresponding to described mark, calculating the Rank scores value of current attribute information respectively;

Metrics calculation unit, is suitable for, using the Rank scores value of the attribute information of described putting person as an object, calculating the distance between this object barycenter corresponding with each cluster respectively;

Cluster determining unit, is suitable for the cluster of cluster belonging to the described attribute information identifying corresponding putting person determining that the barycenter minimum with the spacing of described object is corresponding.

The detection model of above-mentioned business object impression information generates in conjunction with a large amount of historical information analyses, therefore, it is possible to more objective, exactly the business object impression information that putting person uploads is detected, and without the need to carrying out analytic statistics to the input effect of the putting person detected again when detecting according to this detection model, testing process is easier, improves detection efficiency.

Intrinsic not relevant to any certain computer, virtual system or miscellaneous equipment with display at this algorithm provided.Various general-purpose system also can with use based on together with this teaching.According to description above, the structure constructed required by this type systematic is apparent.In addition, the present invention is not also for any certain programmed language.It should be understood that and various programming language can be utilized to realize content of the present invention described here, and the description done language-specific is above to disclose preferred forms of the present invention.

In instructions provided herein, describe a large amount of detail.But can understand, embodiments of the invention can be put into practice when not having these details.In some instances, be not shown specifically known method, structure and technology, so that not fuzzy understanding of this description.

Similarly, be to be understood that, in order to simplify the disclosure and to help to understand in each inventive aspect one or more, in the description above to exemplary embodiment of the present invention, each feature of the present invention is grouped together in single embodiment, figure or the description to it sometimes.But, the method for the disclosure should be construed to the following intention of reflection: namely the present invention for required protection requires feature more more than the feature clearly recorded in each claim.Or rather, as claims below reflect, all features of disclosed single embodiment before inventive aspect is to be less than.Therefore, the claims following embodiment are incorporated to this embodiment thus clearly, and wherein each claim itself is as independent embodiment of the present invention.

Those skilled in the art are appreciated that and adaptively can change the module in the equipment in embodiment and they are arranged in one or more equipment different from this embodiment.Module in embodiment or unit or assembly can be combined into a module or unit or assembly, and multiple submodule or subelement or sub-component can be put them in addition.Except at least some in such feature and/or process or unit be mutually repel except, any combination can be adopted to combine all processes of all features disclosed in this instructions (comprising adjoint claim, summary and accompanying drawing) and so disclosed any method or equipment or unit.Unless expressly stated otherwise, each feature disclosed in this instructions (comprising adjoint claim, summary and accompanying drawing) can by providing identical, alternative features that is equivalent or similar object replaces.

In addition, those skilled in the art can understand, although embodiments more described herein to comprise in other embodiment some included feature instead of further feature, the combination of the feature of different embodiment means and to be within scope of the present invention and to form different embodiments.Such as, in the following claims, the one of any of embodiment required for protection can use with arbitrary array mode.

All parts embodiment of the present invention with hardware implementing, or can realize with the software module run on one or more processor, or realizes with their combination.It will be understood by those of skill in the art that the some or all functions that microprocessor or digital signal processor (DSP) can be used in practice to realize according to the some or all parts in the checkout equipment of the business object impression information of the embodiment of the present invention.The present invention can also be embodied as part or all equipment for performing method as described herein or device program (such as, computer program and computer program).Realizing program of the present invention and can store on a computer-readable medium like this, or the form of one or more signal can be had.Such signal can be downloaded from internet website and obtain, or provides on carrier signal, or provides with any other form.

The present invention will be described instead of limit the invention to it should be noted above-described embodiment, and those skilled in the art can design alternative embodiment when not departing from the scope of claims.In the claims, any reference symbol between bracket should be configured to limitations on claims.Word " comprises " not to be got rid of existence and does not arrange element in the claims or step.Word "a" or "an" before being positioned at element is not got rid of and be there is multiple such element.The present invention can by means of including the hardware of some different elements and realizing by means of the computing machine of suitably programming.In the unit claim listing some devices, several in these devices can be carry out imbody by same hardware branch.Word first, second and third-class use do not represent any order.Can be title by these word explanations.

