CN101271558A - Multi-policy commercial product recommending system based on context information - Google Patents
Multi-policy commercial product recommending system based on context information Download PDFInfo
- Publication number
- CN101271558A CN101271558A CNA2008100374964A CN200810037496A CN101271558A CN 101271558 A CN101271558 A CN 101271558A CN A2008100374964 A CNA2008100374964 A CN A2008100374964A CN 200810037496 A CN200810037496 A CN 200810037496A CN 101271558 A CN101271558 A CN 101271558A
- Authority
- CN
- China
- Prior art keywords
- user
- parts
- information
- strategy
- recommendation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a multi-strategy commodity recommendation system basing on context information. The recommendation system acquires the operation information of a user through an information acquisition part which is operated by the user, analyzes the operation action of the user and establishes the interest description model of the user. During the interaction process between the user and an electronic commerce website, a recommendation strategy fitting the present user and the context information of the system is dynamically selected according to a strategy selection rule. The recommendation strategy describes and generates a personalized commodity recommendation list according with the interest and the requirement of the user according to the interest of the user. Through the selection of the recommendation strategy, the multi-strategy commodity recommendation system basing on context information improves the adaptability of the system to various applications and system dynamic changes. And compared with the existing recommendation system, the multi-strategy commodity recommendation system basing on context information is improved in the recommendation quality, the recommendation scale and the recommendation performance.
Description
Technical field
The present invention relates to a kind of how tactful commending system of networking technology area based on contextual information, be used for analysis at e-commerce website user operation behavior, at the merchandise news demand of user's individuality, provide the service of how tactful personalized commercial information recommendation based on the contextual information of user and system.
Background technology
Along with the rapid growth of internet information service business, the information service website of ecommerce also presents the powerful situation of fast development, and e-commerce website has become the critical services platform that enterprise realizes goods marketing.Simultaneously, e-commerce website has also become the portable tool that the user carries out commodity selection and purchase.But, increase along with the e-commerce website scale, its commodity amount that provides also sharply increases, from tens of kinds of merchandise newss are provided in early days, ten hundreds of merchandise news services is provided till now, user artificially fully obtains the merchandise news that meets the demands according to the needs of individuality in the so big commodity space, produced so-called " information overload " problem.
At the information overload problem, a large amount of e-commerce websites have been disposed the commending system (Recommender System) that is used to realize the user personalized information service, commending system can be collected the operation information with recording user, and user's operation information is analyzed by certain algorithm, the various interest of digging user form the description of corresponding user interest and demand.Carry out in the mutual process with e-commerce website at user's individuality, commending system can be described (User Profile) based on the user merchandise news of website is filtered, selectively meet the merchandise news of user interest and information requirement, thereby realize individual info service to the individual initiatively recommendation of user.
The general recommend method that adopts based on pure strategy of existing commending system, common commending system mainly contain content-based commending system, cooperation recommending system, based on the commending system of user's individual information with based on the commending system of knowledge.The single strategy of recommending can guarantee certain recommendation quality under the fewer situation of number of users and the number of entry, but for large-scale recommendation task, the pure strategy recommend method is recommending there is certain defective aspect accuracy and the data processing scale, as the very difficult potential interest of finding the user of content-based commending system, cooperation recommending system is difficult to produce to the commodity score data the user more after a little while makes customer satisfaction system recommendation results etc.But there be having complementary functions to a certain degree between the different single recommendation strategies, by different commending systems is mixed, formed the mixing commending system (Hybrid Recommender) of a definite form, can make the allomeric function of system and performance obtain certain improvement.But, the suitable environment and the condition relative fixed of mixing commending system, and less than contextual information at user and system, be difficult to adapt to the dynamic change of user interest and system state, can't satisfy user's function and performance requirement, reduce the satisfaction of user the e-commerce website service.
Summary of the invention
Fundamental purpose of the present invention is the problem that exists in the existing commending system, proposes a kind of how tactful commending system based on contextual information.This commending system is at existing pure strategy commending system and mix on the basis of commending system, sets up a kind ofly based on user and system context information, has many policy selection mechanism of adaptive ability.This commending system can be according to contextual informations such as user's mode of operation and electronic business web station system states, the satisfactory recommendation strategy of dynamic auto selection, and then realize user's personalized commercial is recommended.
Related commending system is different from existing commending system among the present invention, it has improved the dynamically adapting ability of commending system, thereby the defective that solves existing commending system is used in the mixing by multiple strategy, the integral body of final raising system is recommended quality and performance, has satisfied the requirement of user to personalized commercial information service quality.
The key issue that needs among the present invention to solve is, according to the user's operation information of collecting, the interest of digging user therefrom, and along with user's operation and time change the interest model that upgrades the user, when the user capture e-commerce website, recommend method is according to current system and user's contextual information, and Dynamic Selection is fit to the recommendation strategy of contextual information, and, generate the personalized commercial recommendation information of optimizing by dynamically selected recommendation strategy at user's interest.
For achieving the above object, the main function components that comprises based on the multi-policy commercial product recommending system of contextual information has: user's operation information collecting part, user's operation information analysis component, user interest updating component, user interest is represented and memory unit, context information management parts, policy selection parts, mixed strategy recommend parts, pure strategy to recommend parts, commending system external interface parts, processed offline parts, contextual information database, policy selection rule base, user interest descriptive data base and information of goods information data storehouse.
The present invention adopts the client/server mode of operation, and various functional parts are deployed in user's client computer and e-business network site server respectively; Wherein the user's operation information collecting part works in user's client computer this locality, miscellaneous part all works in the e-commerce website server end, has reduced system as far as possible and the software and hardware of client computer is required and has relied on.
The user's operation information collecting part is responsible for collecting user's various operation informations, and by the internet user's operation information is sent to the e-business network site server, user's operation information is forwarded to commending system external interface parts by Website server again.Commending system external interface parts are used to realize all data and the message of server end commending system and e-business network site server, and realize message exchange according to the internal part of data and type of message and commending system.The user's operation information analysis component is analyzed and is excavated the user's operation information that receives from commending system external interface parts, thereby finds user's explicit and implicit expression interest, and interest information is submitted to the user interest updating component.The user interest updating component can be according to the user interest contextual information decision concrete update method to user interest information, and the notice user interest is represented and memory unit.User interest is represented can represent according to the different needs of strategy of recommending abstract user interest information with memory unit, and it is stored in the user interest descriptive data base.The context information management parts can manage the current multiple contextual information of user and system.The policy selection rule base has been stored the rule knowledge that commending system carries out policy selection, the policy selection rule is the foundation that the policy selection parts carry out policy selection, in order to guarantee that commending system has certain reusability, the policy selection rule can be customized as required by the implementor of commending system.The policy selection parts are according to the contextual information of active user and system, the policy selection rule of storing in the policy selection rule base according to customization recommends to carry out Dynamic Selection in the strategy at multiple single recommendation strategy and mixing, thereby guarantees the recommendation service quality of commending system.It is a plurality of software and hardware entities of realizing multiple single recommendation policing feature that pure strategy is recommended parts.It is the mixing recommendation function entities that are based upon on the pure strategy recommendation parts that mixed strategy is recommended parts.The processed offline parts carry out the macrooperation amount tasks such as extraction, the calculating of user's similarity and the calculating of commodity similarity of merchandise news feature in the mode of off-line.The user interest descriptive data base is used to store at difference recommends the user interest of strategy to describe.The contextual information database is represented and has been stored when select rule required user and system context information at strategy.The information of goods information data storehouse is used to preserve all merchandise newss on the current e-commerce website and feature description data of merchandise news etc.
