CN104090894B - The method of online parallel computation recommendation information, device and server - Google Patents
The method of online parallel computation recommendation information, device and server Download PDFInfo
- Publication number
- CN104090894B CN104090894B CN201310698031.4A CN201310698031A CN104090894B CN 104090894 B CN104090894 B CN 104090894B CN 201310698031 A CN201310698031 A CN 201310698031A CN 104090894 B CN104090894 B CN 104090894B
- Authority
- CN
- China
- Prior art keywords
- information
- recommended
- user
- real time
- information list
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
Abstract
The invention discloses a kind of method of online parallel computation recommendation information, device and server, belong to Internet technical field.Comprise: obtain the application identities of first user under current made application and information bit mark, obtain the information list that each information bit mark is corresponding; Real-time online obtains real time data corresponding to application identities, pulls the historical data that application identities is corresponding, according to the prediction weight of the information to be recommended in real time data and each information list of historical data parallel computation; Often kind of prediction weight according to each information to be recommended is screened the information to be recommended in each information list; The information to be recommended filtered out in each information list is defined as the recommendation information of the information bit that information bit corresponding to each information list identifies.The real time data that the present invention is got by real-time online and the prediction weight of historical data parallel computation information to be recommended pulled, according to prediction weight determination recommendation information, make the recommendation information determined more accurate.
Description
Technical field
The present invention relates to Internet technical field, particularly a kind of method of online parallel computation recommendation information, device and server.
Background technology
Along with the high speed development of internet and perfect, compared to these traditional three large propagation media of newspaper, broadcast and TV, the network media has the advantages such as powerful real-time, dirigibility due to it and becomes the grand strategy ingredient of communications media gradually.Wherein, when information is propagated in a network, circulation way, except tradition formula of casting net is thrown in, is more adopt the data determination recommendation information that collects of a kind of basis, then the recommendation information determined is carried out the method for recommending.Because information adopts the defining method of different recommendation informations can bring different benefits when propagating, therefore, the computing method selecting more reasonably recommendation information according to actual conditions are needed.
Provide a kind of computing method of recommendation information in correlation technique, in the method, when terminal detects that user surfs the Net, terminal obtains the application identities of this user under current made application and at least one information bit mark, and obtains the related data of user.Afterwards, the application identities got, each information bit mark and the related data of this user are sent to server by terminal again.Server is according to the concern type of this user of correlation data calculation of this user, and after obtaining information list corresponding to each information bit mark, concern type according to this user is screened the information can thrown in information list, thus the information that user pays close attention to is defined as the information of each information bit recommendation.Wherein, the related data of user is the data got after terminal accumulates a period of time in off-line case.
Realizing in process of the present invention, inventor finds that said method at least exists following problem:
Related data due to user is the data got after terminal accumulates a period of time in off-line case, the real-time of the data thus got is poor, thus the concern type of the user making server determine according to the related data of user cannot reflect the concern situation of user truly, cause the Method compare that screens the information can thrown in information list according to the concern type of user coarse, fineness is inadequate, thus the recommendation information determined is accurate not.
Summary of the invention
In order to solve the problem of correlation technique, embodiments provide a kind of method of online parallel computation recommendation information, device and server.Described technical scheme is as follows:
On the one hand, provide a kind of method of online parallel computation recommendation information, described method comprises:
Obtain the application identities of first user under current made application and at least one information bit mark, obtain the information list that each information bit mark is corresponding, in described information list, comprise at least one information to be recommended;
Real-time online obtains real time data corresponding to described application identities, pull the historical data that described application identities is corresponding, and predict weight according at least one of the information to be recommended in the described real time data got and each information list of historical data parallel computation pulled, described prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
If the time calculating at least one prediction weight of the information to be recommended in each information list according to the described real time data got and the historical data pulled exceeds preset time threshold, then stop calculating operation, and obtain the prediction weight of the information each to be recommended in each information list according to current result of calculation;
Often kind of prediction weight according to each information to be recommended is screened the information to be recommended in each information list, obtains the information to be recommended filtered out in each information list;
The information to be recommended filtered out in each information list is defined as the recommendation information of the information bit that information bit corresponding to each information list identifies.
On the other hand, provide a kind of device of online parallel computation recommendation information, described device comprises:
First acquisition module, for obtaining the application identities of first user under current made application and at least one information bit mark;
Second acquisition module, for obtaining information list corresponding to each information bit mark, comprises at least one information to be recommended in described information list;
3rd acquisition module, obtains real time data corresponding to described application identities for real-time online, pulls the historical data that described application identities is corresponding;
Computing module, at least one prediction weight of the information to be recommended in the real time data got described in basis and each information list of historical data parallel computation pulled, described prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
Described computing module, also for when at least one that the real time data got described in basis and the historical data pulled calculate the information to be recommended in each information list predicts that the time of weight exceeds preset time threshold, stop calculating operation, and obtain the prediction weight of the information each to be recommended in each information list according to current result of calculation;
Screening module, screens the information to be recommended in each information list for often kind of prediction weight according to each information to be recommended, obtains the information to be recommended filtered out in each information list;
Determination module, for being defined as the recommendation information of the information bit that information bit corresponding to each information list identifies by the information to be recommended filtered out in each information list.
Additionally provide a kind of server, described server includes storer, and one or more than one program, one of them or more than one program are stored in storer, and be configured to be performed by one or more than one processor, described one or more than one routine package are containing the instruction for carrying out following operation:
Obtain the application identities of first user under current made application and at least one information bit mark, obtain the information list that each information bit mark is corresponding, in described information list, comprise at least one information to be recommended;
Real-time online obtains real time data corresponding to described application identities, pull the historical data that described application identities is corresponding, and predict weight according at least one of the information to be recommended in the described real time data got and each information list of historical data parallel computation pulled, described prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
If the time calculating at least one prediction weight of the information to be recommended in each information list according to the described real time data got and the historical data pulled exceeds preset time threshold, then stop calculating operation, and obtain the prediction weight of the information each to be recommended in each information list according to current result of calculation;
Often kind of prediction weight according to each information to be recommended is screened the information to be recommended in each information list, obtains the information to be recommended filtered out in each information list;
The information to be recommended filtered out in each information list is defined as the recommendation information of the information bit that information bit corresponding to each information list identifies.