The detection method of A1, a kind of business object impression information, wherein, comprising:

Upload the attribute information of the putting person of business object impression information in advance based on multiple history, generate the detection model of business object impression information;

Receive the detection request carrying putting person's mark;

According to the detection model of described business object impression information, the business object impression information that corresponding putting person uploads is identified to described putting person and detects.

A2, method as described in A1, wherein, the described attribute information uploading the putting person of business object impression information in advance based on multiple history, the step generating the detection model of business object impression information comprises:

Many of gathering in advance in browser show daily record and many click logs;

The attribute information that each history uploads the putting person of business object impression information is added up respectively according to described displaying daily record and described click logs;

Attribute information based on described multiple putting person generates the detection model of business object impression information.

A3, method as described in A2, wherein,

Described displaying daily record comprises: whether the business object of the mark of the putting person that the mark of the business object of displaying, the business object of described displaying belong to, described displaying is pushed away the massfraction of the information on a left side, the business object of described displaying;

Described click logs comprises: whether the business object of the mark of the putting person that the mark of the business object of click, the business object of described click belong to, described click is pushed away the consumption figures of the information on a left side, the business object of described click;

The attribute information of described putting person comprises: pageview, and/or left side pageview, and/or click volume, and/or left side click volume, and/or massfraction, and/or consumption figures, and/or clicking rate.

A4, method as described in A3, wherein, describedly add up according to described displaying daily record and described click logs the step that each history uploads the attribute information of the putting person of business object impression information respectively and comprise:

Add up in all displaying daily records the quantity of business object that belong to same putting person, that show, using the pageview of this quantity as described putting person; And/or,

Add up in all displaying daily records and belong to same putting person and pushed away the quantity of the business object of left displaying, using the left side pageview of this quantity as described putting person; And/or,

Add up in all click logs the quantity of business object that belong to same putting person, that click, using the click volume of this quantity as described putting person; And/or,

Add up in all click logs and belong to same putting person and pushed away the quantity of the business object of left click, using the left side click volume of this quantity as described putting person; And/or,

Add up the quantity of the business object of same displaying in all displaying daily records, using the pageview of this quantity as described business object; Add up the quantity of the business object of same click in all click logs, using the click volume of this quantity as described business object; Calculate the click volume of each business object and the quotient of pageview respectively, using the clicking rate of described quotient as described business object; Calculate the mean value belonging to the clicking rate of all business objects of same putting person, as the clicking rate of described putting person; And/or,

Calculate in all displaying daily records the mean value of the massfraction of business object that belong to same putting person, that show, using the massfraction of this mean value as described putting person; And/or,

Calculate in all click logs the summation of the consumption figures of business object that belong to same putting person, that click, using the consumption figures of this summation as described putting person.

A5, method as described in A2, wherein, the step that the described attribute information based on described multiple putting person generates the detection model of business object impression information comprises:

For each attribute information of each putting person, calculate the Rank scores value of current attribute information in the attribute information that all putting persons are identical with this attribute information respectively;

Using the Rank scores value of all properties information of a putting person as an object, cluster is carried out to all objects;

For each cluster, carry out linear regression analysis respectively, obtain the regression parameter that current cluster is corresponding;

For each cluster, regression parameter corresponding to current cluster is utilized to determine the detection model of the business object impression information that this cluster is corresponding respectively.

A6, method as described in A5, wherein, describedly to comprise the step that all objects carry out cluster:

Hierarchical clustering is carried out to all objects, determines the initial clustering of destination number;

The barycenter of each initial clustering of random selecting;

For each object, calculate the distance between existing object and each barycenter respectively, and existing object is referred in cluster corresponding to the barycenter minimum with the spacing of this object;

Whether the cluster that each barycenter that judgement obtains is corresponding meets the condition of convergence;

If not, then recalculate the barycenter of each initial clustering, and return described for each object, calculate the distance between existing object and each barycenter respectively, and existing object is referred to the step in cluster corresponding to the barycenter minimum with the spacing of this object;

If so, then determine that the cluster that each barycenter of obtaining is corresponding is cluster result.

