[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

CN102957722A - Network service Method and system for generating personalized recommendation - Google Patents

Network service Method and system for generating personalized recommendation Download PDF

Info

Publication number
CN102957722A
CN102957722A CN2011102438977A CN201110243897A CN102957722A CN 102957722 A CN102957722 A CN 102957722A CN 2011102438977 A CN2011102438977 A CN 2011102438977A CN 201110243897 A CN201110243897 A CN 201110243897A CN 102957722 A CN102957722 A CN 102957722A
Authority
CN
China
Prior art keywords
commending system
recommendation
content
weight
user
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN2011102438977A
Other languages
Chinese (zh)
Inventor
吴晓明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Vital Shining Software & Technology Co Ltd
Original Assignee
Vital Shining Software & Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Vital Shining Software & Technology Co Ltd filed Critical Vital Shining Software & Technology Co Ltd
Priority to CN2011102438977A priority Critical patent/CN102957722A/en
Publication of CN102957722A publication Critical patent/CN102957722A/en
Pending legal-status Critical Current

Links

Images

Landscapes

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

Multiple defects exist in an existing personalized recommendation system, for example, operation and maintenance of the personalized recommendation system are expensive. A complex personalized recommendation system is only oriented to relative large-scale enterprises generally. In addition, different recommendation system algorithms are applicable to different recommendation system environments only. The invention provides a method for personalized network service, and a personalized recommendation system can be developed according to the method. An operator of content websites recommends content for terminal users by the personalized recommendation system. The recommendation system can be optimized by personalized network service in terms of different terminal users, user sets and content websites. The personalized recommendation system tracks and records related user activities, such as commodity purchase, commodity browsing and commodity lease, analyzes collected data and performs detection and quantification to relations among special commodities. The personalized recommendation system recommends personalized content for users based on purchase history, browsing history, commodity evaluation and other data.