Based on the above-mentioned functions parts, be divided into two workflows based on the course of work of the multi-policy commercial product recommending system of contextual information according to client-server, be responsible for the work of treatment of client-side and server end respectively.The workflow of client-side is:
Step 1, user start client browser;
Step 2, user capture target electronic business web site if the user visits this target electronic business web site first, are then pointed out user installation user's operation information collecting part, and obtain overall unique sign of this user from server end;
Step 3, startup user's operation information collecting part;
Step 4, listen for user operation, acquisition user's various operation informations;
Step 5, user's operation information is encoded, be sent to e-commerce website by the client network interface;
Close browser message if step 6 user does not operate for a long time or listens to, then stop the user's operation information collecting part; Otherwise, go to step 4 and continue to intercept.
The business processing of server end is based on the message-driven mode, and the message that commending system is mainly handled comprises user's operation information and recommendation list request message, and the workflow of server end is:
Step 1, the how tactful commending system of startup;
If the how tactful commending system of step 2 is to start first, then initialization is carried out in the contextual information database in the system, policy selection rule base, user interest descriptive data base and information of goods information data storehouse;
Step 3, the processed offline parts carry out the merchandise news feature with offline mode the intensive tasks such as extraction, the calculating of user's similarity and the calculating of commodity similarity that start;
Step 4, commending system external interface component awaits external message.If obtain system closing message, then execution in step 7; Ask recommendation list message if obtain the user, then execution in step 6; If obtain user's operation information, then execution in step 5;
Step 5, the user's operation information analysis component that starts, the relevant genus generating structure user's operation information of analysis user operation information, and the structuring user's operation information is transferred to the user interest updating component handle, the user interest updating component is mapped as user's operation information the update instruction of user interest model, and represent to carry out corresponding update instruction with memory unit by user interest, realization is to the concrete renewal of user interest database, and returns execution in step 4;
Step 6, startup policy selection parts, these parts are by the contextual information of contextual information database acquisition at the specific user, and call suitable pure strategy and recommend parts or mixed strategy to recommend parts according to being stored in custom strategies selective rule in the policy selection rule base, and by the personalized commercial information recommendation tabulation of latter's generation at the targeted customer, by commending system external interface parts recommendation list is returned to e-commerce server, and return execution in step 4;
All parts of step 7, end commending system stop the operation of commending system.
The present invention can realize existing pure strategy commending system and all functions of mixing commending system, than existing commending system, have the following advantages: made full use of the contextual information of user and system, improved the personalized degree of commending system, improved the recommendation quality of commending system; On existing commending system, added the Dynamic Selection of difference being recommended strategy, improved the adaptability of system various application and system dynamics variation by the selection of recommending strategy; Related policy selection rule and recommendation policy component all can customize and expand, and guarantee the reusability and the versatility of commending system; And all more existing commending system increases on recommendation quality, recommendation scale and recommendation performance.
Description of drawings
Below in conjunction with accompanying drawing the multi-policy commercial product recommending system based on contextual information that relates among the present invention is elaborated, the parts that identical in the accompanying drawings label is corresponding identical, wherein:
Fig. 1 is a structural representation of the present invention
Fig. 2 is the client workflow diagram
Fig. 3 is that the server end user interest is described the renewal workflow diagram
Fig. 4 is that the server end recommendation list generates workflow diagram
Embodiment
Fig. 1 is illustrated one-piece construction of the present invention, and all functions parts among the present invention are deployed in client and e-commerce website server end respectively.In order to guarantee the efficient of client computer, reduce dependence to client software and hardware, most of functional part all is deployed in server end, has reduced the processing pressure of client computer.Client computer 20 among the present invention is a general browser software, and the user can finish various operating functions to merchandise news object on the e-commerce server by client computer 20.Client computer 20 communicates by the Internet 30 and e-business network site server 160, and commending system external interface 40 is realized request and the response with the e-commerce website service.
User's operation information collecting part 10 works in user's client computer 20 this locality among Fig. 1, these parts 10 start when the user capture target electronic business web site automatically with the form of intelligent agent (Agent), and these parts mainly are responsible for collection, coding and the transmission to user's operation information.In order to protect user's individual privacy, the user's operation information collecting part must work within the information gathering scope of subscriber authorisation, does not allow the information source that user's unauthorized uses is illegally collected and handled.The various operation informations that 10 pairs of these parts are collected are encoded and are represented.User's operation information collecting part 10 is independent of client computer with communicating by letter of e-business network site server 160, can carry out message exchange by the Internet 30 and e-business network site server 160 separately, and by the message exchange of latter's realization with commending system external interface parts.User's operation information can be with reference to following exemplary forms:
User_ActionType{
Action_Type?action；
Action_Param?param；
}
Wherein, action represents user's operation types, and param represents the various parameters that the user operates.
Commending system external interface parts 40 work in the front end of commending system among Fig. 1, exchanges such as the data of realization commending system and e-business network site server 160 and message, message exchange between all external systems and the commending system all must realize the unification of guarantee information Interchange Format and definition by logical parts 40.In commending system, message exchange commonly used mainly comprises user's operation information, commending system administrative messag, recommendation list request and response message, will carry out following processing to different message commending system external interface parts:
If commending system external interface parts 40 are received user's operation information, these parts to user's operation information analysis component 50, and are further analyzed forwards and are handled user's operation information by the latter.
If commending system external interface parts 40 are received recommendation list request message, then these parts with forwards to policy selection parts 80, and the selection and the final recommendation list that generates of recommending strategy by the indicated state of latter's based on contextual information.
If commending system external interface parts 40 are received the recommendation list response message, then these parts are forwarded to e-business network site server 160 with message and recommendation list, determine the form of expression of recommendation list by the latter, and are sent to client.