A7, method as described in A6, wherein, described hierarchical clustering is carried out to all objects, determine that the step of the initial clustering of destination number comprises:

Using an object as an initial clustering, calculate the distance between every two initial clusterings respectively;

An initial clustering is merged into by apart from minimum two initial clusterings;

Utilize B (k) value that initial clustering described in following formulae discovery is corresponding:

$B\left(k\right)=\underset{1}{\overset{C_k^2}{\mathrm{\Σ}}}\mathrm{interDis}+\underset{1}{\overset{k}{\mathrm{\Σ}}}\mathrm{intraDis}$

Wherein, interDis is the distance between every two initial clusterings, and intraDis is the distance sum between inner every two objects of initial clustering, and k is the quantity of initial clustering;

Calculating the distance between the initial clustering after merging and other each initial clusterings, and return and describedly will merge into the step of an initial clustering apart from minimum two initial clusterings, is till 1 until the number of initial clustering;

Search minimum B (k) value in all B (k) values, k corresponding for described minimum B (k) value initial clustering is defined as the initial clustering of destination number.

A8, method as described in A6, wherein, the step whether cluster that each barycenter that described judgement obtains is corresponding meets the condition of convergence comprises:

Utilize the A value corresponding to cluster that each barycenter of obtaining described in following formulae discovery is corresponding:

$A=\min {\mathrm{\Σ}}_{i=1}^I{\mathrm{\Σ}}_{x_j\∈C_i}\mathrm{dist}{(\mathrm{center}\left(i\right),x_j)}^2$

Wherein, I is the quantity of cluster, C _ibe the combination of object in i-th cluster, x _jbe the jth object in i-th cluster, center (i) is the center of i-th cluster, and the center of i-th cluster is the mean value of all objects in i-th cluster;

The A value that before obtaining, preset times calculates, and two adjacent A values every in the A value of this calculating and the A value that calculates of front preset times are compared;

If the amplitude of variation of every two adjacent A values is all in preset range, then determine that the cluster that each barycenter of obtaining is corresponding meets the condition of convergence.

A9, method as described in A5, wherein, described attribute information comprises consumption figures,

Described for each cluster, carry out linear regression analysis respectively, the step obtaining regression parameter corresponding to current cluster comprises:

For each cluster, determine the following formula that each object in current cluster is corresponding respectively:

Y _n＝β ₀+β ₁X ₁+β ₂X ₂+…+β _mX _m+e

Wherein, β ₀~ β _mfor regression parameter, X ₁~ X _mbe respectively the Rank scores value of the attribute information of the n-th object, Y _nbe the Rank scores value of the consumption figures of the n-th object, m is the quantity of the attribute information of the n-th object, and e is stochastic error;

The system of equations that the formula corresponding according to each object in current cluster forms calculates regression parameter β corresponding to current cluster ₀~ β _mvalue.

A10, method as described in A9, wherein, described for each cluster, the step of the detection model of the business object impression information that this cluster is corresponding comprises to utilize regression parameter corresponding to current cluster to determine respectively:

For each cluster, utilize the regression parameter β that current cluster is corresponding respectively ₀~ β _mvalue calculate following formula:

$\mathrm{Score}=\frac{{\mathrm{\β}}_0+{\mathrm{\β}}_1{X}_1+{\mathrm{\β}}_2{X}_2+...+{\mathrm{\β}}_m{X}_m}{{\mathrm{\β}}_0+{\mathrm{\β}}_1+{\mathrm{\β}}_2+...+{\mathrm{\β}}_m}$

The formula calculated is defined as the detection model of business object impression information corresponding to this cluster.

A11, method as described in A6, wherein, the described detection model according to described business object impression information, identifies to described putting person the step that business object impression information that corresponding putting person uploads detects and comprises:

Obtain the attribute information that described putting person identifies corresponding putting person;

Determine the cluster belonging to attribute information of the putting person that described mark is corresponding;

Using the input of the Rank scores value of the attribute information of putting person corresponding for described mark as the detection model of business object impression information corresponding to the cluster determined;

The output of the detection model of described business object impression information is identified the score value of the business object impression information that corresponding putting person uploads as described putting person.

A12, method as described in A11, wherein, describedly determine that the step of the cluster belonging to attribute information of the putting person that described mark is corresponding comprises:

For each attribute information of the putting person of described mark correspondence, calculate the Rank scores value of current attribute information respectively;

Using the Rank scores value of the attribute information of described putting person as an object, calculate the distance between this object barycenter corresponding with each cluster respectively;

Determine the cluster of cluster belonging to the described attribute information identifying corresponding putting person that the barycenter minimum with the spacing of described object is corresponding.