The beneficial effect of the technical scheme that the embodiment of the present invention provides is:
By obtaining information list corresponding to each information bit mark, the real time data that application identities under current the made application of real-time online acquisition user is corresponding, pull the historical data that application identities is corresponding, the data got are made to have more real-time, and according to the prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, more accurate according to the recommendation information that often kind of prediction weight of each information to be recommended is determined, and improve computing velocity.
Accompanying drawing explanation
In order to be illustrated more clearly in the technical scheme in the embodiment of the present invention, below the accompanying drawing used required in describing embodiment is briefly described, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these accompanying drawings.
Fig. 1 is the method flow diagram of a kind of online parallel computation recommendation information that the embodiment of the present invention one provides;
Fig. 2 is the method flow diagram of a kind of online parallel computation recommendation information that the embodiment of the present invention two provides;
Fig. 3 is the schematic flow sheet of the first online parallel computation recommendation information that the embodiment of the present invention two provides;
Fig. 4 is the schematic flow sheet of the online parallel computation recommendation information of the second that the embodiment of the present invention two provides;
Fig. 5 is the schematic flow sheet of the third online parallel computation recommendation information that the embodiment of the present invention two provides;
Fig. 6 is the apparatus structure schematic diagram of a kind of online parallel computation recommendation information that the embodiment of the present invention three provides;
Fig. 7 is the structural representation of a kind of 3rd acquisition module that the embodiment of the present invention three provides;
Fig. 8 is the structural representation of a kind of server that the embodiment of the present invention four provides.
Embodiment
For making the object, technical solutions and advantages of the present invention clearly, below in conjunction with accompanying drawing, embodiment of the present invention is described further in detail.
Embodiment one
Because existing correlation technique is when determining recommendation information, the related data of the user used for terminal accumulate a period of time in off-line case after the data that get, therefore, cause the real-time of the data got poor, thus the concern type of the user making server determine according to the related data of user cannot reflect the concern situation of user truly, cause the Method compare that screens the information in information list according to the concern type of user coarse, fineness is inadequate, thus makes the recommendation information determined accurate not.
In order to prevent above-mentioned situation, embodiments provide a kind of method of online parallel computation recommendation information, the method can be applicable to server, and see Fig. 1, for server as executive agent, the method flow that the present embodiment provides comprises:
101: obtain the application identities of first user under current made application and at least one information bit mark, obtain the information list that each information bit mark is corresponding, in information list, comprise at least one information to be recommended;
102: real-time online obtains real time data corresponding to application identities, pull the historical data that application identities is corresponding, and at least one prediction weight of information to be recommended in the real time data got according to real-time online and each information list of historical data parallel computation pulled, prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
Wherein, according at least one prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, include but not limited to:
According to the prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, and according to often kind of the information to be recommended in the real time data got and the same information list of historical data parallel computation pulled prediction weight;
Wherein, predict that weight at least comprises the one in clicking rate and probability of transaction.
Wherein, real-time online obtains real time data corresponding to application identities, pulls the historical data that application identities is corresponding, includes but not limited to:
Obtain the real time data of first user according to application identities real-time online and pull the historical data of first user, and determining the current place group of first user;
Obtain the real time data of each the second user in the current place group of first user according to application identities real-time online, and pull the historical data of each the second user in the current place group of first user;
The real time data corresponding according to the Real time data acquisition application identities of each second user and first user, and obtain historical data corresponding to application identities according to the historical data of each second user and first user.
Wherein, determine the current place group of first user, include but not limited to:
According to historical data and the current place group of real time data determination first user of first user;
Or, according to the current place group of application identities determination first user.
103: often kind of prediction weight according to each information to be recommended is screened the information to be recommended in each information list, obtains the information to be recommended filtered out in each information list;
104: the recommendation information information to be recommended filtered out in each information list being defined as the information bit that information bit corresponding to each information list identifies.
As a kind of preferred embodiment, the method also comprises:
If the time calculating at least one prediction weight of the information to be recommended in each information list according to the real time data got and the historical data pulled exceeds preset time threshold, then stop calculating operation, and obtain the prediction weight of the information each to be recommended in each information list according to current result of calculation.
The method that the present embodiment provides, by obtaining information list corresponding to each information bit mark, the real time data that application identities under current the made application of real-time online acquisition user is corresponding, pull the historical data that application identities is corresponding, the data got are made to have more real-time, and according to the prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, more accurate according to the recommendation information that often kind of prediction weight of each information to be recommended is determined, and improve computing velocity.
Embodiment two
Embodiments provide a kind of method of online parallel computation recommendation information, in conjunction with the content of above-described embodiment one, the present embodiment take executive agent as server is example, is illustrated the method that the present embodiment provides.See Fig. 2, the method flow that the present embodiment provides comprises:
201: obtain the application identities of first user under current made application and at least one information bit mark, obtain the information list that each information bit mark is corresponding;
The present embodiment does not do concrete restriction to the obtain manner obtaining the application identities of first user under current made application and at least one information bit mark, includes but not limited to: the application identities of the first user got by terminal that server receiving terminal sends under current made application and at least one information bit mark.When terminal obtains the application identities of first user under current made application and at least one information bit mark, can by detecting the current first user application process that use, the information bit mark that the information bit obtained under the application identities of this application correspondence and this application process according to the application process determined is corresponding.Wherein, the corresponding application identities of each application process, the corresponding information bit mark of each information bit, information bit is designated at least one.Such as, terminal detects that the current application process used of user is QICQ.exe, terminal obtains the database file that application process QICQ.exe stores in this locality, the information bit mark that application identities corresponding to application process QICQ.exe is corresponding with the information bit under this application process is obtained from database file, application identities as got is QICQ_ID, gets two information bit marks and is respectively Table_ID1 and Table_ID2.