Description

A kind of network service method and system thereof for generating personalized recommendation
Technical field
The present invention relates to a kind of commending system, be specifically related to a kind of personalized recommendation system based on User Activity.
Background technology
Relatively large-scale and complicated website utilizes personalized recommendation system to recommend individualized content to the user usually, such as personalized commercial.Particularly, some personalized recommendation systems can track record relevant user activities such as commodity purchasing, goods browse and commodity lease, analyzes the data collected and the relation between the particular commodity.If the user accesses the certain content of certain website, such as certain commodity in the e-commerce website or certain piece of article in the news website, then the website can be shown to the information of user's related content simultaneously.
Summary of the invention
The present invention proposes a kind of method of personalized network service, can develop a kind of personalized recommendation system according to the method.Content sites operator by personalized recommendation system to terminal use's content recommendation.The personalized network service can be optimized commending system for different terminal uses, user's group and content sites.Personalized recommendation system track record relevant user activities is analyzed the data of collecting, the relation between detection and the quantitative particular commodity.Personalized recommendation system can the based target user purchase history, browsing histories, commodity evaluation and other data recommend individualized content to the user.
Description of drawings
Fig. 1 is that content recommendation is to the network environment schematic diagram of content sites;
Fig. 2 is the process schematic diagram of responding recommendation request;
Fig. 3 is the process schematic diagram of calculated recommendation system weight;
Fig. 4 is the process schematic diagram of selecting the commending system weight;
Fig. 5 is the process schematic diagram to commending system supplier defrayment;
Fig. 6 is the process schematic diagram at personalized network service registry commending system.
Embodiment
In some specific embodiments, the operating personnel of interactive computer system can alleviate Part Development personnel's task by the computer system service, such as exploitation commending system software.The computer system service refers to network service, can be by internet access.Network service can obtain the content that network service operator or supplier recommend by outside commending system.Content sites preferably applies to different proposed algorithms different recommendation environment by network service.Network service can detect the commending system of recommending environment performance the best automatically.
The content sites environment is described network service, generates or select content recommendation to the user by network service.Content sites is a kind of interactive computer system of using network service, and other forms have in interactive TV system, online service network, the shop service booth and by the system of Email to user's content recommendation.Content sites mainly comprises can be by purchasing, lease, download, describe, browse or with commodity electric catalog or the data bank of other forms consumption.Typical content sites has retail website, audio-visual download website, online travel service website, news website, Auction Site, buying portal website, friend-making website, social network sites, and the comprehensive website of above-mentioned several services can be provided simultaneously.Content sites provides many services by network service, comprises the activity of browsing that stores and safeguard the specific user, such as commodity purchasing, browse, lease, search history, and provide personalized recommendation etc. for the user.
Fig. 1 is that content recommendation is to network environment 100 schematic diagrames of content sites.Network environment 100 comprises PNS (personalized network service) 120, and the personalized network service can be applied to comprise the server system of physical server and computer equipment (not shown).PNS120 also can be applied to Internet resources such as network service.In some specific embodiments, PNS120 by outside commending system 180 preferably Recommendations to the terminal use of content sites 102.
Content sites 102 operators and developer set up and access PNS120 account.The operator operation website of content sites 102 or interactive site or the system of other types operate custom system 104 content recommendations to the terminal use by PNS120.The developer refers to develop tissue or the individual of commending system.The operator of registered content sites and commending system supplier's account data is stored in the data bank 175.Different developers provide different commending systems 180, and certain developer also can provide a plurality of commending systems 180.PNS120 crosses to the expense of user's content recommendation based on the operating position PayPal of commending system 180 in the operator of content sites 102.The developer obtains the part ultimate yield according to circumstances based on the performance of commending system 180.
Content sites 102 is communicated by letter with PNS120 by network 110a.Content sites 102 comprises Internet resources, sends the website of request by browser or other software as responding the terminal use.Custom system 104 can be computer equipment such as desktop computer, personal digital assistant (PDAs), mobile phone, set-top box, media player, laptop computer, panel computer, electronic book reader, terminating machine etc.Network 110a can be the Internet, local area network (LAN) or wide area network etc.
Content sites 102 can be moved or be provided by other operators, the supplier of different operators, supplier or non-PNS120.Commodity electric catalog or data bank in the custom system 104 common accessed content websites 102.Custom system is carried out the relevant user activities of particular commodity, and content sites 102 reports to the API that the network service interface 130 of PNS120 provides by the API request with User Activity.In some specific embodiments, content sites 102 can be used as network service client terminal and communicates by letter with PNS120.
The client end AP I module 106 of content sites 102 is carried out the API request in network service interface 130.Client end AP I module 106 comprises the component software of a plurality of execution API.Content sites 102 can utilize non-mechanism report User Activity based on API, such as the REST framework.Content sites 102 also can directly report to User Activity outside commending system 180, and in the process of report User Activity to PNS120, this step is alternative also can be additional to above-mentioned steps.
PNS120 is alternative or be additional to other modes for collecting user activity data.For example, the operator of content sites 102 can add the graphic symbol code in webpage.During custom system 104 running browser downloading web pages, browser is directly reported the page download activity to PNS120 by the graphic symbol code.Activity data report or be recorded in the following scene: the URL of user's access, the product identification system of accessed web page, identification user's terminal data, User Activity type.
In another embodiment, the operator of PNS120 can operate the e-commerce website of display of commodity electronic directory.The User Activity of e-commerce website record is used for the user activity data of supplemental content website 102 reports.In other embodiments, the terminal use can utilize custom system 104 directly to PNS120 report user activity data by browser toolbar or plug-in unit.
Network service interface 130 comprises a plurality of software modules, and the User Activity that content sites 102 is sent is transferred to data, services 140.Data, services 140 comprises a plurality of component softwares, stored user activity data in data bank 150.Data, services 140 can store the activity data relevant with the recognition system of the content sites 102 of report User Activity with terminal use's recognition system.The operator of content sites 102 determines to report to User Activity and the Activity Type of PNS120.Content sites 102 also can be by network service interface 130 batch transfer activity data.
Network service interface 130 comprises can carry out authentication, the payment assembly 132 of calculating and paying.In some specific embodiments, content sites 102 request of obtaining, stored user activity data need to be paid corresponding expense.In addition, the user buys commodity because of the recommendation of commending system and finishes transaction, and the operator of outside commending system 180 and supplier obtain corresponding income (as shown in Figure 5) by payment assembly 132.
Network service interface 130 also can be responded the request of data that content sites 102 sends.Request of data comprises to the request of the dependent merchandise inventory of particular commodity with to specific user's the request that personalized recommendation is provided.In some specific embodiments, the web-page requests that custom system 104 sends is responded in content sites 102 generated data requests.PNS120 is with specific or general target language such as the above request of extend markup language (XML) response.Content sites 102 main parsings that send request are responded, and again obtain or generate HTML or other guide by responding, and show content recommendation.For example, the website of transmission request is translated as the HTML sequence of describing custom system 104 to display of commodity with the inventory of the Recommendations of XML-based.