A13, method as described in A1, wherein, also comprise:

Show that described putting person identifies the testing result of the business object impression information that corresponding putting person uploads.

A14, method as described in A1, wherein, also comprise:

Obtain the attribute information that described putting person identifies corresponding putting person, and show described attribute information.

The pick-up unit of B15, a kind of business object impression information, wherein, comprising:

Generation module, is suitable for the attribute information of the putting person uploading business object impression information in advance based on multiple history, generates the detection model of business object impression information;

Receiver module, is suitable for receiving the detection request carrying putting person's mark;

Detection module, is suitable for the detection model according to described business object impression information, identifies the business object impression information that corresponding putting person uploads detect described putting person.

B16, device as described in B15, wherein, described generation module comprises:

Log collection submodule, is suitable for many of gathering in advance in browser and shows daily record and many click logs;

Statistics of attributes submodule, is suitable for adding up according to described displaying daily record and described click logs the attribute information that each history uploads the putting person of business object impression information respectively;

Model generation submodule, is suitable for the detection model generating business object impression information based on the attribute information of described multiple putting person.

B17, device as described in B16, wherein,

Described displaying daily record comprises: whether the business object of the mark of the putting person that the mark of the business object of displaying, the business object of described displaying belong to, described displaying is pushed away the massfraction of the information on a left side, the business object of described displaying;

Described click logs comprises: whether the business object of the mark of the putting person that the mark of the business object of click, the business object of described click belong to, described click is pushed away the consumption figures of the information on a left side, the business object of described click;

The attribute information of described putting person comprises: pageview, and/or left side pageview, and/or click volume, and/or left side click volume, and/or massfraction, and/or consumption figures, and/or clicking rate.

B18, device as described in B17, wherein, described statistics of attributes submodule comprises:

Pageview statistic unit, is suitable for adding up in all displaying daily records the quantity of business object that belong to same putting person, that show, using the pageview of this quantity as described putting person; And/or,

Left side pageview statistic unit, is suitable for adding up in all displaying daily records and belongs to same putting person and pushed away the quantity of the business object of left displaying, using the left side pageview of this quantity as described putting person; And/or,

Click volume statistic unit, is suitable for adding up in all click logs the quantity of business object that belong to same putting person, that click, using the click volume of this quantity as described putting person; And/or,

Left side click volume statistic unit, is suitable for adding up in all click logs and belongs to same putting person and pushed away the quantity of the business object of left click, using the left side click volume of this quantity as described putting person; And/or,

Clicking rate statistic unit, is suitable for the quantity of the business object of adding up same displaying in all displaying daily records, using the pageview of this quantity as described business object; Add up the quantity of the business object of same click in all click logs, using the click volume of this quantity as described business object; Calculate the click volume of each business object and the quotient of pageview respectively, using the clicking rate of described quotient as described business object; Calculate the mean value belonging to the clicking rate of all business objects of same putting person, as the clicking rate of described putting person; And/or,

Fractional statistics unit, is suitable for calculating in all displaying daily records the mean value of the massfraction of business object that belong to same putting person, that show, using the massfraction of this mean value as described putting person; And/or,

Consumption statistics unit, is suitable for calculating in all click logs the summation of the consumption figures of business object that belong to same putting person, that click, using the consumption figures of this summation as described putting person.

B19, device as described in B16, wherein, described model generation submodule comprises:

Computing unit, is suitable for each attribute information for each putting person, calculates the Rank scores value of current attribute information in the attribute information that all putting persons are identical with this attribute information respectively;

Cluster cell, is suitable for, using the Rank scores value of all properties information of a putting person as an object, carrying out cluster to all objects;

Analytic unit, is suitable for, for each cluster, carrying out linear regression analysis respectively, obtains the regression parameter that current cluster is corresponding;

Determining unit, is suitable for for each cluster, utilizes regression parameter corresponding to current cluster to determine the detection model of the business object impression information that this cluster is corresponding respectively.