After terminal gets application identities under current the made application of user and information bit mark, schematic flow sheet shown in Figure 3, the access distribution that the application identities got and information bit mark are sent to server is gathered layer by terminal, realize the application identities got and information bit mark to be sent to server, receive application identities and information bit mark by server.Server can identify information list corresponding to obtaining information bit-identify according to information bit after getting application identities and information bit mark.Wherein, server can prestore information list corresponding to each information bit mark in this locality, when the list of needs obtaining information, directly can be identified at this locality according to information bit and retrieves and extract.Certainly, server can also adopt the information list that alternate manner obtaining information bit-identify is corresponding, and the present embodiment does not do concrete restriction to this.It should be noted that, when server gets multiple information bit mark, need the information bit list that repeatedly obtaining information bit-identify is corresponding, get in information list corresponding to each information bit mark and at least comprise an information to be recommended, the present embodiment does not do concrete restriction to the number of the information to be recommended comprised in information list.In addition, the information order that server can provide according to advertiser, calculates the information list that each information bit mark is corresponding.Such as, sorted by the information order of certain classification according to price, the information in the information order of first 100 of getting is as the information to be recommended in information list corresponding to one of them message identification.Certainly, can also adopt the list of alternate manner obtaining information, the present embodiment does not do concrete restriction to this yet.
Further, in order to subsequent calculations is convenient, server, when obtaining information list, can calculate often kind of customer group, the information list of each information bit.Such as, customer group is divided into unmarried crowd, married crowd and mother and baby crowd's three kinds of customer groups, calculate the information list of each information bit for often kind of customer group respectively.Certainly, also customer group can be divided further according to actual conditions, the present embodiment does not do concrete restriction to this.Because under an application, the number of information bit and position are all fixing, therefore, the information bit mark that information bit is corresponding is also fixing.Server can precompute under each customer group, the information list that different information bit mark is corresponding, as when browsing a webpage, this webpage there are two fixing information bits, corresponding information bit mark is respectively Table_ID1 and Table_ID2, when unmarried crowd browses this webpage, server can precompute the information list of information bit corresponding to these two information bit marks for unmarried crowd; When married crowd browses this webpage, server can precompute the information list of information bit corresponding to these two information bit marks for married crowd; When mother and baby crowd browses this webpage, server can precompute the information list of information bit corresponding to these two information bit marks for mother and baby crowd.Such as, for the relevant information of more amusement can be comprised in the information list of unmarried crowd, as digital product information etc.For the relevant information of more household can be comprised in the information list of married crowd, as different furniture information etc.For the product information that can comprise more infant in the information list of mother and baby crowd and be correlated with, as care product information etc.
Preliminary screening can be done by the information list corresponding to each information bit mark by above-mentioned steps, remove redundant information, thus make information to be recommended in information list can the needs of properer user, and the time of subsequent calculations cost can be saved.Certainly, according to actual conditions, this step can be done and optimize further, and the present embodiment does not do concrete restriction to the implementation procedure of this step.
202: real-time online obtains real time data corresponding to application identities, pulls the historical data that application identities is corresponding;
The present embodiment does not obtain real time data corresponding to application identities to real-time online and does concrete restriction, also concrete restriction is not done to the mode that pulls pulling historical data corresponding to application identities, include but not limited to: obtain the real time data of first user according to application identities real-time online and pull the historical data of first user, and determining the current place group of first user; Obtain the real time data of each the second user in the current place group of first user according to application identities real-time online, and pull the historical data of each the second user in the current place group of first user; The real time data corresponding according to the Real time data acquisition application identities of each second user and first user, and obtain historical data corresponding to application identities according to the historical data of each second user and first user.
Wherein, determine that the current place group of first user can adopt the following two kinds mode:
First kind of way: according to the current place group of application identities determination first user.
First user mark can be divided to corresponding group by server in advance, due to user ID corresponding to first user can be comprised in application identities, after getting user ID corresponding to the first user in application identities, the current place group of user ID determination first user that can be directly corresponding according to first user.
The second way: according to historical data and the current place group of real time data determination first user of first user.
If server has prestored the historical data of first user corresponding to this user ID, as the behavioral data of user when using this application and the representation data of first user input, when server gets application identities, then can according to the user ID in application identities, pull behavioral data and the representation data of first user corresponding to this user ID, and then the current place group of first user can be determined according to behavioral data and representation data two item number certificate.
Such as, user is divided into unmarried male group, unmarried female group, married male group and married Nv Qunsige group by server in advance, and server has prestored the representation data of first user input for " sex: man " " marital status: unmarried ".Now, after server gets application identities, the representation data of first user can be got according to the user ID of first user in application identities, can determine that first user belongs to unmarried this group of male group according to the representation data of first user.Certainly, according to application identities determination first user, current place group can also adopt alternate manner, and the present embodiment does not do concrete restriction to this.
Certainly, historical data can also comprise other more content, and the present embodiment does not do concrete restriction to this.
When pulling the historical data of user, in order to accelerate the speed pulling historical data, server can arrange a L2 cache, and comparatively internal memory is faster for this buffer memory reading speed.Therefore, can the historical data often used recently being deposited in this L2 cache, when pulling the historical data less than needing in L2 cache, can pulling from internal memory, and upgrade the historical data stored in L2 cache simultaneously.Certainly, for embody rule occasion, can also arrange the speed that three grades of buffer memorys or alternate manner accelerate to pull historical data, the present embodiment does not do concrete restriction to this.It should be noted that, the historical data stored constantly is accumulated by server and obtains, as the behavioral data of the user in historical data, server can collect user surf the Net at every turn input keyword as in the behavioral data of user.
Due to when first user first time access server system, server may not store the historical data of first user corresponding to first user mark in advance, now, and can according to the current place group of real time data determination first user of first user.The data of the current generation of application that real time data can use for first user, as currently in user watching a video, the video information of this video can be current real time data, and user is browsing commodity, and the design parameter information of these commodity also can be current real time data.Certainly, real time data also can be other content, and the present embodiment does not do concrete restriction to this yet.Server, after getting the real time data of user, can carry out calculated off-line equally, and become a certain item content in historical data with the real time data this got, the present embodiment does not do concrete restriction to this yet.
Mode according to the real time data determination first user current place group of first user with reference to the mode according to the current place group of historical data determination first user, can repeat no more herein.Certainly, can also adopt other method according to actual conditions, the present embodiment does not do concrete restriction to this.