Network service interface 130 sends request of data to recommendation service module 160.In some specific embodiments, recommendation service module 160 is communicated by letter with a plurality of outside commending systems 180 by network 110b.External server 180 can be by operator and supplier's operation of PNS120 or other services.Outside commending system 180 can move in the server outside being independent of PNS120.For example, the operator of outside commending system 180 or supplier are responsible for determining mode and the position of commending system operation separately.In addition, commending system 180 also can move in the server of a plurality of PNS120.Be the commending system of difference PNS120 operator exploitation, commending system 180 only refers to outside commending system.Outside commending system 180 can utilize the algorithms of different generating recommendations.The developer of outside commending system 180 or the competition between the supplier can impel content sites 102 to obtain best the recommendation.
Recommendation service module 160 comprises Registering modules 162, and this module comprises a plurality of component softwares of outside commending system 180 in 160 registrations of recommendation service module.For example, outside 180 the supplier that recommends can provide the machine readable of outside commending system 180 to describe by Registering modules 162, describes to comprise the network address such as URL or identify outside commending system 180.Registering modules 162 machine readable of commending system can be described and the readable description of user is stored in the data bank 175.Registering modules 162 also can be collected the information about outside commending system 180, and content sites 102 obtains this information by PNS120 supplier's website or extranet.The supplier of content sites 102 selects outside commending system 180 to recommend (as shown in Figure 6) to receive according to circumstances.
Recommendation service 160 comprises recommends retrieval module 164, comprise a plurality of component softwares with inquiry, call outside commending system recommendation be provided.In some specific embodiments, recommend retrieval module 164 to send the recommendation of request to outside commending system 180 query response content sites 102.Recommending retrieval module 164 to send the request of obtaining recommendation can be that network service request is such as the request based on REST or SOAP.Recommendation request also can be to obtain the recognition system corresponding to terminal use of recommendation, the system of identification content sites 102, the system of identification user group (according to user's sex, age, hobby, the divisions such as occupation).In some specific embodiments, terminal use's computer equipment directly sends recommendation request.For example, the graphic symbol code embeds the webpage of content sites 102, and custom system 104 is recommended to network service interface 130 acquisition request by network 110a.
Content sites 102 or recommendation service 160 create a plurality of recognition systems.For example, recommendation service 160 can send to the terminal data that comprises user's recognition system terminal use's browser, confirms not create user's recognition system.In addition, content sites 102 can create terminal data or create user's recognition system by other mechanism.Content sites 102 or recommendation service 160 can be associated the recognition system of user's group with other user's recognition systems.The user organizes recognition system and can be associated with the user of certain content sites 102 or the user of a plurality of content sites 102.
For responding the commending system request, outside commending system 180 can be communicated by letter with data interface 170 in real time, again to obtain the user activity data in the data bank 150.The transfer of these active user activity datas need to be passed through signature and authentication, with the protection privacy of user.In addition, outside commending system 180 can be communicated by letter with data interface 170, thereby obtains the commodity data in the data bank 150.Commodity data comprises that content sites 102 offers the catalogue data of PNS120.In some specific embodiments, outside commending system 180 is by at least part of user activity data of data acquisition or commodity data in the batch transferring data interface 170.In certain specific embodiments, the batch of user activity data shifts by FTP over SSL and connects protection user data privacy.Batch data shifts off line carries out, and can carry out once every day.In some specific embodiments, outside commending system 180 directly communicates by letter to obtain user activity data or commodity data with content sites 102 by network 110.
Thereby outside commending system 180 can analysis of user activities and the commodity data content recommendation to content sites 102.Thereby different outside commending systems 180 are analyzed type and the individual subscriber data generating recommendations of different pieces of information, commodity, item property.Different outside commending systems 180 also can aim at different browsing environments, the Type of website, user's group, Activity Type and the type of merchandise provides recommendation.Outside commending system 180 can return product identification system and reflect the corresponding scores of associated recommendation intensity.Custom system 104 shows by content sites 102 recommends, and such as the request that sends according to custom system 104 content recommendation is presented in the webpage.For improving commending system supplier's brand recognition, content sites 102 can be presented at commending system supplier's brand identity near the position of content recommendation.
The recommendation results that the optimization module 166 of recommendation service module 160 can be returned based on the outside commending system 180 of various criterion optimization can have a detailed description in conjunction with Fig. 3.In some specific embodiments, optimizing module 166 is the quantitative weight of recommendation of outside commending system 180.Weight reflects the performance of outside commending system 180 according to circumstances.Content sites 102 can select to optimize the quantitative weight of module 166.Optimizing module 166 can be with the data storing of weight and outside commending system 180 in data bank 175.
Except outside commending system 180 provides the recommendation, PNS120 also can generate content the most popular with users by user activity data.Different commodity tabulations can generate dissimilar commodity and activity.Content sites 102 can obtain recommendation list by API.In some specific embodiments, except outside commending system 180, but also generating recommendations of recommendation service 160.Recommendation service 160 generates commodity to the similar mapping of commodity based on user activity data and commodity data, and by the personalized recommendation of mapping generation for the user.Outside commending system 180 is with the basis of similitude as self commending system algorithm.
In some specific embodiments, PNS120 preferably provides and recommends content sites 102, be not limited to the content sites of certain particular type, comprise batch marketing system and the search engine of e-commerce website, news website, content-aggregated website, music download website, social network sites, interactive television system, Email-based.Content sites 102 can be to terminal use's Recommendations and other forms of content of recommending based on User Activity by PNS120.For example, if the terminal use has browsed a pair of shoes in certain commercial content sites, the website then can be recommended commodity like user and this pair footwear simultaneously.Equally, if the user has browsed in certain news website about the somewhere custom news in red-letter day, the website then can be recommended other local customs of user news in red-letter day simultaneously.If the user has browsed certain retail website, the website then can provide personalized recommendation to the user according to this user's transactions history or other activities.Conversely, these recommend also can impel browsing content website of more time of user effort 102 and the final commodity of buying in the content sites.
In some specific embodiments, content sites 102 operators pass through the preferably task of load sharing developer generating recommendations systems soft ware of PNS120.And by optimizing the recommendation of outside commending system 180, PNS120 can improve the quality that content sites 102 obtains recommendation.
Except above-mentioned embodiment, part commending system 180 can be developed by the operator of PNS120 is inner, and moves in the PNS120 server.In some specific embodiments, optimisation technique is preferably quantitatively given the internal referral system with weight based on the performance of commending system.The internal referral system can compare with outside commending system.
Fig. 2 is process 200 schematic diagrames of responding recommendation request.Recommendation process 200 can be applied to PNS120, specifically is applied to recommendation service module 160.Recommendation process 200 is responded the request that content sites sends, and obtains recommendation by external browser.
Module 202 receives the recommendation request that content sites sends, and for example sends recommendation request by the API request to PNS.Content sites is recommended for specific user and commodity request, and further request is recommended to be optimized to specific user's Activity Type.
The outside commending system of module 204 inquiries can be by recommending retrieval module 164.Commending system provides the raw score of recommendation and commodity for responding request.Recommendation process 200 can provide the standardization mark with the raw score standardization.
Module 206 is determined the recommendation results that receives in limiting time.Limiting time is by definite a period of time of content sites or PNS and SLA (SLA).In some specific embodiments, if do not obtain recommendation results in the limiting time, can replace original recommendation results by other guide.Popular with users, salable or other previously selected contents that module 208 is recommended are to the user, and then recommendation process 200 stops.