B20, device as described in B19, wherein, described cluster cell comprises:

Hierarchical clustering subelement, is suitable for carrying out hierarchical clustering to all objects, determines the initial clustering of destination number;

Choose subelement, be suitable for the barycenter of each initial clustering of random selecting;

Sort out subelement, be suitable for for each object, calculate the distance between existing object and each barycenter respectively, and existing object is referred in cluster corresponding to the barycenter minimum with the spacing of this object;

Judgment sub-unit, whether the cluster that each barycenter being suitable for judging to obtain is corresponding meets the condition of convergence; If not, then recalculate the barycenter of each initial clustering, and call described classification subelement; If so, then determine that the cluster that each barycenter of obtaining is corresponding is cluster result.

B21, device as described in B20, wherein, described hierarchical clustering subelement, is specifically suitable for:

Using an object as an initial clustering, calculate the distance between every two initial clusterings respectively;

An initial clustering is merged into by apart from minimum two initial clusterings;

Utilize B (k) value that initial clustering described in following formulae discovery is corresponding:

$B\left(k\right)=\underset{1}{\overset{C_k^2}{\mathrm{\Σ}}}\mathrm{interDis}+\underset{1}{\overset{k}{\mathrm{\Σ}}}\mathrm{intraDis}$

Wherein, interDis is the distance between every two initial clusterings, and intraDis is the distance sum between inner every two objects of initial clustering, and k is the quantity of initial clustering;

Calculating the distance between the initial clustering after merging and other each initial clusterings, and return and describedly will merge into the step of an initial clustering apart from minimum two initial clusterings, is till 1 until the number of initial clustering;

Search minimum B (k) value in all B (k) values, k corresponding for described minimum B (k) value initial clustering is defined as the initial clustering of destination number.

B22, device as described in B20, wherein, described judgment sub-unit, is specifically suitable for:

Utilize the A value corresponding to cluster that each barycenter of obtaining described in following formulae discovery is corresponding:

$A=\min {\mathrm{\Σ}}_{i=1}^I{\mathrm{\Σ}}_{x_j\∈C_i}\mathrm{dist}{(\mathrm{center}\left(i\right),x_j)}^2$

Wherein, I is the quantity of cluster, C _ibe the combination of object in i-th cluster, x _jbe the jth object in i-th cluster, center (i) is the center of i-th cluster, and the center of i-th cluster is the mean value of all objects in i-th cluster;

The A value that before obtaining, preset times calculates, and two adjacent A values every in the A value of this calculating and the A value that calculates of front preset times are compared;

If the amplitude of variation of every two adjacent A values is all in preset range, then determine that the cluster that each barycenter of obtaining is corresponding meets the condition of convergence.

B23, device as described in B19, wherein, described attribute information comprises consumption figures, and described analytic unit comprises:

Formula determination subelement, is suitable for for each cluster, determines the following formula that each object in current cluster is corresponding respectively:

Y _n＝β ₀+β ₁X ₁+β ₂X ₂+…+β _mX _m+e

Wherein, β ₀~ β _mfor regression parameter, X ₁~ X _mbe respectively the Rank scores value of the attribute information of the n-th object, Y _nbe the Rank scores value of the consumption figures of the n-th object, m is the quantity of the attribute information of the n-th object, and e is stochastic error;

Parameter computation unit, is suitable for calculating regression parameter β corresponding to current cluster according to the system of equations of formula composition corresponding to each object in current cluster ₀~ β _mvalue.

B24, device as described in B23, wherein, described determining unit comprises:

Formulae discovery subelement, is suitable for for each cluster, utilizes the regression parameter β that current cluster is corresponding respectively ₀~ β _mvalue calculate following formula:

$\mathrm{Score}=\frac{{\mathrm{\β}}_0+{\mathrm{\β}}_1{X}_1+{\mathrm{\β}}_2{X}_2+...+{\mathrm{\β}}_m{X}_m}{{\mathrm{\β}}_0+{\mathrm{\β}}_1+{\mathrm{\β}}_2+...+{\mathrm{\β}}_m}$

Clustering Model determination subelement, is suitable for the detection model formula calculated being defined as business object impression information corresponding to this cluster.

B25, device as described in B20, wherein, described detection module comprises:

Attribute obtains submodule, is suitable for obtaining the attribute information that described putting person identifies corresponding putting person;

Cluster determination submodule, is suitable for the cluster belonging to attribute information determining the putting person that described mark is corresponding;

Information scoring submodule, is suitable for the input of the Rank scores value of the attribute information of putting person corresponding for described mark as the detection model of business object impression information corresponding to the cluster determined; The output of the detection model of described business object impression information is identified the score value of the business object impression information that corresponding putting person uploads as described putting person.