Wherein, can determine the current place group of first user by above-mentioned steps, the second user refers to other users in the group of first user place.The real time data of each the second user in the current place group of first user is obtained at real-time online, and when pulling the historical data of each the second user in the current place group of first user, can the real time data of first user be obtained with reference to real-time online and pull the method for historical data of first user, repeat no more herein.Certainly, other method real-time online can also be adopted to obtain the real time data of each the second user in the group of first user current place and pull the historical data of each the second user in the current place group of first user, the present embodiment does not do concrete restriction to this, see the step of query caching Cache data Layer in Fig. 3.
When after the real time data and historical data of the real time data and historical data and first user of determining each the second user, preferably, owing to having redundant data unavoidably in any data, therefore, can screen the real time data of the real time data of each the second user and historical data and first user and historical data, the present embodiment does not limit concrete screening mode herein, includes but not limited to be fallen etc. by the data screening that the time is excessively of a specified duration according to the time.Certainly, according to actual conditions, can also adopt other screening modes, the present embodiment does not do concrete restriction to this.
203: according at least one prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
The present embodiment does not do concrete restriction to the account form of at least one prediction weight of the information to be recommended in the real time data got according to real-time online and each information list of historical data parallel computation pulled, include but not limited to: according to the prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, and according to often kind of the information to be recommended in the real time data got and the same information list of historical data parallel computation pulled prediction weight.
For the ease of understanding, existing citing explains above-mentioned computing method, and concrete explaination is as follows:
After server gets information list corresponding to each information bit mark, multiple calculation task can be split into according to information bit mark, as identified Table_ID1 and Table_ID2 according to 2 information bits, split into 2 calculation task Task1 and Task2 respectively, each calculation task is used for the prediction weight of the information to be recommended in information list corresponding to calculating information bit mark.Preferably, in order to save time, multiple calculation tasks of fractionation can parallel computation, and namely calculation task Task1 and Task2 can executed in parallel, and this example does not do concrete restriction to this.Wherein, prediction weight is worth for the recommendation weighing each information to be recommended, prediction weight can be divided into polytype as required, include but not limited to Information rate, probability of transaction and the degree of correlation etc., the present embodiment does not do concrete restriction to the kind of prediction weight, does not also do concrete restriction to the content of prediction weight.
Further, in order to simplify subsequent computational task, can before often kind of prediction weight of information to be recommended in each information list of parallel computation, the real time data got according to real-time online and the historical data pulled are screened the information to be recommended in information list.Such as, according to the user's representation data in historical data or user behavior data etc., the information to be recommended in information list is screened, or, real time data according to user is screened the information to be recommended in information list, this screening process can do the step of filtering distribution as access distribution in Fig. 3 gathers layer query filter condition users data etc., namely screens the information to be recommended in information list.
Because difference prediction weight correspond to different algorithms respectively, therefore, calculation task needs to split further according to algorithm.Such as, to predict that weight is for Information rate and the degree of correlation, two kinds of algorithms corresponding to prediction weight are respectively Alg1 and Alg2.Therefore, calculation task Task1 can be split as two computing unit task task 1_Alg1, Task1_Alg2 according to two kinds of algorithms.Similarly, calculation task Task2 also can be split as two computing unit task task 2_Alg1, Task2_Alg2 according to two kinds of algorithms.For the ease of understanding, now carry out relevant explanation explanation for computing unit task task 1_Alg1, what computing unit task task 1_Alg1 calculated is the Information rate that information bit identifies all information to be recommended in information list corresponding to Table_ID1.Preferably, in order to save time, the multiple computing unit tasks split according to algorithm can parallel computation, and namely computing unit task task 2_Alg1 and Task2_Alg2 also can executed in parallel, and this example does not do concrete restriction to this.
Owing at least comprising an information to be recommended in an information list, therefore, computing unit task also can split further according to the number of information to be recommended in algorithm information list.Such as, for computing unit task task 1_Alg1, if include three information to be recommended in information list, then now computing unit task task 1_Alg1 can be split as three computing unit subtasks Task1_Alg1_1, Task1_Alg1_2, Task1_Alg1_3 according to the number of information to be recommended.Similarly, computing unit task task 1_Alg2, Task2_Alg1, Task2_Alg2 also can be split as several computing unit subtasks according to the number of information to be recommended in corresponding informance list.This process can carry out the step of distributing by information to be recommended as shown in Figure 3.For the ease of understanding, now carry out relevant explanation explanation for computing unit subtask Task1_Alg1_1, what computing unit subtask Task1_Alg1_1 calculated is the Information rate that information bit identifies first information to be recommended in information list corresponding to Table_ID1.Preferably, in order to save time, the multiple computing unit subtasks split into according to the number of information in information list can parallel computation, namely computing unit subtask Task1_Alg1_1, Task1_Alg1_2, Task1_Alg1_3 equally can executed in parallel, and the present embodiment does not do concrete restriction to this.
It should be noted that, the split process of parallel computation can content in reference diagram 5, flow process is split consistent with in the split process principle of this step in Fig. 5, actual in carrying out parallel computation, fractionation flow process or method for splitting can be determined according to server arrangement, and the present embodiment does not do concrete restriction to this.
For the ease of understand, below for calculate prediction weight for computing information clicking rate, see the step of the prediction weight of computing information in Fig. 3, the process of the Information rate of a calculating information to be recommended is explained:
User can be comprised in the real time data got due to real-time online and the historical data pulled and click feedback data.Wherein, the total degree that an information that what user clicked that feedback data represents is exposes in information bit and user click the total degree of this information.The total degree * 100% that total degree/information that user clicks this information exposes in information bit is the Information rate of this information.Such as, the cumulative exposure total degree on an advertisement position of advertisement is first had to be 1000 times, and user is in the process of this advertisement exposure, the number of times clicking altogether this advertisement is 100 times, now, click feedback data according to user and can determine that the ad click rate of this advertisement is 100/1000*100%=10%.
Certainly, comprise above-mentioned Information rate, in the process that reality calculates prediction weight, can according to actual conditions, adopt different algorithms to calculate prediction weight, the present embodiment does not do concrete restriction to the computing method of prediction weight.