If commending system external interface parts 40 are received the commending system administrative messag, the commending system administrative messag mainly comprises system start-up, closes, parameter setting, status poll etc., and then these parts will directly carry out corresponding bookkeeping to commending system according to the type of message.
User's operation information analysis component 50 is finished the analysis to user's operation information among Fig. 1, these parts receive from commending system external interface parts and transmit user's operation information of coming, operation information to the user excavates, therefrom find user's interest, and interest information represented, realize renewal operation that user interest is represented with the form notice user interest updating component 60 of user interest updating message at last.
After user interest updating component 60 obtains the user interest updating message among Fig. 1, these parts 60 will be according to user's beacon information subsidiary in the message, judge the mode of upgrading this user interest, and the notice user interest represents to carry out with memory unit the physical mappings and the storage of interest information.
User interest represents that the major function with memory unit 70 is that lastest imformation with particular user interests is converted into the concrete operations instruction to user interest descriptive data base 130 among Fig. 1, these parts 70 can be set up corresponding mapping relations and standard between the logical expressions of user interest and physical representation storage, and the physics and the logical expressions that guarantee user interest have certain customizability, reduce the degree of coupling between the two, it is can not produce too much influence to another person that any one in the two changes.
The physical store that user interest database 130 is realized user interest among Fig. 1, the concrete representation of user interest also has different representations according to employed recommended technology difference.Simultaneously in order to guarantee the renewable property of user interest, in the user interest database, also preserved user's existing operation note, can follow the tracks of the variation of user interest according to time response according to user's operation note, thus interest that can the more accurate representation user.
In content-based recommend method commonly used, it is that content information feature by user's access products constitutes that user's interest is described, the content characteristic of commodity can adopt the VSM vector space model to represent, for particular commodity i, can from the descriptor of these commodity, extract the feature description information of these commodity, and constitute by the commodity proper vector that a plurality of Keyword Weight constituted, its form is as follows:
fv
i＝(w
i1，w
i2，L，w
im，L，w
in)
W wherein
IjThe weight of j the feature of expression commodity i, this weight is used for portraying the importance of j feature in the commodity feature description.
For the proper vector of all commodity that the user visited, can construct user's interest characteristics vector by the mode of machine learning or simple weighted summation, the form of this proper vector is identical with the proper vector of commodity.For user i, its interest characteristics vector is as follows:
uv
i＝(x
i1，x
i2，L，x
im，L，x
in)
X wherein
IjThe weight of j the feature key of expression user i, this weight is used for portraying the importance that j feature described at user interest.
And for the cooperation recommending method, user's interest describe to be by the user scoring of commodity to be represented that the user is not the scoring operation of narrow sense to the scoring of commodity, and all users all can specific scoring be mapped as the scoring of commodity to the evaluation of commodity.For specific user i, its all scorings can constitute this user's scoring vector, and are as follows:
rv
i＝(r
i1，r
i2，L，r
im，L，r
in)
R wherein
IjExpression user i is to the score value of j commodity, and this score value has been represented the favorable rating of user to these commodity.All users' scoring vector will constitute user-commodity rating matrix, and will be as follows:
The user interest database can also be described storage with multi-form user interest according to the needs of real system, as at the recommended technology based on knowledge, can store the mapping relations of specific user's interest to commodity in the user interest database.And, can in the user interest database, store specific user's individual information at recommended technology based on user's individual information.Simultaneously, for the efficient that guarantees to recommend to predict, can also preserve in the user interest database such as user's similarity etc. with calculate relevant information.
Store commending system in the policy selection rule base 155 among Fig. 1 and carried out the rule of policy selection, the policy selection rule is recommended the important evidence of policy selection, the policy selection rule can use the IF-THEN structure that is similar to first-order predicate logic to realize, followingly lists several example rules:
IF (user of request recommendation list is new user) THEN (call based on the pure strategy of user's individual information and recommend parts)
IF (degree of rarefication＞0.99 of active user-commodity rating matrix) THEN (calling content-based single recommendation parts)
IF (recommendation precision＞0.9 of customer requirements) THEN (calling content-based cooperation recommending parts)
From the description of above-mentioned policy selection rule as seen, policy selection parts 80 are in the process of carrying out policy selection, the policy selection rule is to depend on the current contextual information of system and user, contextual information is generally all handled and is stored in the contextual information database 150 by context information management parts 110, also can be by policy selection parts 80 online calculating.Simultaneously, in order to guarantee the expandability of framework, 155 designs of policy selection rule base have interface to allow the user to customize according to the demand of system.
Tactful alternative pack 80 is the core components in the how tactful commending system framework among Fig. 1, these parts can receive the recommendation list request message of being transmitted by commending system external interface parts 40, these parts obtain overall unique sign of user from recommendation list request message, be denoted as the recommendation target with this user, the policy selection parts are according to system and user's contextual information, select to be fit to the strategy of recommendation that has most of current context state according to the policy selection rule that is stored in the policy selection rule base 155, and recommend parts 100 or mixed strategy to recommend parts 90 to finish concrete recommendation list according to policy selection result notification pure strategy to generate.
Pure strategy recommendation parts 100 are the proposed algorithm set that are made of multiple different pure strategy recommend methods among Fig. 1, comprise specific implementation in these parts for various pure strategy proposed algorithms, it can receive calling of policy selection parts, and, generate personalized recommendation tabulation at the specific user according to the relevant information that the policy selection parts are notified.Pure strategy is recommended parts 100 also can accept mixed strategy and is recommended calling of parts 90, thereby mixing, realization recommends, and are the bases of realizing recommendation so pure strategy is recommended parts.In these parts 100 proposed algorithm commonly used comprise content-based recommend method, cooperation recommending algorithm, based on userspersonal information's proposed algorithm with based on the proposed algorithm of knowledge etc.Along with to the going deep into of proposed algorithm research, can also recommend to add in the parts 100 new proposed algorithm in pure strategy.Proposed algorithm in these parts can obtain needed associated user of implementation algorithm and merchandise news from user interest descriptive data base 130 and information of goods information data storehouse 140.The recommendation list that these parts generate is responsible for recommendation results is passed to e-business network site server 160 by parts 40 with the form notice commending system external interface parts 40 of recommendation list response message.
Mixed strategy recommendation parts 90 are to comprise the multiple strategy set that the pure strategy recommend method is mixed among Fig. 1, and it has comprised and has mixed method explanation and the parameter setting of recommending.These parts can reception strategy alternative pack 80 call, and the relevant information of being notified according to parts 80 is called relevant pure strategy according to the needs of mixed strategy and is recommended parts 100, and generates the personalized recommendation tabulation at the specific user.The mixing commonly used that comprises in the mixed strategy recommendation parts recommends strategy to have weight to mix, switch mixing, combined hybrid, characteristic mixing, cascade mixing, characteristic amplification mixing and metadata mixing etc., and these parts 90 also allow the implementor of framework to add and customize new mixing recommend method.The mixing proposed algorithm that these parts 90 are realized can obtain needed associated user of implementation algorithm and merchandise news from user interest descriptive data base 130 and information of goods information data storehouse 140.The recommendation list that parts 90 generate is responsible for recommendation results is passed to e-business network site server 160 by parts 40 with the form notice commending system external interface parts 40 of recommendation list response message.
In order to improve the efficient that proposed algorithm realizes, processed offline parts 120 among Fig. 1 are handled the relevant information that proposed algorithm need be used on the backstage in the off-line operation mode, particularly the intensive task at user's similarity, commodity similarity and commodity feature extraction in the various algorithms realizes, and corresponding result is stored in user interest descriptive data base 130 and the information of goods information data storehouse 140, allow mixed strategy to recommend parts 90 and pure strategy to recommend the parts 100 online information that need of obtaining.
The major function of context information management parts 110 is expression, processing and storages of finishing various dynamic context information in the system among Fig. 1.These parts 110 can carry out the required various contextual informations of policy selection by collection strategy alternative pack 80, so these parts 110 have the capture ability that system dynamics is changed.This 110 pairs of system context information processing of parts is the result be stored in the contextual information database, to offer 80 visits of policy selection parts.Treated system and user context information have been preserved in the contextual information database 150, these information are important evidence that policy selection parts 80 carry out policy selection, and the contextual information in these parts 150 can customize according to the needs of policy selection rule.
All merchandise newss of selling in the e-commerce website have been preserved among Fig. 1 in the information of goods information data storehouse 140, mainly comprise characteristic information, the similarity between the different commodity of various descriptors, the commodity of commodity, the score data and the prediction scoring of commodity, above-mentioned information spinner will offer mixed strategy and recommend parts 90 and pure strategy to recommend parts 100 to realize that proposed algorithms use.Obtaining of merchandise news can be from e-commerce website 160, also can be from the result of processed offline parts 120.
Groundwork process of the present invention comprises: client workflow, server end user interest are described and are upgraded workflow and server end recommendation list generation workflow.Three flow processs have realized the main business process of this commending system, and Fig. 2 has described the workflow of client:
Step 1, user start client browser 170
The user starts browser according to the method for operating of client computer 20 operating system of using.
Step 2, user capture target electronic business web site 180
The user is according to the operation of client computer 20 browsers, initiates visit to target electronic business web site server 160 with the URL form of target electronic business web site.
Step 3, judge whether access destination e-commerce website 190 first
If e-business network site server 160 is not found user's beacon information in the request message of client computer 20 access destination e-commerce websites, flow process goes to step 4; Otherwise flow process goes to step 6.
Step 4, installation user's operation information collecting part 200
User's operation information collecting part 10 is downloaded and installed to e-business network site server 160 prompting client computer 20.
Step 5, obtain overall unique user and indicate 210
User's operation information collecting part 10 for new installation, these parts 10 will indicate to the unique user of the e-business network site server 160 request overall situations, and being stored in client computer 20 this locality, the user indicates the unique sign that will carry out message exchange as user and commending system.
Step 6, startup user's operation information collecting part 220
Client computer 20 browsers start user's operation information collecting part 10, and finish all initialization operations of these parts 10.
Step 7, collection user's operation information 230
User's operation information collecting part 10 is collected the various operations of subscriber authorisation with the form of intelligent agent.
Step 8, transmission user's operation information are to target electronic business web site 240
User's operation information collecting part 10 carries out simple code with the user's operation information that obtains, and is sent to target electronic business web site server 160 with channel independently.
Step 9, judge whether client computer requires to withdraw from browser 250
User's operation information collecting part 10 waiting system message judge whether to receive according to type of message to require to withdraw from the system message of browser, if receive and withdraw from message, then flow process goes to step 10; Otherwise flow process goes to step 7.
Step 10, stop user's operation information collecting part 260
Stop the intelligent agent of user's operation information collecting part 10 correspondences, finish the operation information of this session of user is collected.
Fig. 3 has described the server end user interest and has described the renewal workflow:
Step 1, commending system external interface parts are received user's operation information 270
Commending system external interface parts 40 are received user's operation information of being transmitted by e-business network site server 160, have comprised sign and corresponding user's operation information to the user in this message.
Step 2, forwarding user operation information 280
Commending system external interface parts 40 are forwarded to user's operation information analysis component 50 according to type of message with user's operation information.
Step 3, user's operation information analysis component analysis user operation information 290
User's operation information analysis component 50 is received user's operation information, the user's operation information that relates in this message is analyzed, and excavated corresponding user interest information, and notice user interest updating component 60 is carried out the user interest renewal.
Step 4, user interest updating component generate user interest updating message 300
User interest updating component 60 with reference to different user interest descriptive models, generates the updating message at the particular user interests model according to the relative users interest information, and this message has comprised user interest and upgraded information such as classification, operation and parameter.
Step 5, forwarding user interest updating message 310
User interest updating component 60 is forwarded to user interest with the user interest updating message that generates and represents memory unit 70.
Step 6, user interest are represented and memory unit operation user interest descriptive data base 320
User interest is represented with memory unit 70 the user interest renewal to be converted into 130 concrete operations of operation user interest descriptive data base, realizes the renewal that user interest is described.
Fig. 4 has described the server end recommendation list and has generated workflow:
Step 1, commending system external interface parts are received recommendation list request message 330
E-business network site server 160 sends recommendation list request message to commending system external interface parts 40, has comprised user's beacon information of request recommendation list in this message.
Step 2, commending system external interface parts are transmitted recommendation list request message 340
Commending system external interface parts 40 with the recommendation list request forwards received to policy selection parts 80.
Step 3, policy selection parts carry out policy selection 350
Policy selection parts 80 are according to the recommendation list request message that receives, from the contextual information database, obtain with the user and indicate relevant user and system context information, and carry out policy selection according to the policy selection rule in the policy selection rule base 155 and calculate.
Step 4, recommend strategy 360 according to the policy selection call by result
Policy selection parts 80 are implemented concrete recommendation strategy according to the result of calculation of step 3 and are called, if single recommendation strategy is used in the result of calculation indication, then flow process goes to step 5; Otherwise flow process goes to step 6.
Step 5, pure strategy recommend parts to generate recommendation list 370
Pure strategy recommends parts 100 according to user's beacon information, in conjunction with user interest descriptor and merchandise news, move concrete pure strategy proposed algorithm, and the personalized recommendation tabulation at the targeted customer is recommended in generation, and recommendation list is sent to commending system external interface parts 40, flow process goes to step 7.
Step 6, mixed strategy recommend parts to generate recommendation list 380
Mixed strategy recommends parts 90 according to user's beacon information, in conjunction with user interest descriptor and merchandise news, call the pure strategy proposed algorithm with selected mixed strategy, and the personalized recommendation tabulation at the targeted customer is recommended in generation, and recommendation list is sent to commending system external interface parts 40, flow process goes to step 7.
Step 7, commending system external interface are transmitted recommendation list 390
Commending system external interface parts 40 are forwarded to e-business network site server 160 with the recommendation list that receives with the form of message.
Claims (6)
1, a kind of multi-policy commercial product recommending system based on contextual information is characterized in that it comprises:
Policy selection parts are used for recommending tactful Dynamic Selection according to the current contextual information of user and system;
A contextual information database is used to store the contextual information of user and system;
A pure strategy is recommended parts, is used to realize the pure strategy proposed algorithm;
A mixed strategy is recommended parts, is used for realizing mixing proposed algorithm;
A user's operation information collecting part, these parts work in client-side with the form of intelligent agent, are used to collect user's operation information, and the user's operation information of collecting is encoded and transmitted;
Commending system external interface parts work in the e-commerce website server end, realize the interface of commending system and external system, and this interface can the process user operation information, commending system administrative messag and recommendation list request and response message;
A user's operation information analysis component works in the e-commerce website server end, realizes analysis and excavation to user's operation information;
A user interest updating component is used for the mode of determining that user interest upgrades;
A user interest is represented and memory unit, is used for user interest is carried out physical representation and storage;
A user interest descriptive data base, the interest that is used to preserve the user is described;
Context information management parts, these parts can be followed the tracks of the state of user and system, and state is stored in the contextual information database with the form of contextual information;
A policy selection rule base has been preserved the rule that the policy selection parts are recommended policy selection in this rule base, be the foundation that the policy selection parts carry out policy selection;
An information of goods information data storehouse, this database is used to preserve the various attribute informations of commodity, merchandise news in the database can be represented to use with memory unit by user interest, be used to represent user's interest description, information in this database can also be recommended parts to use by single strategy, is used to realize single proposed algorithm;
Processed offline parts, these parts are worked in the off-line operation mode, finish the macrooperation amount tasks such as extraction, the calculating of user's similarity and the calculating of commodity similarity of merchandise news feature.
2, commending system as claimed in claim 1 is characterized in that the mode of operation of commending system adopts Client, wherein:
The user's operation information collecting part works in client;
User interest descriptive data base, contextual information database, information of goods information data storehouse, policy selection rule base, policy selection parts, pure strategy are recommended parts, mixed strategy recommendation parts, commending system external interface parts, user's operation information analysis component, user interest updating component, user interest is represented and memory unit, context information management parts, processed offline parts all work in the e-commerce website server end.
3, commending system as claimed in claim 1 is characterized in that rule in the described policy selection rule base needs the support of contextual information, and the rule in the rule base can be customized.
4, commending system as claimed in claim 1 is characterized in that it is to recommend parts by calling pure strategy that described mixed strategy is recommended parts, realizes that the mixing of multiple different proposed algorithms is carried out.
5, commending system as claimed in claim 1 is characterized in that described commending system external interface parts: receive user's operation information of e-commerce website server notification, and notice user's operation information analysis component is carried out Message Processing;
Receive the recommendation list request message of e-commerce website server notification, and the notification strategy alternative pack carries out Message Processing;
Receive pure strategy and recommend parts or mix the recommendation list response message of recommending the policy component notice, and notice e-business network site server carries out Message Processing.
6, commending system as claimed in claim 1, it is characterized in that the contextual information stored in the described policy selection parts based on contextual information database, policy selection rule in the tactful selective rule storehouse is calculated, and the pure strategy of selecting to be fit to the current context state is recommended or the mixed strategy proposed algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008100374964A CN101271558A (en) | 2008-05-16 | 2008-05-16 | Multi-policy commercial product recommending system based on context information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008100374964A CN101271558A (en) | 2008-05-16 | 2008-05-16 | Multi-policy commercial product recommending system based on context information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101271558A true CN101271558A (en) | 2008-09-24 |
Family
ID=40005511
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2008100374964A Pending CN101271558A (en) | 2008-05-16 | 2008-05-16 | Multi-policy commercial product recommending system based on context information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101271558A (en) |
Cited By (51)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102194183A (en) * | 2010-03-04 | 2011-09-21 | 深圳市腾讯计算机***有限公司 | Service promotion system and method for supporting cooperative implementation of a plurality of promotional activities |
| WO2011150591A1 (en) * | 2010-06-02 | 2011-12-08 | 中兴通讯股份有限公司 | Method and apparatus for providing application list |
| CN102402765A (en) * | 2011-12-27 | 2012-04-04 | 纽海信息技术（上海）有限公司 | Electronic-commerce recommendation method based on user expression analysis |
| CN102426686A (en) * | 2011-09-29 | 2012-04-25 | 南京大学 | Internet information product recommending method based on matrix decomposition |
| CN101431485B (en) * | 2008-12-31 | 2012-07-04 | 深圳市迅雷网络技术有限公司 | Method and system for automatically recommending internet information |
| CN102567899A (en) * | 2011-12-27 | 2012-07-11 | 纽海信息技术（上海）有限公司 | Goods recommending method based on geographic information |
| CN102663019A (en) * | 2012-03-21 | 2012-09-12 | 北京英孚斯迈特信息技术有限公司 | Instant recommendation system |
| CN102722524A (en) * | 2012-05-07 | 2012-10-10 | 北京邮电大学 | Website recommendation result displaying method and device and terminal with the device |
| CN102800006A (en) * | 2012-07-23 | 2012-11-28 | 姚明东 | Real-time goods recommendation method based on customer shopping intention exploration |
| WO2013029233A1 (en) * | 2011-08-30 | 2013-03-07 | Nokia Corporation | Method and apparatus for providing recommendations based on context-aware group modeling |
| CN103064856A (en) * | 2011-10-21 | 2013-04-24 | ***通信集团重庆有限公司 | Resource recommendation method and device based on belief network |
| CN103324619A (en) * | 2012-03-20 | 2013-09-25 | 阿里巴巴集团控股有限公司 | Recommending method and recommending system based on internet |
| CN103325047A (en) * | 2012-03-22 | 2013-09-25 | 纽海信息技术（上海）有限公司 | Device and method for guiding online shopping |
| CN103489108A (en) * | 2013-08-22 | 2014-01-01 | 浙江工商大学 | Large-scale order form matching method in community commerce cloud |
| CN103632278A (en) * | 2012-08-21 | 2014-03-12 | 镇江雅迅软件有限责任公司 | Multi-strategy commodity recommendation system based on context information |
| CN103744917A (en) * | 2013-12-27 | 2014-04-23 | 东软集团股份有限公司 | Mixed recommendation method and system |
| CN103793753A (en) * | 2012-10-31 | 2014-05-14 | Ncr公司 | Techniques for forecasting retail activity |
| CN103870269A (en) * | 2012-12-17 | 2014-06-18 | 三星电子株式会社 | Method and apparatus to provide advertisement data |
| CN104025076A (en) * | 2012-12-20 | 2014-09-03 | 英特尔公司 | Provision of anonymous context information and generation of targeted content |
| CN104079640A (en) * | 2014-06-24 | 2014-10-01 | 袁蔚 | User service delivery method and system |
| CN104102696A (en) * | 2014-06-26 | 2014-10-15 | 海信集团有限公司 | Content recommendation method and device |
| CN104268760A (en) * | 2014-09-24 | 2015-01-07 | 同济大学 | User interest obtaining and transmitting method and system |
| CN104281896A (en) * | 2013-07-01 | 2015-01-14 | 富士通株式会社 | Information integrating system and information integrating method |
| CN104462373A (en) * | 2014-12-09 | 2015-03-25 | 南京大学 | Personalized recommendation engine implementing method based on multiple Agents |
| WO2015085969A1 (en) * | 2013-12-13 | 2015-06-18 | 腾讯科技（深圳）有限公司 | Recommendation algorithm optimization method, device, and system |
| CN104965897A (en) * | 2015-06-26 | 2015-10-07 | 百度在线网络技术（北京）有限公司 | Information recommendation method and device |
| CN105072139A (en) * | 2015-06-26 | 2015-11-18 | 腾讯科技（深圳）有限公司 | Recommendation method and recommendation device |
| CN105095219A (en) * | 2014-04-23 | 2015-11-25 | 华为技术有限公司 | Weibo recommendation method and terminal |
| CN105184618A (en) * | 2015-10-20 | 2015-12-23 | 广州唯品会信息科技有限公司 | Commodity individual recommendation method for new users and system |
| CN105320719A (en) * | 2015-01-16 | 2016-02-10 | 焦点科技股份有限公司 | Crowdfunding website project recommendation method based on project tag and graphical relationship |
| WO2016058153A1 (en) * | 2014-10-16 | 2016-04-21 | Yahoo! Inc. | Personalizing user interface (ui) elements |
| CN105653543A (en) * | 2014-11-11 | 2016-06-08 | 阿里巴巴集团控股有限公司 | Method and device for setting user label in information system |
| CN106934498A (en) * | 2017-03-14 | 2017-07-07 | 携程旅游网络技术（上海）有限公司 | The recommendation method and system of hotel's house type in OTA websites |
| CN106980629A (en) * | 2016-01-19 | 2017-07-25 | 华为技术有限公司 | A kind of network resource recommended method and computer equipment |
| CN107122989A (en) * | 2017-03-21 | 2017-09-01 | 浙江工业大学 | Multi-angle mixed recommendation method for cosmetics |
| CN107424043A (en) * | 2017-06-15 | 2017-12-01 | 北京三快在线科技有限公司 | A kind of Products Show method and device, electronic equipment |
| CN107526811A (en) * | 2017-08-24 | 2017-12-29 | 广东欧珀移动通信有限公司 | Information recommendation method, device, terminal device and storage medium |
| CN103608811B (en) * | 2011-06-17 | 2018-02-23 | 微软技术许可有限责任公司 | For the context-aware applications model of the equipment connected |
| CN108090810A (en) * | 2017-12-19 | 2018-05-29 | 中国科学院合肥物质科学研究院 | A kind of Products Show system based on big data |
| CN108242016A (en) * | 2018-01-25 | 2018-07-03 | 阿里巴巴集团控股有限公司 | A kind of method and apparatus of Products Show |
| CN108874998A (en) * | 2018-06-14 | 2018-11-23 | 华东师范大学 | A kind of dialog mode music recommended method indicated based on composite character vector |
| CN109064278A (en) * | 2018-07-26 | 2018-12-21 | 北京三快在线科技有限公司 | Target object recommended method and device, electronic equipment, storage medium |
| CN109062963A (en) * | 2018-06-27 | 2018-12-21 | 广州优视网络科技有限公司 | From Media Recommendation Method, device and electronic equipment |
| CN109255581A (en) * | 2017-07-13 | 2019-01-22 | Sk普兰尼特有限公司 | For providing the method and device thereof of personalized ventilation system based on Email |
| CN109344328A (en) * | 2018-09-21 | 2019-02-15 | 百度在线网络技术（北京）有限公司 | Obtain the method and device of recommender system best parameter group |
| CN109461024A (en) * | 2018-10-17 | 2019-03-12 | 浙江忧点网络科技有限公司 | A kind of recommender system of prepackaged food marking system |
| CN111125518A (en) * | 2019-12-10 | 2020-05-08 | 海尔优家智能科技（北京）有限公司 | System and method for recommending household appliance information |
| CN107463615B (en) * | 2017-07-03 | 2020-06-09 | 天津科技大学 | Real-time going and dealing recommendation method based on context and user interest in open network |
| CN111339411A (en) * | 2020-02-21 | 2020-06-26 | 京东数字科技控股有限公司 | Method, device, equipment and storage medium for controlling use of recommendation strategy |
| CN111460279A (en) * | 2020-02-25 | 2020-07-28 | 拉扎斯网络科技（上海）有限公司 | Information recommendation method and device, storage medium and computer equipment |
| CN112055075A (en) * | 2020-09-02 | 2020-12-08 | 韦张恒 | Internet product monitoring method, device, server and readable storage medium |
-
2008
- 2008-05-16 CN CNA2008100374964A patent/CN101271558A/en active Pending
Cited By (75)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101431485B (en) * | 2008-12-31 | 2012-07-04 | 深圳市迅雷网络技术有限公司 | Method and system for automatically recommending internet information |
| CN102194183A (en) * | 2010-03-04 | 2011-09-21 | 深圳市腾讯计算机***有限公司 | Service promotion system and method for supporting cooperative implementation of a plurality of promotional activities |
| CN102194183B (en) * | 2010-03-04 | 2015-09-16 | 深圳市腾讯计算机***有限公司 | Support service promotion system and the method for multiple popularization activity coordinated implementation |
| US9460450B2 (en) | 2010-06-02 | 2016-10-04 | Zte Corporation | Method and apparatus for providing application list |
| WO2011150591A1 (en) * | 2010-06-02 | 2011-12-08 | 中兴通讯股份有限公司 | Method and apparatus for providing application list |
| CN103608811B (en) * | 2011-06-17 | 2018-02-23 | 微软技术许可有限责任公司 | For the context-aware applications model of the equipment connected |
| WO2013029233A1 (en) * | 2011-08-30 | 2013-03-07 | Nokia Corporation | Method and apparatus for providing recommendations based on context-aware group modeling |
| CN103890812A (en) * | 2011-08-30 | 2014-06-25 | 诺基亚公司 | Method and apparatus for providing recommendations based on context-aware group modeling |
| CN102426686A (en) * | 2011-09-29 | 2012-04-25 | 南京大学 | Internet information product recommending method based on matrix decomposition |
| CN103064856B (en) * | 2011-10-21 | 2016-03-30 | ***通信集团重庆有限公司 | A kind of resource recommendation method based on Belief Network and device |
| CN103064856A (en) * | 2011-10-21 | 2013-04-24 | ***通信集团重庆有限公司 | Resource recommendation method and device based on belief network |
| CN102567899A (en) * | 2011-12-27 | 2012-07-11 | 纽海信息技术（上海）有限公司 | Goods recommending method based on geographic information |
| CN102402765A (en) * | 2011-12-27 | 2012-04-04 | 纽海信息技术（上海）有限公司 | Electronic-commerce recommendation method based on user expression analysis |
| CN102402765B (en) * | 2011-12-27 | 2017-07-28 | 北京京东尚科信息技术有限公司 | Electronic commerce recommending method based on user's Expression analysis |
| CN103324619A (en) * | 2012-03-20 | 2013-09-25 | 阿里巴巴集团控股有限公司 | Recommending method and recommending system based on internet |
| CN103324619B (en) * | 2012-03-20 | 2016-10-05 | 阿里巴巴集团控股有限公司 | A kind of recommendation method based on the Internet and commending system |
| CN102663019A (en) * | 2012-03-21 | 2012-09-12 | 北京英孚斯迈特信息技术有限公司 | Instant recommendation system |
| CN102663019B (en) * | 2012-03-21 | 2017-06-20 | 北京英孚斯迈特信息技术有限公司 | A kind of instant commending system |
| CN103325047B (en) * | 2012-03-22 | 2018-01-16 | 北京京东尚科信息技术有限公司 | Net purchase guide device and method |
| CN103325047A (en) * | 2012-03-22 | 2013-09-25 | 纽海信息技术（上海）有限公司 | Device and method for guiding online shopping |
| CN102722524A (en) * | 2012-05-07 | 2012-10-10 | 北京邮电大学 | Website recommendation result displaying method and device and terminal with the device |
| CN102800006A (en) * | 2012-07-23 | 2012-11-28 | 姚明东 | Real-time goods recommendation method based on customer shopping intention exploration |
| CN103632278A (en) * | 2012-08-21 | 2014-03-12 | 镇江雅迅软件有限责任公司 | Multi-strategy commodity recommendation system based on context information |
| CN103793753A (en) * | 2012-10-31 | 2014-05-14 | Ncr公司 | Techniques for forecasting retail activity |
| CN103870269A (en) * | 2012-12-17 | 2014-06-18 | 三星电子株式会社 | Method and apparatus to provide advertisement data |
| CN104025076B (en) * | 2012-12-20 | 2018-02-13 | 英特尔公司 | The offer of anonymous contextual information and the generation of specific aim content |
| CN104025076A (en) * | 2012-12-20 | 2014-09-03 | 英特尔公司 | Provision of anonymous context information and generation of targeted content |
| CN104281896A (en) * | 2013-07-01 | 2015-01-14 | 富士通株式会社 | Information integrating system and information integrating method |
| CN103489108A (en) * | 2013-08-22 | 2014-01-01 | 浙江工商大学 | Large-scale order form matching method in community commerce cloud |
| WO2015085969A1 (en) * | 2013-12-13 | 2015-06-18 | 腾讯科技（深圳）有限公司 | Recommendation algorithm optimization method, device, and system |
| CN103744917B (en) * | 2013-12-27 | 2018-04-06 | 东软集团股份有限公司 | Method and system are recommended in mixing |
| CN103744917A (en) * | 2013-12-27 | 2014-04-23 | 东软集团股份有限公司 | Mixed recommendation method and system |
| CN105095219B (en) * | 2014-04-23 | 2019-02-01 | 华为技术有限公司 | Micro-blog recommendation method and terminal |
| CN105095219A (en) * | 2014-04-23 | 2015-11-25 | 华为技术有限公司 | Weibo recommendation method and terminal |
| CN104079640B (en) * | 2014-06-24 | 2017-12-12 | 袁蔚 | A kind of user service provides method and system |
| CN104079640A (en) * | 2014-06-24 | 2014-10-01 | 袁蔚 | User service delivery method and system |
| CN104102696A (en) * | 2014-06-26 | 2014-10-15 | 海信集团有限公司 | Content recommendation method and device |
| CN104268760A (en) * | 2014-09-24 | 2015-01-07 | 同济大学 | User interest obtaining and transmitting method and system |
| CN104268760B (en) * | 2014-09-24 | 2017-06-13 | 同济大学 | A kind of user interest is obtained and transmission method and its system |
| WO2016058153A1 (en) * | 2014-10-16 | 2016-04-21 | Yahoo! Inc. | Personalizing user interface (ui) elements |
| CN105653543A (en) * | 2014-11-11 | 2016-06-08 | 阿里巴巴集团控股有限公司 | Method and device for setting user label in information system |
| CN104462373A (en) * | 2014-12-09 | 2015-03-25 | 南京大学 | Personalized recommendation engine implementing method based on multiple Agents |
| CN105320719A (en) * | 2015-01-16 | 2016-02-10 | 焦点科技股份有限公司 | Crowdfunding website project recommendation method based on project tag and graphical relationship |
| CN105072139B (en) * | 2015-06-26 | 2018-09-04 | 腾讯科技（深圳）有限公司 | Recommend method and apparatus |
| CN104965897A (en) * | 2015-06-26 | 2015-10-07 | 百度在线网络技术（北京）有限公司 | Information recommendation method and device |
| CN105072139A (en) * | 2015-06-26 | 2015-11-18 | 腾讯科技（深圳）有限公司 | Recommendation method and recommendation device |
| CN105184618A (en) * | 2015-10-20 | 2015-12-23 | 广州唯品会信息科技有限公司 | Commodity individual recommendation method for new users and system |
| US10839032B2 (en) | 2016-01-19 | 2020-11-17 | Huawei Technologies Co., Ltd. | Network resource recommendation method and computer device |
| CN106980629A (en) * | 2016-01-19 | 2017-07-25 | 华为技术有限公司 | A kind of network resource recommended method and computer equipment |
| CN106934498A (en) * | 2017-03-14 | 2017-07-07 | 携程旅游网络技术（上海）有限公司 | The recommendation method and system of hotel's house type in OTA websites |
| CN107122989A (en) * | 2017-03-21 | 2017-09-01 | 浙江工业大学 | Multi-angle mixed recommendation method for cosmetics |
| CN107122989B (en) * | 2017-03-21 | 2021-06-18 | 浙江工业大学 | Multi-angle mixed recommendation method for cosmetics |
| CN107424043B (en) * | 2017-06-15 | 2021-06-25 | 北京三快在线科技有限公司 | Product recommendation method and device and electronic equipment |
| CN107424043A (en) * | 2017-06-15 | 2017-12-01 | 北京三快在线科技有限公司 | A kind of Products Show method and device, electronic equipment |
| CN107463615B (en) * | 2017-07-03 | 2020-06-09 | 天津科技大学 | Real-time going and dealing recommendation method based on context and user interest in open network |
| CN109255581B (en) * | 2017-07-13 | 2023-11-14 | Sk普兰尼特有限公司 | Method for providing personalized recommendation service based on e-mail and apparatus therefor |
| CN109255581A (en) * | 2017-07-13 | 2019-01-22 | Sk普兰尼特有限公司 | For providing the method and device thereof of personalized ventilation system based on Email |
| CN107526811A (en) * | 2017-08-24 | 2017-12-29 | 广东欧珀移动通信有限公司 | Information recommendation method, device, terminal device and storage medium |
| CN108090810A (en) * | 2017-12-19 | 2018-05-29 | 中国科学院合肥物质科学研究院 | A kind of Products Show system based on big data |
| CN108242016A (en) * | 2018-01-25 | 2018-07-03 | 阿里巴巴集团控股有限公司 | A kind of method and apparatus of Products Show |
| CN108874998B (en) * | 2018-06-14 | 2021-10-19 | 华东师范大学 | Conversational music recommendation method based on mixed feature vector representation |
| CN108874998A (en) * | 2018-06-14 | 2018-11-23 | 华东师范大学 | A kind of dialog mode music recommended method indicated based on composite character vector |
| CN109062963A (en) * | 2018-06-27 | 2018-12-21 | 广州优视网络科技有限公司 | From Media Recommendation Method, device and electronic equipment |
| CN109062963B (en) * | 2018-06-27 | 2021-06-04 | 阿里巴巴（中国）有限公司 | Self-media recommendation method and device and electronic equipment |
| CN109064278A (en) * | 2018-07-26 | 2018-12-21 | 北京三快在线科技有限公司 | Target object recommended method and device, electronic equipment, storage medium |
| CN109064278B (en) * | 2018-07-26 | 2021-04-02 | 北京三快在线科技有限公司 | Target object recommendation method and device, electronic equipment and storage medium |
| CN109344328B (en) * | 2018-09-21 | 2021-01-05 | 百度在线网络技术（北京）有限公司 | Method and device for obtaining optimal parameter combination of recommendation system |
| CN109344328A (en) * | 2018-09-21 | 2019-02-15 | 百度在线网络技术（北京）有限公司 | Obtain the method and device of recommender system best parameter group |
| CN109461024A (en) * | 2018-10-17 | 2019-03-12 | 浙江忧点网络科技有限公司 | A kind of recommender system of prepackaged food marking system |
| CN111125518A (en) * | 2019-12-10 | 2020-05-08 | 海尔优家智能科技（北京）有限公司 | System and method for recommending household appliance information |
| CN111125518B (en) * | 2019-12-10 | 2023-12-22 | 海尔优家智能科技（北京）有限公司 | Household appliance information recommendation system and method |
| CN111339411A (en) * | 2020-02-21 | 2020-06-26 | 京东数字科技控股有限公司 | Method, device, equipment and storage medium for controlling use of recommendation strategy |
| CN111339411B (en) * | 2020-02-21 | 2023-09-29 | 京东科技控股股份有限公司 | Method, device, equipment and storage medium for controlling recommendation policy use |
| CN111460279A (en) * | 2020-02-25 | 2020-07-28 | 拉扎斯网络科技（上海）有限公司 | Information recommendation method and device, storage medium and computer equipment |
| CN112055075A (en) * | 2020-09-02 | 2020-12-08 | 韦张恒 | Internet product monitoring method, device, server and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101271558A (en) | Multi-policy commercial product recommending system based on context information | |
| US11188928B2 (en) | Marketing method and apparatus based on deep reinforcement learning | |
| CN110162693B (en) | Information recommendation method and server | |
| Raghuwanshi et al. | Collaborative filtering techniques in recommendation systems | |
| CN102708131B (en) | By consumer's automatic classification in fine point | |
| CN102737334B (en) | Micro-segment definition system | |
| Gozhyj et al. | Web resources management method based on intelligent technologies | |
| CA2756198C (en) | Digital analytics system | |
| KR20200003106A (en) | Information retrieval methods, devices and systems | |
| US20070244747A1 (en) | Method and system for recommending products to consumers by induction of decision trees | |
| CN111242748B (en) | Method, apparatus, and storage medium for recommending items to a user | |
| CN104254851A (en) | Method and system for recommending content to a user | |
| CN111008335B (en) | Information processing method, device, equipment and storage medium | |
| CN105144227A (en) | Social media impact assessment | |
| CN102105897A (en) | Near field communication transactions with user profile updates in a mobile environment | |
| Xue et al. | Computational experiment-based evaluation on context-aware O2O service recommendation | |
| KR20190007875A (en) | Method for providing marketing management data for optimization of distribution and logistic and apparatus therefor | |
| CN105160545A (en) | Delivered information pattern determination method and device | |
| CN107241914A (en) | The system and method rewritten for search inquiry | |
| US20230034820A1 (en) | Systems and methods for managing, distributing and deploying a recursive decisioning system based on continuously updating machine learning models | |
| KR102457359B1 (en) | Marketing cost efficiency calculation method and electronic system for performing the method using neural networks | |
| CN103632278A (en) | Multi-strategy commodity recommendation system based on context information | |
| US20230099627A1 (en) | Machine learning model for predicting an action | |
| CN115222433A (en) | Information recommendation method and device and storage medium | |
| CN111400567B (en) | AI-based user data processing method, device and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20080924 |