B26, device as described in B25, wherein, described cluster determination submodule comprises:

Score calculation unit, is suitable for, for each attribute information of putting person corresponding to described mark, calculating the Rank scores value of current attribute information respectively;

Metrics calculation unit, is suitable for, using the Rank scores value of the attribute information of described putting person as an object, calculating the distance between this object barycenter corresponding with each cluster respectively;

Cluster determining unit, is suitable for the cluster of cluster belonging to the described attribute information identifying corresponding putting person determining that the barycenter minimum with the spacing of described object is corresponding.

B27, device as described in B15, wherein, also comprise:

Result display module, is suitable for showing that described putting person identifies the testing result of the business object impression information that corresponding putting person uploads.

B28, device as described in B15, wherein, also comprise:

Attribute display module, is suitable for obtaining the attribute information that described putting person identifies corresponding putting person, and shows described attribute information.

Claims (10)

1. a detection method for business object impression information, is characterized in that, comprising:

Upload the attribute information of the putting person of business object impression information in advance based on multiple history, generate the detection model of business object impression information;

Receive the detection request carrying putting person's mark;

According to the detection model of described business object impression information, the business object impression information that corresponding putting person uploads is identified to described putting person and detects.

2. the method for claim 1, is characterized in that, the described attribute information uploading the putting person of business object impression information in advance based on multiple history, and the step generating the detection model of business object impression information comprises:

Many of gathering in advance in browser show daily record and many click logs;

The attribute information that each history uploads the putting person of business object impression information is added up respectively according to described displaying daily record and described click logs;

Attribute information based on described multiple putting person generates the detection model of business object impression information.

3. method as claimed in claim 2, is characterized in that,

Described displaying daily record comprises: whether the business object of the mark of the putting person that the mark of the business object of displaying, the business object of described displaying belong to, described displaying is pushed away the massfraction of the information on a left side, the business object of described displaying;

Described click logs comprises: whether the business object of the mark of the putting person that the mark of the business object of click, the business object of described click belong to, described click is pushed away the consumption figures of the information on a left side, the business object of described click;

The attribute information of described putting person comprises: pageview, and/or left side pageview, and/or click volume, and/or left side click volume, and/or massfraction, and/or consumption figures, and/or clicking rate.

4. method as claimed in claim 3, is characterized in that, describedly adds up according to described displaying daily record and described click logs the step that each history uploads the attribute information of the putting person of business object impression information respectively and comprises:

Add up in all displaying daily records the quantity of business object that belong to same putting person, that show, using the pageview of this quantity as described putting person; And/or,

Add up in all displaying daily records and belong to same putting person and pushed away the quantity of the business object of left displaying, using the left side pageview of this quantity as described putting person; And/or,

Add up in all click logs the quantity of business object that belong to same putting person, that click, using the click volume of this quantity as described putting person; And/or,

Add up in all click logs and belong to same putting person and pushed away the quantity of the business object of left click, using the left side click volume of this quantity as described putting person; And/or,

Add up the quantity of the business object of same displaying in all displaying daily records, using the pageview of this quantity as described business object; Add up the quantity of the business object of same click in all click logs, using the click volume of this quantity as described business object; Calculate the click volume of each business object and the quotient of pageview respectively, using the clicking rate of described quotient as described business object; Calculate the mean value belonging to the clicking rate of all business objects of same putting person, as the clicking rate of described putting person; And/or,

Calculate in all displaying daily records the mean value of the massfraction of business object that belong to same putting person, that show, using the massfraction of this mean value as described putting person; And/or,

Calculate in all click logs the summation of the consumption figures of business object that belong to same putting person, that click, using the consumption figures of this summation as described putting person.

5. method as claimed in claim 2, is characterized in that, the step that the described attribute information based on described multiple putting person generates the detection model of business object impression information comprises:

For each attribute information of each putting person, calculate the Rank scores value of current attribute information in the attribute information that all putting persons are identical with this attribute information respectively;

Using the Rank scores value of all properties information of a putting person as an object, cluster is carried out to all objects;

For each cluster, carry out linear regression analysis respectively, obtain the regression parameter that current cluster is corresponding;

For each cluster, regression parameter corresponding to current cluster is utilized to determine the detection model of the business object impression information that this cluster is corresponding respectively.