Further, in order to determine recommendation information quickly, do not allow user wait for the long period, during often kind of the information to be recommended in the real time data got according to real-time online and each information list of historical data parallel computation pulled prediction weight, a time threshold can be pre-set.During concrete enforcement, carry out timing at first from computation process, when accumulate overtime threshold value computing time time, now, if do not complete calculation task, then using the prediction weight of information to be recommended complete as calculated as result of calculation.Any result is not calculated if cause for some reason, now can using the information in the information recommendation list of acquiescence as the recommendation information determined.If do not pre-set the information recommendation list of acquiescence, then now illustrate that information bit that the information bit that this information list is corresponding identifies may lose the value of impression information, now, can select this information bit undercarriage from application corresponding to application identities.Certainly, can also process further according to concrete situation in a practical situation, the present embodiment does not do concrete restriction to this.
204: often kind of prediction weight according to each information to be recommended is screened the information to be recommended in each information list, obtains the information to be recommended filtered out in each information list;
After calculating often kind of prediction weight of each information to be recommended, can screen the information to be recommended in each information list according to often kind of prediction weight of each information to be recommended.Wherein, carry out can adopting with the following method in the process of screening:
In advance for often kind of prediction weight arranges a weighted value, such as, Information rate weighted value is 60%.Probability of transaction weighted value is 40% etc., the numerical value that the weighted value that often kind of each information to be recommended prediction weighted value is multiplied by correspondence obtains is superposed, obtain a fractional value, the fractional value according to each information to be recommended screens the information to be recommended in each information list.
Such as, to predict that weight is for Information rate and probability of transaction, the weighted value that Information rate accounts for is 40%, and the weighted value that probability of transaction accounts for is 60% is example.Three information to be recommended are included in information list, the Information rate of first information to be recommended is 40%, probability of transaction is 20%, the Information rate that the Information rate of second information to be recommended is 90%, probability of transaction is the 10%, three information to be recommended is 20%, probability of transaction is 80%.For the ease of calculating, the Information rate of each information to be recommended and probability of transaction can be converted to the form of fractional value.For first information to be recommended, Information rate mark is that 40%*100=40 divides, and probability of transaction mark is that 20%*100=20 divides.Often kind of each information to be recommended prediction weight is multiplied by the numerical value that corresponding weighted value obtains to superpose, obtains a final fractional value.Such as, the fractional value of first information to be recommended is that 40*40%+20*60%=28 divides, and the fractional value of second information to be recommended is that 90*40%+10*60%=42 divides, and the fractional value of the 3rd information to be recommended is that 20*40%+80*60%=56 divides.
After the fractional value determining each information to be recommended, can screen the information to be recommended in each information list according to the fractional value of each information to be recommended.Such as, can sort according to fractional value, filter out the information to be recommended that fractional value is maximum, as above-mentioned example mid-score value is 56 points to the maximum, the third namely corresponding information to be recommended.Or can also filter out first three information to be recommended of fractional value rank, the present embodiment does not do concrete restriction to the process of screening according to fractional value.
Further, in order to the recommendation information determined is more humane, after being screened the information in information list by said method, the user behavior data in the real time data that can also get according to real-time online and the historical data pulled does classification equalization operation and redirect operation to the information in the information list after screening.For the ease of understanding, be that advertisement is illustrated below with recommendation information.
Such as, by user behavior data, server can determine that user is browsing Mobile phone recently, but a user is when buying mobile phone, also may buy the marginal-product that mobile phone is relevant simultaneously, as sticking film for mobile phone, and mobile phone shell etc.Therefore, in order to make the demand of the properer user of the advertisement in the advertising listing after screening, classification equalization operation can be done to the advertisement in advertising listing.As in said circumstances, if there is no the relevant advertisements of sticking film for mobile phone in the advertising listing after screening, the advertisement of at least one sticking film for mobile phone can be added in the advertising listing after screening.Again because a user is when preparing purchase commodity, in the process of screening commodity, after having browsed commodity often, the of a sort similar commodity of these commodity of removal search, the price and the design parameter that contrast similar commodity again determine the final commodity bought, namely, in the process of contrast, before browsed commodity are browsed possibly again.Based on above-mentioned principle, in order to make, the properer user's of the advertisement in the advertising listing after screening browse custom, can do redirect operation, add the advertisement that at least one user browses recently in the advertising listing namely after screening to the advertisement in advertising listing.
Certainly, except above-mentioned screening technique, other method can also be adopted to screen the information to be recommended in each information list according to often kind of prediction weight of each information to be recommended, and the present embodiment does not do concrete restriction to often kind of prediction weight according to each information to be recommended to the screening technique that the information to be recommended in each information list is screened.And after the information in information list being screened by said method, some other Optimum Operation can also be done, the present embodiment does not do concrete restriction to the content optimized, and does not also do concrete restriction to the method optimized and subsequent step.
205: the recommendation information information to be recommended filtered out in each information list being defined as the information bit that information bit corresponding to each information list identifies.
After the information to be recommended filtered out in each information list is defined as the recommendation information of the information bit that information bit corresponding to each information list identifies by server, the recommendation information of application identities, information bit mark and correspondence can be sent to terminal.
Further, terminal is after the recommendation information receiving application identities, information bit mark and correspondence that server sends, corresponding application can be determined according to application identities, according to information bit mark comformed information position, recommendation information be shown in the information bit of correspondence.When the recommendation information of an information bit mark correspondence has multiple, multiple recommendation information can also be shown at the enterprising road wheel stream of information bit.Certainly, can also have other operation, the present embodiment does not do concrete restriction to the subsequent step of the recommendation information determining the information bit that the information bit that each information list is corresponding identifies.
It should be noted that, the steps flow chart of above-mentioned execution can reference diagram 3 or Fig. 4.Wherein, when Fig. 3 is for determining recommendation information, after first user operates terminal, the interaction flow of terminal and server.When Fig. 4 is for determining recommendation information, the treatment scheme of server internal determination recommendation information.The method that Fig. 3, Fig. 4 and the present embodiment provide is consistent in principle and step framework, in the process that reality is implemented, can any one in the method in reference diagram 3, Fig. 4 or the present embodiment, and the present embodiment does not do concrete restriction to this.