If module 210 has been obtained recommendation results in limiting time, then can select the weight of the commending system of the recommendation results of returning, as can from database or question blank, selecting.Weight reflects the quality of each recommendation results according to circumstances.Recommend in advance weight generation by optimizing module 166 based on the best for specific user, user's group, commodity, Activity Type etc.Therefore, a kind of commending system also may have multiple weight.Generate and the process of selected weight has in conjunction with Fig. 3 and Fig. 4 and further describes.
Module 212 is with the raw score standardization of each commending system.The standardization mark is used for contrasting the recommendation that different commending systems provide.The standardization mark is calculated as follows:
Figure BSA00000561751000061
Standardization mark i represents i the standardization mark of recommending, and mark i represents i the raw score of recommending, and lowest fractional is commending system is observed in the limiting time minimum minute, and highest score is the best result that commending system is observed in this time period.
Module 214 combination with standard marks and commending system weight generate comprehensive mark.Therefore, the standardization mark is according to the weight adjustment, as: comprehensive mark=(standardization mark i) * (commending system weight) (2)
Standardization mark i is i the comprehensive mark of recommending, and the commending system weight is the weight that is assigned to commending system.Accordingly, the performance of high weight reflection commending system is high, and the performance of low weight reflection commending system is low.Can describe the process of Determining Weights in detail in conjunction with Fig. 3.In addition, if a plurality of commending system has been recommended same commodity, the comprehensive mark of each commending system can generate the new comprehensive mark of these commodity.
Module 216 selects to recommend subset based on comprehensive mark, as selecting the highest recommendation subset of mark.In the module 218, the content that selected recommendation and corresponding commending system meeting recording, tracking commending system are recommended.In the module 220, selected content recommendation returned content website, recommendation process 200 stops.
Fig. 3 is the schematic diagram of calculated recommendation system weight optimizing process 300.Optimizing process 300 can be applied to PNS120, specifically is applied to the optimization module 166 of recommendation service module 160.Optimizing process 300 can be recommended to optimize by weight generation based on the performance of outside commending system.Optimizing process 300 can move automatically, and weight is assigned to commending system.Optimizing process 300 also can cycling service, recomputates or again optimize weight according to the User Activity of content sites report, for example recomputates weight every day.
Optimize module 166 and calculate the weight of specific recommendations system based on the performance of commending system specific environment.The specific environment performance refers to the commending system performance for specific user, user's group, content sites etc.Optimizing process 300 can generate the specific user's weight with statistical significance.If the weight of selected certain content website is optimized module 166 and can be selected specific user's weight, particular group of users weight, specific website weight or overall weight, can have a detailed description in conjunction with Fig. 4.If without overall weight, the weight of distributing to commending system is default-weight.
Module 302 is obtained the commending system record in limiting time.User Activity and commending system record can obtain from data bank by optimizing module 166.User activity data comprises that commodity purchasing, goods browse, commodity evaluation and show events data storing are at the timestamp of data bank 150.Commending system record comprises the recommendation information that offers content sites and the commending system information (as shown in Figure 2) of content recommendation.In addition, the commending system record also can be the timestamp that commending system is recommended to the user.
The User Activity number that module 304 is determined owing to commending system is determined the recommendation request number that commending system is processed simultaneously.In some specific embodiments, when comprising that owing to the User Activity of certain specific recommendations system commending system is recommended, the commodity correlated activation that is recorded in the given time by PNS.For example, if commending system provides in a hour of recommendation, the activity that the user buys certain commodity is recorded, and then this time purchase activity is owing to this commending system.In the different embodiments, time length can be different.In some specific embodiments, relatively large-scale time window be applied to exponential decay model and the recommendation record data aggregate of User Activity, can detect the User Activity of more heterogeneous pass.Decision is to provide time period between twice recommendation to the specific user owing to the time quantum of certain commending system.
User Activity number owing to the specific recommendations system is classified based on Activity Type.For example, ten commodity purchasings, five goods browses can be owing to certain commending system.In addition, no matter all be a kind of activity polymerization owing to certain commending system User Activity Activity Type.
Can be according to the weight analysis User Activity how owing to commending system during each commending system content recommendation.Furtherly, first commending system weight is that 2, the second commending system weights are 0.5, and then 80% activity is owing to first commending system, and 20% activity is owing to second commending system.How the comprehensive recommender score analysis of user activities that other specific embodiments can provide based on commending system is owing to commending system.For example, comprehensive mark is respectively 0.9 and has recommended identical content with two commending systems of 0.6.Based on this mark, 60% and 33% activity is respectively owing to above two commending systems.
The overall weight of each commending system is calculated or recomputated to module 306.Have in conjunction with Fig. 4 and to further describe, if can't obtain more certain weights, such as the weight of user, user's group, content sites, then can use the overall weight of each commending system.In some specific embodiments, overall weight can be calculated as follows:
Figure BSA00000561751000071
Overall situation weight Commending system, typeRefer to the commending system overall situation weight of specific activities type, the attribution activity TypeRefer in limiting time the type owing to the User Activity of commending system, handled request Commending systemRefer to the number of request that commending system is processed in limiting time.Overall situation weight is stored in the database, and database can be non-relational database.
The corresponding different Activity Types of commending system have different overall weights.Content sites is optimized commending system by the different overall weight of different Activity Types based on the User Activity type.For example, content sites usually wishes based on the user browse record but not purchaser record obtains best recommendation.In addition, each commending system has independent overall weight based on all User Activities owing to commending system.
Decision module 308 determines whether commending systems have processed request enough in the content sites.For example, this module can be determined whether commending system has been processed statistical significance in the content sites number of request.For example, module determines that 10000 number of requests have statistical significance.If determine the request number statistical significance is arranged, module 310 is calculated the weight of the certain content website of corresponding each commending system.The specific website weight can utilize the equation of similar equation 3 to calculate, and the attribution Activity Type refers in special time the Activity Type owing to commending system.The specific website weight is stored in the database.
If commending system is not processed enough requests to generate the specific website weight for content sites, decision module 324 has determined whether the extra content website.If there is not the extra content website, optimizing process 300 stops.Otherwise optimizing process 300 returns module 308 and 310, for calculating the specific website weight in the other guide website.
Module 310 to decision module 312 is optimizing processs 300.Decision module 312 determines whether commending system is that particular group of users has been processed enough requests.Enough number of requests refer to possess the number of request of statistical significance.If commending system has been processed enough number of requests, module 314 is calculated as the particular group of users weight that the user organizes the commending system of generating recommendations.The particular group of users weight also can calculate by the equation of similar equation 3, and the attribution Activity Type refers to the interior user's group activity owing to commending system of limiting time.The particular group of users weight can be stored in the database.
If commending system is not processed enough requests for the user organizes, then module 320 has determined whether that the additional customer organizes.If there is not the additional customer to organize, optimizing process 300 is carried out decision module 324.Otherwise optimizing process 300 returns module 312 and 314 and is other user's batch totals calculation particular group of users weights.
Optimizing process 300 continues Executive Module 314 to 316, and decision module 316 determines whether commending system is that the specific user processes enough requests.Enough number of requests refer to possess the number of request of statistical significance.If commending system has been processed enough number of requests, module 318 is calculated the commending system of respective user certain weights and is the recommendation of user's generation.Specific user's weight also can calculate by the equation of similar equation 3, and the attribution Activity Type refers to the interior User Activity owing to commending system of limiting time.Specific user's weight is stored in the database.