6. method as claimed in claim 5, is characterized in that, describedly comprises the step that all objects carry out cluster:

Hierarchical clustering is carried out to all objects, determines the initial clustering of destination number;

The barycenter of each initial clustering of random selecting;

For each object, calculate the distance between existing object and each barycenter respectively, and existing object is referred in cluster corresponding to the barycenter minimum with the spacing of this object;

Whether the cluster that each barycenter that judgement obtains is corresponding meets the condition of convergence;

If not, then recalculate the barycenter of each initial clustering, and return described for each object, calculate the distance between existing object and each barycenter respectively, and existing object is referred to the step in cluster corresponding to the barycenter minimum with the spacing of this object;

If so, then determine that the cluster that each barycenter of obtaining is corresponding is cluster result.

7. method as claimed in claim 6, is characterized in that, describedly carries out hierarchical clustering to all objects, determines that the step of the initial clustering of destination number comprises:

Using an object as an initial clustering, calculate the distance between every two initial clusterings respectively;

An initial clustering is merged into by apart from minimum two initial clusterings;

Utilize B (k) value that initial clustering described in following formulae discovery is corresponding:

$B\left(k\right)=\underset{1}{\overset{C_k^2}{\mathrm{\Σ}}}\mathrm{interDis}+\underset{1}{\overset{k}{\mathrm{\Σ}}}\mathrm{intraDis}$

Wherein, interDis is the distance between every two initial clusterings, and intraDis is the distance sum between inner every two objects of initial clustering, and k is the quantity of initial clustering;

Calculating the distance between the initial clustering after merging and other each initial clusterings, and return and describedly will merge into the step of an initial clustering apart from minimum two initial clusterings, is till 1 until the number of initial clustering;

Search minimum B (k) value in all B (k) values, k corresponding for described minimum B (k) value initial clustering is defined as the initial clustering of destination number.

8. method as claimed in claim 6, it is characterized in that, the step whether cluster that each barycenter that described judgement obtains is corresponding meets the condition of convergence comprises:

Utilize the A value corresponding to cluster that each barycenter of obtaining described in following formulae discovery is corresponding:

$A=\min {\mathrm{\Σ}}_{i=1}^I{\mathrm{\Σ}}_{x_j\∈C_i}\mathrm{dist}{(\mathrm{center}\left(i\right),x_j)}^2$

Wherein, I is the quantity of cluster, C _ibe the combination of object in i-th cluster, x _jbe the jth object in i-th cluster, center (i) is the center of i-th cluster, and the center of i-th cluster is the mean value of all objects in i-th cluster;

The A value that before obtaining, preset times calculates, and two adjacent A values every in the A value of this calculating and the A value that calculates of front preset times are compared;

If the amplitude of variation of every two adjacent A values is all in preset range, then determine that the cluster that each barycenter of obtaining is corresponding meets the condition of convergence.

9. method as claimed in claim 5, it is characterized in that, described attribute information comprises consumption figures,

Described for each cluster, carry out linear regression analysis respectively, the step obtaining regression parameter corresponding to current cluster comprises:

For each cluster, determine the following formula that each object in current cluster is corresponding respectively:

Y _n＝β ₀+β ₁X ₁+β ₂X ₂+…+β _mX _m+e

Wherein, β ₀~ β _mfor regression parameter, X ₁~ X _mbe respectively the Rank scores value of the attribute information of the n-th object, Y _nbe the Rank scores value of the consumption figures of the n-th object, m is the quantity of the attribute information of the n-th object, and e is stochastic error;

The system of equations that the formula corresponding according to each object in current cluster forms calculates regression parameter β corresponding to current cluster ₀~ β _mvalue.

10. a pick-up unit for business object impression information, is characterized in that, comprising:

Generation module, is suitable for the attribute information of the putting person uploading business object impression information in advance based on multiple history, generates the detection model of business object impression information;

Receiver module, is suitable for receiving the detection request carrying putting person's mark;

Detection module, is suitable for the detection model according to described business object impression information, identifies the business object impression information that corresponding putting person uploads detect described putting person.