The method that the present embodiment provides, by obtaining information list corresponding to each information bit mark, the real time data that application identities under current the made application of real-time online acquisition user is corresponding, pull the historical data that application identities is corresponding, the data got are made to have more real-time, and according to the prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, more accurate according to the recommendation information that often kind of prediction weight of each information to be recommended is determined, and improve computing velocity.
Embodiment three
Embodiments provide a kind of device of online parallel computation recommendation information, the method that this device provides for performing above-described embodiment one or embodiment two.See Fig. 6, this device comprises:
First acquisition module 601, for obtaining the application identities of first user under current made application and at least one information bit mark;
Second acquisition module 602, for obtaining information list corresponding to each information bit mark, comprises at least one information to be recommended in information list;
3rd acquisition module 603, obtains real time data corresponding to application identities for real-time online, pulls the historical data that application identities is corresponding;
Computing module 604, at least one prediction weight according to the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, described prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
Screening module 605, screens the information to be recommended in each information list for often kind of prediction weight according to each information to be recommended, obtains the information to be recommended filtered out in each information list;
Determination module 606, for being defined as the recommendation information of the information bit that information bit corresponding to each information list identifies by the information to be recommended filtered out in each information list.
As a kind of preferred embodiment, computing module 604, for the prediction weight according to the information to be recommended in the real time data that gets and each information list of historical data parallel computation pulled, and according to often kind of prediction weight of the information to be recommended in the real time data got and the same information list of historical data parallel computation pulled; Wherein, predict that weight at least comprises the one in clicking rate and probability of transaction.
As a kind of preferred embodiment, computing module 604, also for when the time of at least one prediction weight calculating the information to be recommended in each information list according to the real time data got and the historical data pulled exceeds preset time threshold, stop calculating operation, and obtain the prediction weight of the information each to be recommended in each information list according to current result of calculation.
As a kind of preferred embodiment, see Fig. 7, the 3rd acquisition module 603, includes but not limited to:
First acquiring unit 6031, for obtaining the real time data of first user according to application identities real-time online and pulling the historical data of first user;
Determining unit 6032, for determining the current place group of first user;
Second acquisition unit 6033, obtains the real time data of each the second user in the current place group of first user, and pulls the historical data of each the second user in the current place group of first user for real-time online;
3rd acquiring unit 6034, for the real time data corresponding according to the Real time data acquisition application identities of each second user and first user, and obtains historical data corresponding to application identities according to the historical data of each second user and first user.
As a kind of preferred embodiment, determining unit 6032, for according to the real time data of first user and the current place group of historical data determination first user, or, according to the current place group of application identities determination first user.
The device that the present embodiment provides, by obtaining information list corresponding to each information bit mark, the real time data that application identities under current the made application of real-time online acquisition user is corresponding, pull the historical data that application identities is corresponding, the data got are made to have more real-time, and according to the prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, more accurate according to the recommendation information that often kind of prediction weight of each information to be recommended is determined, and improve computing velocity.
Embodiment four
Present embodiments provide a kind of server, this server may be used for the method performing the online parallel computation recommendation information provided in above-described embodiment.See Fig. 8, this server 800 comprises:
Server 800 can produce larger difference because of configuration or performance difference, one or more central processing units (central processing units can be comprised, CPU) 1122 (such as, one or more processors) and storer 1132, one or more store the storage medium 1130 (such as one or more mass memory units) of application program 1142 or data 1144.Wherein, storer 1132 and storage medium 1130 can be of short duration storages or store lastingly.The program being stored in storage medium 1130 can comprise one or more modules (diagram does not mark), and each module can comprise a series of command operatings in server.Further, central processing unit 1122 can be set to communicate with storage medium 1130, and server 800 performs a series of command operatings in storage medium 1130.
Server 800 can also comprise one or more power supplys 1126, one or more wired or wireless network interfaces 1150, one or more IO interface 1158, and/or, one or more operating systems 1141, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM etc.
More than one or one program is stored in storer, and is configured to be performed by more than one or one processor, and described more than one or one routine package is containing the instruction for carrying out following operation:
Obtain the application identities of first user under current made application and at least one information bit mark, obtain the information list that each information bit mark is corresponding, in information list, comprise at least one information to be recommended;
Real-time online obtains real time data corresponding to described application identities, pull the historical data that application identities is corresponding, and predict weight according at least one of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
Often kind of prediction weight according to each information to be recommended is screened the information to be recommended in each information list, obtains the information to be recommended filtered out in each information list;
The information to be recommended filtered out in each information list is defined as the recommendation information of the information bit that information bit corresponding to each information list identifies.
Suppose that above-mentioned is the first possible embodiment, then, in the embodiment that the second provided based on the embodiment that the first is possible is possible, in the storer of server, also comprise the instruction for performing following operation:
According to the prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, and according to often kind of the information to be recommended in the real time data got and the same information list of historical data parallel computation pulled prediction weight;
Wherein, predict that weight at least comprises the one in clicking rate and probability of transaction.
In the third the possible embodiment provided, in the storer of server, also comprise the instruction for performing following operation based on any one embodiment of the possible embodiment of the first or the second:
If the time calculating at least one prediction weight of the information to be recommended in each information list according to the real time data got and the historical data pulled exceeds preset time threshold, then stop calculating operation, and obtain the prediction weight of the information each to be recommended in each information list according to current result of calculation.
In the 4th kind of possible embodiment provided, in the storer of server, also comprise the instruction for performing following operation based on any one embodiment of the possible embodiment of the first or the second:
Obtain the real time data of described first user according to application identities real-time online and pull the historical data of first user, and determining the current place group of described first user;
Obtain the real time data of each the second user in the current place group of first user according to application identities real-time online, and pull the historical data of each the second user in the current place group of first user;
The real time data corresponding according to the Real time data acquisition application identities of each second user and first user, and obtain historical data corresponding to application identities according to the historical data of each second user and first user.
In the 5th kind of possible embodiment provided based on the 4th kind of possible embodiment, in the storer of server, also comprise the instruction for performing following operation:
According to historical data and the current place group of real time data determination first user of first user;
Or, according to the current place group of application identities determination first user.