If commending system is not processed enough requests for the user, module 320 has determined whether the additional customer.If there is not the additional customer, optimizing process 300 is carried out decision module 322.Otherwise optimizing process 300 returns module 316 and 318 and calculates specific user's weight for other users.Accordingly, optimizing process 300 continues to carry out until do not have additional customer, user's group and content sites.
Any request if commending system almost is untreated, commending system then can obtain default-weight.Commending system according to the default-weight content recommendation to commending system.For example, default-weight can be 0.1.
Optimization module basis indirectly user feedback is browsed the movable optimization recommendations such as commodity as buying commodity, also can be optimized according to direct user feedback.For example, the terminal use provides based on the user feedback of voting to optimizing module 166.For different user, user's group or content sites, user feedback has different forms.The type of user feedback is selected based on the performance of commending system.
Fig. 4 selects commending system weight process 400 schematic diagrames.Weight selection course 400 can be applied to PNS120, specifically is applied to the optimization module 166 of recommendation service module 160.The weight that weight selection course 400 selects optimizing process 300 to generate is responded the recommendation service module 160 that receives recommendation request.In some specific embodiments, how weight selection course 400 display optimization modules 166 return best commending system weight.
The request that module 401 receives the commending system weight.Decision module 402 determines whether the certain content website user can obtain the commending system weight.If can obtain the commending system weight, 404 of modules are returned specific user's commending system weight, and then weight selection course 400 stops.If the user can't obtain weight in the decision module 402,406 of modules continue right of execution reselection procedure process 400.
Decision module 406 determines whether user's group can obtain the commending system weight.If can obtain the commending system weight, particular group of users commending system weight is returned module 408, and then weight selects module 400 to stop.If user's group can't be obtained weight in the decision module 406, decision module 410 continues right of execution reselection procedure process 400.
Decision module 410 determines whether content sites can obtain the commending system weight.If can obtain the commending system weight, then module 412 is returned certain content recommendation of websites system weight, and then weight selection course 400 stops.If content sites can't obtain the commending system weight in the decision module 410, then module 414 continues right of execution reselection procedure process 400.
Decision module 414 judges whether to obtain commending system overall situation weight.If can obtain overall weight, then module 416 is returned commending system overall situation weight.Can't obtain commending system overall situation weight if decision module 414 is judged, then module 418 is returned the default-weight of new commending system, and then weight selection course 400 stops.
Fig. 5 is that defrayment is to the schematic diagram of recommending supplier's process 500.Payment process 500 is applied to PNS120, specifically is applied to the payment assembly 132 of network service interface 130.The commending system performance that payment process 500 provides based on supplier is to supplier's defrayment.
Module 502 recording users are movable, and User Activity can be the purchase activity or browse activity.Decision module 504 judges that User Activity is whether owing to the recommendation of commending system.If the recommendation of commending system and User Activity occur in the limiting time section, then this User Activity is owing to the recommendation of commending system.If owing to the recommendation of commending system, then payment process 500 does not stop User Activity.
In addition, content sites is paid corresponding expense in the module 506.For example, in the single purchase activity, the expense of content sites payment can be the part percentage of commodity price, constant expense and based on the constant expense of commodity price.Module 508 determines to pay commending system supplier's expense based on the commending system performance.For example, PNS charges to expense commending system supplier's account.The quality of commending system performance is based on many factors, such as the User Activity number owing to commending system, and the number of request that commending system receives etc.In some specific embodiments, the performance of commending system weight reflection commending system and the expense of paying supplier based on performance.Weight that be used for to determine expense allocation can be that content sites obtains the weight of recommendation or is used for providing the weight of recommendation.The expense of paying can be the part of commodity price or constant expense etc.
In some specific embodiments, PNS120 collects the uniform price of each recommendation and based on recommending number and weight defrayment to commending system supplier.Commending system also can be given tacit consent to free use, but specific commending system supplier can be based on the recommendation charge that provides.PNS120 can collect the expense that the part commending system obtains.
In other specific embodiments, content sites can be selected to share user activity data with the new commending system supplier who does not subscribe to.But content sites is the Free Acquisition content recommendation also, and commending system supplier and PNS obtain income by the user data of collecting online targeted advertisements.
In another embodiment, content sites obtains based on the graphic symbol small tool of HTML but not the recommendation list of service.The graphic symbol small tool comprises advertisement, and accordingly, content sites operator does not need for obtaining the recommendation defrayment.PNS operator and commending system supplier are based on their overall weight revenue sharing.
Fig. 6 is registered commending system supplier 602 and registered content sites operator 606 and PNS120 reciprocation schematic diagram.Above-mentioned interacting activity can betide the network service interface of website, extranet and PNS.In the module 610, commending system supplier 602 can develop a commending system.In the module 612, commending system supplier 602 can by the machine readable of commending system describe and the readable description of user to the description of PNS120 submission about commending system.
PNS120 registration commending system can pass through Registering modules 162 in the module 614.During the registration commending system, PNS120 can store commending system information in computer storage, such as the URL of location commending system.In addition, PNS120 can issue the readable description of user of commending system in the module 616, browses for the registration operator of content sites 102.By the registration commending system, PNS120 can connect commending system and recommendation service.Similarly, PNS120 can change commending system flexibly.
The description of announcing is presented in the commending system catalogue that can browse for content sites 606.Content sites 606 can be accessed the commending system description and be selected commending system in the module 618, and PNS120 is content sites registration commending system in the module 620.Accordingly, content sites 606 can be selected by any commending system content recommendation to content sites 606.
In addition, commending system and commending system supplier 602 tabulations can be accessed commending system supplier 602.Commending system supplier 602 tabulations form vying each other between the commending system supplier 602 based on the success rate rank of commending system algorithm.In addition, PNS120 can send performance report to commending system supplier 602.Performance report comprises owing to the User Activity of recommending, the information such as weight.
A large amount of modules of system of the present invention can be applied in the assembly such as server of software program, hardware or software module or many computers.Although this piece of explanation is described respectively multiple module, the potential logic of these module operations is identical.In addition, above-mentioned each process, assembly, algorithm embed the module of many computers or computer processor execution, or voluntarily operation.These modules can be stored in computer readable medium or the computer memory device of any type.Said process and algorithm can partly or entirely apply to concrete application flow.The result of said process and step can be stored in any computer storage.In some specific embodiments, module can dispose be used to carrying out a plurality of processors, comprises sub-processor.In addition, module can be software or nextport hardware component NextPort, such as OO component software, infrastructure component, task component, processing mode, function, attribute, flow process, subprogram, program code, driving, firmware, microcode, circuit, data, database, data structure, tabulation, variable, and the combination of said modules etc.
As shown in Figure 1, assembly 130,132,140,160,162,163,166,170 can all or part ofly be applied to the program module of computer hardware resource operation among the PNS, comprises the server system of a plurality of physical servers.Being applied to the hardware resource of PNS can colocated or according to area distribution.Data bank 150 can be used as a plurality of databases, and flat file system or other any computer data data bank are used.
Should be appreciated that the present invention is defined by the claims.The change and the modification that do not break away from skill spirit of the present invention all are included within the practical range of the present invention.

Claims (10)

1. one kind provides the system of personalized recommendation to a plurality of content sites, and this system comprises:
Network service interface: receive first request that content sites sends, the stored user activity data, user activity data comprises certain content that User Activity is corresponding and specific user's recognition system, first request receives by computer network;
Receive second request of the content recommendation of content sites transmission, second request receives by computer network;
The user activity data data bank: the user activity data that the stored contents website sends, the user activity data data bank comprises the physical computer holder;
Recommendation service: respond second request, outside commending system provides to be recommended to respond second request, and commending system supplier provides outside commending system;
Payment module: the expense that outside commending system provides recommendation is obtained in the payment of content sites operator, and the part expense is paid supplier based on the performance performance of the commending system that is provided by commending system supplier.
2. described according to claim 1, it is characterized in that payment module is further judged the performance of commending system according to user activity data.
3. described according to claim 1, it is characterized in that User Activity comprises the commodity that user selection or purchase are recommended.
4. described according to claim 1, it is characterized in that the performance of commending system is judged in the request that payment module is further processed according to commending system.
5. computer implemented method to commending system supplier defrayment comprises:
Call a plurality of commending systems by network at the content sites content recommendation to the terminal use;
The user carries out correlated activation because of the recommendation of commending system, and content sites need be paid corresponding expense;
The part reimbursement of expense is to commending system supplier.
6. described according to claim 5, it is characterized in that, determine the distribution of expense according to the weight of a plurality of commending systems.
7. described according to claim 6, it is characterized in that, based on the performance weight generation of a plurality of commending systems.
8. described according to claim 6, it is characterized in that selected weight derives from specific user's weight, particular group of users weight, certain content website weight, overall weight and default-weight.
9. described according to claim 5, the User Activity of part expense content-based website report is distributed to each commending system.
10. described according to claim 5, the part reimbursement of expense counts supplier's account to commending system supplier.
CN2011102438977A 2011-08-24 2011-08-24 Network service Method and system for generating personalized recommendation Pending CN102957722A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2011102438977A CN102957722A (en) 2011-08-24 2011-08-24 Network service Method and system for generating personalized recommendation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2011102438977A CN102957722A (en) 2011-08-24 2011-08-24 Network service Method and system for generating personalized recommendation

Publications (1)

Publication Number Publication Date
CN102957722A true CN102957722A (en) 2013-03-06

Family

ID=47765945

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2011102438977A Pending CN102957722A (en) 2011-08-24 2011-08-24 Network service Method and system for generating personalized recommendation

Country Status (1)

Country Link
CN (1) CN102957722A (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2015176652A1 (en) * 2014-05-20 2015-11-26 Tencent Technology (Shenzhen) Company Limited Network service recommendation method and apparatus
CN105392046A (en) * 2015-11-24 2016-03-09 天脉聚源（北京）科技有限公司 Method and device of interactive television system for recommending program
CN106776660A (en) * 2015-11-25 2017-05-31 阿里巴巴集团控股有限公司 A kind of information recommendation method and device
CN107087222A (en) * 2017-04-25 2017-08-22 深圳市茁壮网络股份有限公司 The display application process and system of a kind of set top box portal page
CN107360221A (en) * 2017-06-27 2017-11-17 金心东 A kind of shared ad system and method based on LBS
CN109493152A (en) * 2017-09-11 2019-03-19 宋梓莘 A kind of network data processing and matched combination method
CN109643422A (en) * 2016-07-06 2019-04-16 电子湾有限公司 Sensor-based Products Show
CN110473038A (en) * 2018-05-10 2019-11-19 北京嘀嘀无限科技发展有限公司 A kind of Products Show method, Products Show system and computer equipment
CN117726416A (en) * 2023-12-29 2024-03-19 深圳叮咚租机科技信息有限公司 Method, system and leasing platform for leasing mobile phone online

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101170430A (en) * 2007-12-06 2008-04-30 北京广视通达网络技术有限公司 An advertisement distribution method based on distributed P2P stream media platform
EP1923811A2 (en) * 2006-11-21 2008-05-21 Thomson Licensing Method and device for providing the device with access rights to access rights controlled digital content
CN101291444A (en) * 2007-04-20 2008-10-22 余光高 Method for improving conventional operating schema of content service in mobile data service and commercial schema thereof
CN101465703A (en) * 2007-12-20 2009-06-24 音乐会技术公司 Method and system for populating a content repository for an internet radio service based on a recommendation network
CN101601026A (en) * 2007-01-30 2009-12-09 索尼株式会社 Customer equipment in electric network provides the system and method for content effectively

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1923811A2 (en) * 2006-11-21 2008-05-21 Thomson Licensing Method and device for providing the device with access rights to access rights controlled digital content
CN101601026A (en) * 2007-01-30 2009-12-09 索尼株式会社 Customer equipment in electric network provides the system and method for content effectively
CN101291444A (en) * 2007-04-20 2008-10-22 余光高 Method for improving conventional operating schema of content service in mobile data service and commercial schema thereof
CN101170430A (en) * 2007-12-06 2008-04-30 北京广视通达网络技术有限公司 An advertisement distribution method based on distributed P2P stream media platform
CN101465703A (en) * 2007-12-20 2009-06-24 音乐会技术公司 Method and system for populating a content repository for an internet radio service based on a recommendation network

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9659256B2 (en) 2014-05-20 2017-05-23 Tencent Technology (Shenzhen) Company Limited Network service recommendation method and apparatus
WO2015176652A1 (en) * 2014-05-20 2015-11-26 Tencent Technology (Shenzhen) Company Limited Network service recommendation method and apparatus
US20170068900A1 (en) * 2014-05-20 2017-03-09 Tencent Technology (Shenzhen) Company Limited Network service recommendation method and apparatus
CN105392046B (en) * 2015-11-24 2019-04-26 天脉聚源（北京）科技有限公司 A kind of interactive television system recommends the method and device of program
CN105392046A (en) * 2015-11-24 2016-03-09 天脉聚源（北京）科技有限公司 Method and device of interactive television system for recommending program
CN106776660A (en) * 2015-11-25 2017-05-31 阿里巴巴集团控股有限公司 A kind of information recommendation method and device
US11507849B2 (en) 2015-11-25 2022-11-22 Advanced New Technologies Co., Ltd. Weight-coefficient-based hybrid information recommendation
CN109643422A (en) * 2016-07-06 2019-04-16 电子湾有限公司 Sensor-based Products Show
CN107087222A (en) * 2017-04-25 2017-08-22 深圳市茁壮网络股份有限公司 The display application process and system of a kind of set top box portal page
CN107087222B (en) * 2017-04-25 2019-09-27 深圳市茁壮网络股份有限公司 A kind of the display application method and system of set-top box portal page
CN107360221A (en) * 2017-06-27 2017-11-17 金心东 A kind of shared ad system and method based on LBS
CN109493152A (en) * 2017-09-11 2019-03-19 宋梓莘 A kind of network data processing and matched combination method
CN110473038A (en) * 2018-05-10 2019-11-19 北京嘀嘀无限科技发展有限公司 A kind of Products Show method, Products Show system and computer equipment
CN117726416A (en) * 2023-12-29 2024-03-19 深圳叮咚租机科技信息有限公司 Method, system and leasing platform for leasing mobile phone online
CN117726416B (en) * 2023-12-29 2024-07-30 深圳叮咚租机科技信息有限公司 Method and system for leasing mobile phone online

Similar Documents

Publication Publication Date Title
US9934510B2 (en) Architecture for distribution of advertising content and change propagation
CN102957722A (en) Network service Method and system for generating personalized recommendation
TWI529642B (en) Promotion method and equipment of product information
US20190012683A1 (en) Method for predicting purchase probability based on behavior sequence of user and apparatus for the same
US20110029382A1 (en) Automated Targeting of Information to a Website Visitor
US20090106108A1 (en) Website management method and on-line system
US20070043583A1 (en) Reward driven online system utilizing user-generated tags as a bridge to suggested links
KR20110048065A (en) System and method for online advertising using user social information
KR20190007875A (en) Method for providing marketing management data for optimization of distribution and logistic and apparatus therefor
US20130013428A1 (en) Method and apparatus for presenting offers
JP2009193465A (en) Information processor, information providing system, information processing method, and program
US7726563B2 (en) System and method for providing optimized shopping list
US20240046298A1 (en) Systems and methods for promoting transaction rewards
JP2018120337A (en) Retrieval device, retrieval method and retrieval program
KR101026544B1 (en) Method and Apparatus for ranking analysis based on artificial intelligence, and Recording medium thereof
JP2019003610A (en) Extraction device, extraction method, and extraction program
JP7157570B2 (en) server
JP6311052B1 (en) Extraction apparatus, extraction method, and extraction program
Cheng Product recommendation system design
KR20230072097A (en) System and method for recommending product based on rate of return
US20150032532A1 (en) Automated targeting of information influenced by geo-location to an application user using a mobile device
KR102563130B1 (en) Apparatus and method for providing merchandise sales page
US11799979B2 (en) Predictive retargeting system and method
US20150032540A1 (en) Automated targeting of information influenced by delivery to a user
Rustagi A near real-time personalization for eCommerce platform

Legal Events

Date Code Title Description
DD01 Delivery of document by public notice

Addressee: Vital Shining Software & Technology Co., Ltd. Wu Xiaoming

Document name: Notification of Passing Preliminary Examination of the Application for Invention

C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
DD01 Delivery of document by public notice

Addressee: Vital Shining Software & Technology Co., Ltd.

Document name: Notification of Publication and of Entering the Substantive Examination Stage of the Application for Invention

DD01 Delivery of document by public notice

Addressee: Vital Shining Software & Technology Co., Ltd.

Document name: the First Notification of an Office Action

C02 Deemed withdrawal of patent application after publication (patent law 2001)
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20130306

DD01 Delivery of document by public notice

Addressee: Wu Xiaoming

Document name: Notification that Application Deemed to be Withdrawn