Server provided by the invention, by obtaining information list corresponding to each information bit mark, the real time data that application identities under current the made application of real-time online acquisition user is corresponding, pull the historical data that application identities is corresponding, the data got are made to have more real-time, and according to the prediction weight of the information to be recommended in the real time data got and each information list of historical data parallel computation pulled, more accurate according to the recommendation information that often kind of prediction weight of each information to be recommended is determined, and improve computing velocity.
It should be noted that: the device of the online parallel computation recommendation information that above-described embodiment provides is when calculated recommendation information, only be illustrated with the division of above-mentioned each functional module, in practical application, can distribute as required and by above-mentioned functions and be completed by different functional modules, inner structure by device is divided into different functional modules, to complete all or part of function described above.In addition, the device of the online parallel computation recommendation information that above-described embodiment provides, the online server of parallel computation recommendation information and the embodiment of the method for online parallel computation recommendation information belong to same design, its specific implementation process refers to embodiment of the method, repeats no more here.
The invention described above embodiment sequence number, just to describing, does not represent the quality of embodiment.
One of ordinary skill in the art will appreciate that all or part of step realizing above-described embodiment can have been come by hardware, the hardware that also can carry out instruction relevant by program completes, described program can be stored in a kind of computer-readable recording medium, the above-mentioned storage medium mentioned can be ROM (read-only memory), disk or CD etc.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (9)
1. a method for online parallel computation recommendation information, is characterized in that, described method comprises:
Obtain the application identities of first user under current made application and at least one information bit mark, obtain the information list that each information bit mark is corresponding, in described information list, comprise at least one information to be recommended;
Real-time online obtains real time data corresponding to described application identities, pull the historical data that described application identities is corresponding, and predict weight according at least one of the information to be recommended in the described real time data got and each information list of historical data parallel computation pulled, described prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
If the time calculating at least one prediction weight of the information to be recommended in each information list according to the described real time data got and the historical data pulled exceeds preset time threshold, then stop calculating operation, and obtain the prediction weight of the information each to be recommended in each information list according to current result of calculation;
Often kind of prediction weight according to each information to be recommended is screened the information to be recommended in each information list, obtains the information to be recommended filtered out in each information list;
The information to be recommended filtered out in each information list is defined as the recommendation information of the information bit that information bit corresponding to each information list identifies.
2. method according to claim 1, is characterized in that, at least one prediction weight of the information to be recommended in the real time data got described in described basis and each information list of historical data parallel computation pulled, comprising:
According to the prediction weight of the information to be recommended in the described real time data got and each information list of historical data parallel computation pulled, and according to often kind of the information to be recommended in the described real time data got and the same information list of historical data parallel computation pulled prediction weight;
Wherein, described prediction weight at least comprises the one in clicking rate and probability of transaction.
3. method according to claim 1 and 2, is characterized in that, described real-time online obtains real time data corresponding to described application identities, pulls the historical data that described application identities is corresponding, comprising:
Obtain the real time data of described first user according to described application identities real-time online and pull the historical data of described first user, and determining the current place group of described first user;
Obtain the real time data of each the second user in the current place group of described first user according to described application identities real-time online, and pull the historical data of each the second user in the current place group of described first user;
The real time data that application identities is corresponding according to the Real time data acquisition of each second user and described first user, and obtain historical data corresponding to described application identities according to the historical data of each second user and described first user.
4. method according to claim 3, is characterized in that, describedly determines the current place group of described first user, comprising:
The current place group of described first user is determined according to the historical data of described first user and real time data;
Or, determine the current place group of described first user according to described application identities.
5. a device for online parallel computation recommendation information, is characterized in that, described device comprises:
First acquisition module, for obtaining the application identities of first user under current made application and at least one information bit mark;
Second acquisition module, for obtaining information list corresponding to each information bit mark, comprises at least one information to be recommended in described information list;
3rd acquisition module, obtains real time data corresponding to described application identities for real-time online, pulls the historical data that described application identities is corresponding;
Computing module, at least one prediction weight of the information to be recommended in the real time data got described in basis and each information list of historical data parallel computation pulled, described prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
Described computing module, also for when at least one that the real time data got described in basis and the historical data pulled calculate the information to be recommended in each information list predicts that the time of weight exceeds preset time threshold, stop calculating operation, and obtain the prediction weight of the information each to be recommended in each information list according to current result of calculation;
Screening module, screens the information to be recommended in each information list for often kind of prediction weight according to each information to be recommended, obtains the information to be recommended filtered out in each information list;
Determination module, for being defined as the recommendation information of the information bit that information bit corresponding to each information list identifies by the information to be recommended filtered out in each information list.
6. device according to claim 5, it is characterized in that, described computing module, for the prediction weight of the information to be recommended in the real time data got described in basis and each information list of historical data parallel computation pulled, and according to often kind of the information to be recommended in the described real time data got and the same information list of historical data parallel computation pulled prediction weight; Wherein, described prediction weight at least comprises the one in clicking rate and probability of transaction.
7. the device according to claim 5 or 6, is characterized in that, described 3rd acquisition module, comprising:
First acquiring unit, for obtaining the real time data of described first user and pulling the historical data of described first user according to described application identities real-time online;
Determining unit, for determining the current place group of described first user;
Second acquisition unit, obtains the real time data of each the second user in the current place group of described first user for real-time online, and pulls the historical data of each the second user in the current place group of described first user;
3rd acquiring unit, for the real time data that application identities according to the Real time data acquisition of each second user and described first user is corresponding, and obtain historical data corresponding to described application identities according to the historical data of each second user and described first user.
8. device according to claim 7, it is characterized in that, described determining unit, for determining the current place group of described first user according to the real time data of described first user and historical data, or, determine the current place group of described first user according to described application identities.
9. a server, it is characterized in that, described server includes storer, and one or more than one program, one of them or more than one program are stored in storer, and be configured to be performed by one or more than one processor, described one or more than one routine package are containing the instruction for carrying out following operation:
Obtain the application identities of first user under current made application and at least one information bit mark, obtain the information list that each information bit mark is corresponding, in described information list, comprise at least one information to be recommended;
Real-time online obtains real time data corresponding to described application identities, pull the historical data that described application identities is corresponding, and predict weight according at least one of the information to be recommended in the described real time data got and each information list of historical data parallel computation pulled, described prediction weight is for weighing the recommendation rate of specific gravity of each information to be recommended;
If the time calculating at least one prediction weight of the information to be recommended in each information list according to the described real time data got and the historical data pulled exceeds preset time threshold, then stop calculating operation, and obtain the prediction weight of the information each to be recommended in each information list according to current result of calculation;
Often kind of prediction weight according to each information to be recommended is screened the information to be recommended in each information list, obtains the information to be recommended filtered out in each information list;
The information to be recommended filtered out in each information list is defined as the recommendation information of the information bit that information bit corresponding to each information list identifies.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310698031.4A CN104090894B (en) | 2013-12-18 | 2013-12-18 | The method of online parallel computation recommendation information, device and server |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310698031.4A CN104090894B (en) | 2013-12-18 | 2013-12-18 | The method of online parallel computation recommendation information, device and server |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104090894A CN104090894A (en) | 2014-10-08 |
| CN104090894B true CN104090894B (en) | 2015-10-21 |
Family
ID=51638610
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310698031.4A Active CN104090894B (en) | 2013-12-18 | 2013-12-18 | The method of online parallel computation recommendation information, device and server |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104090894B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105630852A (en) * | 2014-11-28 | 2016-06-01 | 北京奇立软件技术有限公司 | Information inquiry method and server |
| CN106033419B (en) * | 2015-03-10 | 2020-11-06 | 深圳市腾讯计算机系统有限公司 | Method, device and system for pushing messages in real time |
| CN104867492B (en) * | 2015-05-07 | 2019-09-03 | 科大讯飞股份有限公司 | Intelligent interactive system and method |
| CN106570006B (en) * | 2015-10-08 | 2020-04-21 | 腾讯科技（深圳）有限公司 | Multimedia file recommendation system and delivery information storage method |
| CN105955969B (en) * | 2015-11-12 | 2019-08-13 | 中国银联股份有限公司 | A kind of user behavior data arrangement method and device |
| CN105654341A (en) * | 2015-12-28 | 2016-06-08 | 中国民航信息网络股份有限公司 | Aviation product recommendation system and aviation product recommendation method based on cloud service |
| CN105959374B (en) * | 2016-05-12 | 2019-05-03 | 腾讯科技（深圳）有限公司 | A kind of data recommendation method and its equipment |
| CN107038213B (en) * | 2017-02-28 | 2021-06-15 | 华为技术有限公司 | Video recommendation method and device |
| CN108810056B (en) * | 2017-05-04 | 2021-05-14 | 腾讯科技（北京）有限公司 | Information pushing method and device |
| CN110245990B (en) * | 2019-06-19 | 2023-04-07 | 北京达佳互联信息技术有限公司 | Advertisement recommendation method and device, electronic equipment and storage medium |
| CN112541119B (en) * | 2020-12-08 | 2022-07-05 | 厦门诚创网络股份有限公司 | Efficient and energy-saving small recommendation system |
| CN113254772B (en) * | 2021-05-31 | 2022-01-11 | 山东远桥信息科技有限公司 | Information pushing method based on big data |
| CN114443967B (en) * | 2022-04-08 | 2022-07-08 | 北京并行科技股份有限公司 | Similar application recommendation method, computing device and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101546484A (en) * | 2009-04-30 | 2009-09-30 | 南京航空航天大学 | Flight delay conformance analysis and forecast system based on SOA and operation method thereof |
| CN102088626A (en) * | 2009-12-02 | 2011-06-08 | Tcl集团股份有限公司 | On-line video recommendation method and video portal service system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7032689B2 (en) * | 1996-03-25 | 2006-04-25 | Halliburton Energy Services, Inc. | Method and system for predicting performance of a drilling system of a given formation |
-
2013
- 2013-12-18 CN CN201310698031.4A patent/CN104090894B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101546484A (en) * | 2009-04-30 | 2009-09-30 | 南京航空航天大学 | Flight delay conformance analysis and forecast system based on SOA and operation method thereof |
| CN102088626A (en) * | 2009-12-02 | 2011-06-08 | Tcl集团股份有限公司 | On-line video recommendation method and video portal service system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104090894A (en) | 2014-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104090894B (en) | The method of online parallel computation recommendation information, device and server | |
| US9135348B2 (en) | Method and apparatus for machine-learning based profiling | |
| CN107004205B (en) | System and method for suggesting creative types for online content items to advertisers | |
| JP5881153B2 (en) | Recommendation of correlated information | |
| CN102594870B (en) | Cloud computing platform, cloud computing system and service information publishing method for cloud computing system | |
| US8909644B2 (en) | Real-time adaptive binning | |
| TW201734914A (en) | Content recommendation method, apparatus and system | |
| CN105741134A (en) | Method and apparatus for applying cross-data-source marketing crowds to marketing | |
| CN106296305A (en) | Electric business website real-time recommendation System and method under big data environment | |
| CN110796477A (en) | Advertisement display method and device, electronic equipment and readable storage medium | |
| US20170330239A1 (en) | Methods and systems for near real-time lookalike audience expansion in ads targeting | |
| CN102194015B (en) | Retrieval information heat statistical method | |
| CN104090886A (en) | Method and device for constructing real-time portrayal of user | |
| CN101583964A (en) | Large-scale aggregating and reporting of ad data | |
| JP2014531092A (en) | Distributed data stream processing method and system | |
| Li et al. | Grocery shopping recommendations based on basket-sensitive random walk | |
| CN109074366B (en) | Gain adjustment component for computer network routing infrastructure | |
| CN106815254A (en) | A kind of data processing method and device | |
| CN105184616A (en) | Method and device for targeted delivery of business object | |
| CN102281290A (en) | Emulation system and method for a PaaS (Platform-as-a-service) cloud platform | |
| CN103605736A (en) | Method and device for processing conversion data | |
| TWI720989B (en) | Recommended method and device | |
| US20160180413A1 (en) | Methods and systems that aggregate multi-media reviews of products and services | |
| CN106156113A (en) | A kind of method of video recommendations, system and electronic equipment | |
| CN110503486B (en) | Method and device for screening advertisement strategies |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |