WO2017075980A1 - 信息推送方法及装置 - Google Patents

信息推送方法及装置 Download PDF

Info

Publication number
WO2017075980A1
WO2017075980A1 PCT/CN2016/083732 CN2016083732W WO2017075980A1 WO 2017075980 A1 WO2017075980 A1 WO 2017075980A1 CN 2016083732 W CN2016083732 W CN 2016083732W WO 2017075980 A1 WO2017075980 A1 WO 2017075980A1
Authority
WO
WIPO (PCT)
Prior art keywords
probability matrix
information
probability
preset
matrix
Prior art date
Application number
PCT/CN2016/083732
Other languages
English (en)
French (fr)
Inventor
陈克寒
Original Assignee
北京金山安全软件有限公司
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 北京金山安全软件有限公司 filed Critical 北京金山安全软件有限公司
Publication of WO2017075980A1 publication Critical patent/WO2017075980A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation

Definitions

  • the present application relates to the field of Internet technologies, and in particular, to an information pushing method and apparatus.
  • the demand of the merchant is often considered mainly, for example, the advertisement exposure rate required by the merchant, the advertisement push frequency requested by the merchant, and the like.
  • the server comprehensively analyzes the demand of the existing merchant, selects the advertisement to be pushed from the advertisements included in the current advertisement library according to the analysis result, and then pushes the advertisement to be pushed.
  • Applying the above method can successfully push the advertisement information to the user, but in fact, different users have different degrees of interest in different advertisements, for example, some users are interested in cosmetics advertisements, some users are interested in automobile advertisements, etc., so the application is now There is a technical way to conduct advertisement push when the targeting is not strong and the user experience is poor.
  • the embodiment of the present application discloses an information pushing method and device, which are based on user-specific information push to improve user experience.
  • the embodiment of the present application discloses an information pushing method, and the method includes:
  • the preset user information database obtaining, by the preset user information database, a probability that the target user is interested in each piece of information recorded in the preset information base, wherein the preset user information database is used to record each user to the preset The probability of interest in each piece of information recorded in the repository;
  • the preset user information database is generated in the following manner:
  • Determining whether the probability matrix D satisfies a preset convergence judgment condition If not, adjusting each element in the probability matrix D according to a preset adjustment rule, and updating the probability matrix D according to the adjusted element, Returning to the step of determining whether the probability matrix D satisfies a preset convergence determination condition until the probability matrix D satisfies the preset convergence determination condition;
  • the determining whether the probability matrix D meets a preset convergence determination condition includes:
  • the probability matrix D is decomposed into a probability matrix A′ and a probability matrix B′ according to a preset matrix decomposition algorithm, wherein the probability matrix A′ is a matrix corresponding to the probability matrix A, and the probability matrix B ' is a matrix corresponding to the probability matrix B;
  • the determining, according to the probability matrix A′, the probability matrix B′, and the probability matrix D, whether the probability matrix D satisfies a preset convergence judgment condition include:
  • a vector representing the elements of the i-th row of the probability matrix A' a vector representing the elements of the j-th row of the probability matrix B', ⁇ representing an adjustment coefficient, and d ij representing an element of the probability matrix D;
  • the determining whether the probability matrix D satisfies a preset convergence judgment condition according to the error argmin ⁇ , ⁇ L (D) includes:
  • the probability matrix D is a probability matrix updated according to the adjusted element
  • the absolute difference between the errors argmin ⁇ , ⁇ L (D) and the errors argmin ⁇ , ⁇ L (D)' is determined. Whether it is less than a preset second error threshold, if it is smaller, it is determined that the probability matrix D satisfies a preset convergence judgment condition, wherein the error argmin ⁇ , ⁇ L (D)′ represents the probability based on the update
  • the error obtained by matrix D is calculated.
  • the information pushing method further includes:
  • the probability matrix A and the probability matrix B are updated according to the probability matrix A' and the probability matrix B'.
  • the probability of each user being interested in each piece of information recorded in the preset information base is predicted according to the probability matrix A and the probability matrix B, and the probability is obtained.
  • Matrix D including:
  • d ij represents an element of the probability matrix D
  • ⁇ ik represents an element of the i-th row and the k-th column of the probability matrix A
  • ⁇ jk represents an element of the j-th column of the j-th column of the probability matrix B
  • k represents The number of information classifications obtained.
  • an embodiment of the present application discloses an information pushing device, where the device includes:
  • a push request receiving module configured to receive an information push request for the target user
  • a probability obtaining module configured to obtain, from a preset user information database, a probability that the target user is interested in each piece of information recorded in a preset information base, wherein the preset user information database is used to record each The probability that the user is interested in each piece of information recorded in the preset information base;
  • the to-be-push information determining module is configured to determine, to be pushed, the information to be pushed from the pieces of information recorded in the preset information database according to the obtained probability from high to low;
  • the information pushing module is configured to push the information to be pushed.
  • the information pushing apparatus further includes:
  • a database generating module configured to generate the preset user information database
  • the database generation module includes:
  • the information classification obtains a sub-module for obtaining the currently existing information classification
  • the first probability matrix obtains a sub-module for obtaining a probability that all current users are interested in each of the above information classifications, and generating a probability matrix A;
  • a second probability matrix obtaining sub-module configured to obtain a probability that each piece of information recorded in the preset information library belongs to each of the above information categories, and generate a probability matrix B;
  • a third probability matrix obtaining submodule configured to predict each according to the probability matrix A and the probability matrix B a probability that the user is interested in each piece of information recorded in the preset information base, and obtains a probability matrix D;
  • a convergence determination sub-module configured to determine whether the probability matrix D satisfies a preset convergence determination condition
  • a probability matrix update submodule configured to adjust each element in the probability matrix D according to a preset adjustment rule, and update the location according to the adjusted element, if the judgment result of the convergence determination submodule is negative
  • the probability matrix D is triggered to trigger the convergence determination sub-module to perform the determination until the probability matrix D satisfies the preset convergence determination condition;
  • the database generation submodule is configured to generate the preset user information database according to the probability matrix D.
  • the convergence determining submodule includes:
  • a matrix decomposition unit configured to decompose the probability matrix D into a probability matrix A′ and a probability matrix B′ according to a preset matrix decomposition algorithm, wherein the probability matrix A′ is a matrix corresponding to the probability matrix A
  • the probability matrix B' is a matrix corresponding to the probability matrix B;
  • the convergence determining unit is configured to determine, according to the probability matrix A', the probability matrix B', and the probability matrix D, whether the probability matrix D satisfies a preset convergence determination condition.
  • the convergence determining unit includes:
  • An error calculation subunit for calculating an error argmin ⁇ , ⁇ L (D) between the predicted probability and the true probability according to the following expression
  • a vector representing the elements of the i-th row of the probability matrix A' a vector representing the elements of the j-th row of the probability matrix B', ⁇ representing an adjustment coefficient, and d ij representing an element of the probability matrix D;
  • the convergence determination subunit is configured to determine, according to the error argmin ⁇ , ⁇ L (D), whether the probability matrix D satisfies a preset convergence determination condition.
  • the convergence determining subunit In a specific implementation manner of the application, the convergence determining subunit,
  • the probability matrix D is a probability matrix updated according to the adjusted element, determining the error between argmin ⁇ , ⁇ L (D) and the error argmin ⁇ , ⁇ L (D)′ Whether the absolute difference is less than a preset second error threshold, and if not, determining that the probability matrix D satisfies a preset convergence determination condition, wherein the error argmin ⁇ , ⁇ L (D)′ represents based on the update
  • the probability matrix D calculates the error.
  • the database generating module further includes:
  • a matrix update submodule configured to update the probability matrix A and the probability matrix according to the probability matrix A′ and the probability matrix B′, if it is determined that the probability matrix D satisfies a preset convergence determination condition Probability matrix B.
  • the third probability matrix obtains a sub-module, and is specifically configured to predict, according to the following expression, a probability that each user is interested in each piece of information recorded in the preset information base. , obtain the probability matrix D,
  • d ij represents an element of the probability matrix D
  • ⁇ ik represents an element of the i-th row and the k-th column of the probability matrix A
  • ⁇ jk represents an element of the j-th column of the j-th column of the probability matrix B
  • k represents The number of information classifications obtained.
  • an embodiment of the present application discloses an information pushing device, including: a housing, a processor, a memory, a circuit board, and a power supply circuit, wherein the circuit board is disposed inside the space enclosed by the housing, the processor and the The memory is disposed on the circuit board; the power circuit is configured to supply power to each circuit or device of the information pushing device; the memory is used to store the executable program code; and the processor is operable to read the executable program code stored in the memory.
  • the program corresponding to the program code is executed for executing the information pushing method described in the above embodiment of the present invention.
  • the embodiment of the present application discloses a computer program, which when executed on a processor, executes the information pushing method according to the above embodiment of the present invention.
  • the embodiment of the present application discloses a computer readable storage medium, where the computer storage medium stores one or more modules, and when the one or more modules are executed by the terminal, the terminal: executes the present The information pushing method described in the above embodiment is invented.
  • the probability that each current user is interested in classifying the existing information and the probability that each piece of information recorded in the preset information library belongs to each information classification predicts each user pair.
  • the probability that each piece of information recorded in the preset information base is of interest, and the information to be pushed is determined according to the predicted result, and then the information is pushed. It can be seen that when the information to be pushed is determined by using the solution provided by the embodiment of the present application, the probability that the user is interested in each piece of information is considered, and therefore, the user can be pushed according to the targeted information of the user, thereby improving the user experience.
  • FIG. 1 is a schematic flowchart of a method for pushing information according to an embodiment of the present application
  • FIG. 2 is a schematic flowchart of a database generating method according to an embodiment of the present application.
  • FIG. 3 is a schematic structural diagram of an information pushing apparatus according to an embodiment of the present application.
  • FIG. 4 is a schematic structural diagram of a database generating apparatus according to an embodiment of the present application.
  • FIG. 5 is a schematic structural diagram of an information pushing apparatus according to an embodiment of the present application.
  • FIG. 1 is a schematic flowchart of a method for pushing information according to an embodiment of the present disclosure, where the method includes:
  • S101 Receive an information push request for the target user.
  • the information push request needs to carry at least the identifier of the target user.
  • the request may carry other information, which is not limited in this application.
  • the information involved in the embodiment of the present application may be advertisement information, news information, weather forecast information, and the like, which is not limited in this application.
  • S102 Obtain a probability that the target user is interested in each piece of information recorded in the preset information base from the preset user information database.
  • the preset user information database is used to record the probability that each user is interested in each piece of information recorded in the preset information base.
  • the preset user information database when the preset user information database is generated in advance, the probability that all the users are interested in the existing information classification, the probability that each piece of information in the preset information library belongs to each information classification, and the like may be considered, and the specific generation process may be Referring to the embodiment shown in Figure 2, it will not be described in detail herein.
  • the user information database may be, for example, a preset user information database according to a fixed time interval, such as one-time update, one-week update, and the like.
  • S103 Determine, to be pushed, the information to be pushed from the pieces of information recorded in the preset information database according to the obtained high-to-low probability.
  • the information to be pushed when the information to be pushed is determined from the pieces of information recorded in the preset information database, the information to be pushed may be determined according to the obtained probability from high to low, so that the user may be preferentially pushed to the user.
  • Information when the information to be pushed is determined from the pieces of information recorded in the preset information database, the information to be pushed may be determined according to the obtained probability from high to low, so that the user may be preferentially pushed to the user.
  • the number of times the information in the preset information base has been pushed to the user within the preset time period may be considered, for example, the target user has been pushed and obtained today.
  • the information with the highest probability among the obtained probabilities may not determine the information with the highest probability as the information to be pushed, so as to prevent the user from being resent by repeatedly pushing the same information to the target user in a short period of time.
  • the preset user information database referred to above is described in detail below through a specific embodiment.
  • FIG. 2 is a schematic flowchart of a database generating method according to an embodiment of the present disclosure, where the method includes:
  • the current information classification may be manually divided by the operation and maintenance personnel in the actual application process.
  • S202 Obtain a probability that all current users are interested in classifying each of the foregoing information, and generate a probability matrix A.
  • the probability obtained here is only the initial probability of generating the user information database, in a specific implementation manner of the present application, the probability that all current users are interested in classifying each of the above information may be randomly generated.
  • the number of operations performed by the user for each type of information may be measured in real time for a period of time, for example, The number of times the strategy game is run, the number of times the casual game is installed, the number of times the strategy game advertisement is clicked, etc., and then the machine learning model corresponding to each information classification is obtained by the existing machine learning algorithm, thereby obtaining the current classification of each information by all users. The probability of interest.
  • the probability that a certain user is interested in a certain information classification cannot be obtained, and the user may be interested in classifying the information.
  • the probability is zero.
  • S203 Obtain a probability that each piece of information recorded in the preset information library belongs to each of the above information categories, and generate a probability matrix B.
  • the information classification of the information recorded in the preset information base may be manually set, or may be set according to the information fed back during the browsing of the information by the user, and each piece of information recorded in the preset information base. It can belong to only one information category, or it can belong to multiple information categories.
  • each piece of information recorded in the preset information library belongs to each of the above information classifications, it may be found that a certain record does not belong to any one of the information classifications. In this case, the record may be set. The probability of belonging to each information classification is zero.
  • the probability that each user is interested in each piece of information recorded in the preset information base is predicted, and when the probability matrix D is obtained, the following expression may be expressed.
  • d ij represents the element of the probability matrix D.
  • d ij is the element of the i-th row and the j-th column of the probability matrix D
  • ⁇ ik represents the element of the i-th row and the k-th column of the probability matrix A
  • ⁇ jk represents the probability matrix
  • the element of column k, column k, and k represents the number of information classifications obtained.
  • the probability matrix D is predicted based on the probability matrix A and the probability matrix B, it can be considered whether the probability matrix D satisfies the preset convergence judgment condition by the matrix decomposition method. Specifically, when determining whether the probability matrix D satisfies a preset convergence judgment condition, the probability matrix D may be decomposed into a probability matrix A′ and a probability matrix B′ according to a preset matrix decomposition algorithm, and then according to the probability matrix A′, The probability matrix B' and the probability matrix D determine whether the probability matrix D satisfies a preset convergence judgment condition.
  • the probability matrix A' is a matrix corresponding to the probability matrix A
  • the probability matrix B' is a matrix corresponding to the probability matrix B.
  • the error argmin between the prediction probability and the true probability may be first calculated according to the following expression. ⁇ , ⁇ L (D) , and then judge whether the probability matrix D satisfies the preset convergence judgment condition based on the errors argmin ⁇ , ⁇ L (D) .
  • d ij represents an element of the probability matrix D. Specifically, d ij represents an element of the i-th row and the j-th column of the probability matrix D.
  • determining whether the probability matrix D satisfies a preset convergence judgment condition according to the error argmin ⁇ , ⁇ L (D) may be determined according to the following conditions:
  • the first case determining whether the error argmin ⁇ , ⁇ L (D) is less than a preset first error threshold, and if not, determining that the probability matrix D satisfies a preset convergence judgment condition;
  • the second case in the case where the probability matrix D is a probability matrix updated according to the adjusted element, the absolute difference between the error argmin ⁇ , ⁇ L (D) and the error argmin ⁇ , ⁇ L (D)' is judged. Whether the value is less than a preset second error threshold, and if less, the decision probability matrix D satisfies a preset convergence judgment condition, wherein the error argmin ⁇ , ⁇ L (D)′ represents a calculation based on the probability matrix D before the update error.
  • S206 Adjust each element in the probability matrix D according to a preset adjustment rule, and update the probability matrix D according to the adjusted element, and return to S205.
  • the probability matrix D involved in this step may be the updated probability matrix D, or may be the probability matrix D after one or more adjustments.
  • the probability matrix A and the probability matrix B may also be updated according to the probability matrix A′ and the probability matrix B′. This is equivalent to a reverse update of the probability that each user is currently interested in each information classification, and at the same time updates the probability that the information in the preset information base belongs to each information classification, especially for information that belongs to any one of the information classifications.
  • the classification of the information can be implemented by updating the probability matrix B.
  • updating the probability matrix A and the probability matrix B can facilitate subsequent updating of the preset user information database.
  • the probability that each current user is interested in classifying the existing information and the probability that each piece of information recorded in the preset information library belongs to each information classification predicts each user pair.
  • the probability that each piece of information recorded in the preset information base is of interest, and the information to be pushed is determined according to the predicted result, and then the information is pushed. It can be seen that when the information to be pushed is determined by using the solution provided by the foregoing embodiments, the probability that the user is interested in each piece of information is considered, and therefore, the user can be pushed according to the targeted information of the user, thereby improving the user experience.
  • the embodiment of the present application further provides an information pushing device.
  • FIG. 3 is a schematic structural diagram of an information pushing apparatus according to an embodiment of the present disclosure, where the apparatus includes:
  • the push request receiving module 301 is configured to receive an information push request for the target user
  • the probability obtaining module 302 is configured to obtain, from a preset user information database, a probability that the target user is interested in each piece of information recorded in the preset information base, wherein the preset user information database is used for recording Each The probability that each user is interested in each piece of information recorded in the preset information base;
  • the to-be-push information determining module 303 is configured to determine, to be pushed, the information to be pushed from the pieces of information recorded in the preset information database according to the obtained probability from high to low;
  • the information pushing module 304 is configured to push the information to be pushed.
  • the information pushing apparatus may further include:
  • a database generating module configured to generate the preset user information database.
  • FIG. 4 is a schematic structural diagram of a database generating apparatus according to an embodiment of the present disclosure.
  • the apparatus is a specific device of a database generating module, and includes:
  • the information classification obtaining sub-module 401 is configured to obtain the currently existing information classification
  • the first probability matrix obtaining sub-module 402 is configured to obtain a probability that all current users are interested in each of the above information classifications, and generate a probability matrix A;
  • the second probability matrix obtaining sub-module 403 is configured to obtain a probability that each piece of information recorded in the preset information library belongs to each of the above information categories, and generate a probability matrix B;
  • the third probability matrix obtaining sub-module 404 is configured to predict a probability that each user is interested in each piece of information recorded in the preset information base according to the probability matrix A and the probability matrix B, and obtain a probability matrix D. ;
  • the convergence determination sub-module 405 is configured to determine whether the probability matrix D satisfies a preset convergence determination condition
  • the probability matrix update sub-module 406 is configured to adjust each element in the probability matrix D according to a preset adjustment rule if the determination result of the convergence determination sub-module 405 is negative, and according to the adjusted element Updating the probability matrix D, triggering the convergence determination sub-module 405 to perform a determination until the probability matrix D satisfies the preset convergence determination condition;
  • the database generation sub-module 407 is configured to generate the preset user information database according to the probability matrix D.
  • the convergence determination sub-module 405 can include:
  • a matrix decomposition unit configured to decompose the probability matrix D into a probability matrix A′ and a probability matrix B′ according to a preset matrix decomposition algorithm, wherein the probability matrix A′ is a matrix corresponding to the probability matrix A
  • the probability matrix B' is a matrix corresponding to the probability matrix B;
  • the convergence determining unit is configured to determine, according to the probability matrix A', the probability matrix B', and the probability matrix D, whether the probability matrix D satisfies a preset convergence determination condition.
  • the convergence determining unit may include:
  • An error calculation sub-unit for calculating an error between the predicted probability and the true probability according to the following expression argmin ⁇ , ⁇ L (D) ,
  • a vector representing the elements of the i-th row of the probability matrix A' a vector representing the elements of the j-th row of the probability matrix B', ⁇ representing an adjustment coefficient, and d ij representing an element of the probability matrix D;
  • the convergence determination subunit is configured to determine, according to the error argmin ⁇ , ⁇ L (D), whether the probability matrix D satisfies a preset convergence determination condition.
  • the method may be specifically configured to determine whether the error argmin ⁇ , ⁇ L (D) is less than a preset first error threshold, and if not, determine that the probability matrix D satisfies a preset convergence judgment condition; or
  • the error argmin ⁇ , ⁇ L (D) and the error argmin ⁇ , ⁇ L (D)' in the case where the probability matrix D is a probability matrix updated according to the adjusted element. Whether the absolute difference is less than a preset second error threshold, and if not, determining that the probability matrix D satisfies a preset convergence determination condition, wherein the error argmin ⁇ , ⁇ L (D)' indicates that the update is based on The probability matrix D calculates the error.
  • the database generating module may further include:
  • a matrix update submodule configured to update the probability matrix A and the probability matrix according to the probability matrix A′ and the probability matrix B′, if it is determined that the probability matrix D satisfies a preset convergence determination condition Probability matrix B.
  • the third probability matrix obtains a sub-module, which may be specifically used to predict, according to the following expression, a probability that each user is interested in each piece of information recorded in the preset information base, and obtain a probability matrix D,
  • d ij represents an element of the probability matrix D
  • ⁇ ik represents an element of the i-th row and the k-th column of the probability matrix A
  • ⁇ jk represents an element of the j-th column of the j-th column of the probability matrix B
  • k represents The number of information classifications obtained.
  • the probability that each current user is interested in classifying the existing information and the probability that each piece of information recorded in the preset information library belongs to each information classification predicts each user pair.
  • the probability that each piece of information recorded in the preset information base is of interest, and the information to be pushed is determined according to the predicted result, and then the information is pushed. It can be seen that when the information to be pushed is determined by using the solution provided by the foregoing embodiments, the probability that the user is interested in each piece of information is considered, and therefore, the user can be pushed according to the targeted information of the user, thereby improving the user experience.
  • the description is relatively simple, and the relevant parts can be referred to the description of the method embodiment.
  • an embodiment of the present invention further provides an information pushing apparatus, including: a housing 501, a processor 502, a memory 503, a circuit board 504, and a power supply circuit 505, wherein the circuit board 504 is disposed in the housing 501.
  • processor 502 and memory 503 are disposed on circuit board 504; power supply circuit 505 is used to power various circuits or devices of the electronic device; memory 503 is used to store executable program code; and processor 502 is read by The executable program code stored in the memory 503 is taken to execute a program corresponding to the executable program code for performing the following steps:
  • S101' receiving an information push request for the target user.
  • S102' obtaining, from a preset user information database, a probability that the target user is interested in each piece of information recorded in the preset information base.
  • S103' determining the information to be pushed from among pieces of information recorded in the preset information database according to the obtained high-to-low probability.
  • the processor 502 also runs a program corresponding to the executable program code by reading the executable program code stored in the memory 503 for performing the following steps:
  • S204' predicting a probability that each user is interested in each piece of information recorded in the preset information base according to the probability matrix A and the probability matrix B, and obtaining a probability matrix D.
  • S205' It is judged whether or not the probability matrix D satisfies the preset convergence judgment condition. If not, the execution S206' is satisfied, and S207' is executed.
  • Each element in the probability matrix D is adjusted according to a preset adjustment rule, and the probability matrix D is updated according to the adjusted element, and the process returns to S205'.
  • S207' Generate a preset user information database according to the probability matrix D.
  • embodiments of the present invention also provide a computing program that, when run on a processor, performs the following steps:
  • the preset user information database obtaining, by the preset user information database, a probability that the target user is interested in each piece of information recorded in the preset information base, wherein the preset user information database is used to record each user to the preset Letter The probability of interest in each piece of information recorded in the repository;
  • an embodiment of the present invention further provides a computer readable storage medium.
  • the computer storage medium stores one or more modules.
  • the terminal executes the following steps:
  • the preset user information database obtaining, by the preset user information database, a probability that the target user is interested in each piece of information recorded in the preset information base, wherein the preset user information database is used to record each user to the preset The probability of interest in each piece of information recorded in the repository;
  • An ordered list of executable instructions for implementing logical functions may be embodied in any computer readable medium for use in an instruction execution system, apparatus, or device (eg, a computer-based system, a system including a processor, or other A system, device, or device that fetches instructions and executes instructions for use, or in conjunction with such instructions to execute a system, apparatus, or device.
  • a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport a program for use in an instruction execution system, apparatus, or device, or in conjunction with the instruction execution system, apparatus, or device.
  • computer readable media include the following: electrical connections (electronic devices) having one or more wires, portable computer disk cartridges (magnetic devices), random access memory (RAM), Read only memory (ROM), erasable editable read only memory (EPROM or flash memory), fiber optic devices, and portable compact disk read only memory (CDROM).
  • the computer readable medium may even be a paper or other suitable medium on which the program can be printed, as it may be optically scanned, for example by paper or other medium, followed by editing, interpretation or, if appropriate, other suitable The method is processed to obtain the program electronically and then stored in computer memory.
  • portions of the invention may be implemented in hardware, software, firmware or a combination thereof.
  • multiple steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system.
  • a suitable instruction execution system For example, if implemented in hardware, as in another embodiment, it can be implemented by any one or combination of the following techniques well known in the art: having logic gates for implementing logic functions on data signals. Discrete logic circuits, application specific integrated circuits with suitable combinational logic gates, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.
  • each functional unit in each embodiment of the present invention may be integrated into one processing module, or each unit may exist physically separately, or two or more units may be integrated into one module.
  • the above integrated modules can be implemented in the form of hardware or in the form of software functional modules.
  • the integrated modules, if implemented in the form of software functional modules and sold or used as stand-alone products, may also be stored in a computer readable storage medium.

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本申请实施例公开了一种信息推送方法及装置，涉及互联网技术领域，其中，包括：接收针对目标用户的信息推送请求；从预设的用户信息数据库中获得所述目标用户对预设的信息库中记录的各条信息感兴趣的概率，其中，所述预设的用户信息数据库用于记录各个用户对所述预设的信息库中记录的各条信息感兴趣的概率；根据所获得的概率由高到低的顺序，从所述预设的信息数据库中记录的各条信息中确定待推送信息；推送所述待推送信息。应用本申请实施例提供的方案能够有针对性的进行信息推送。

Description

信息推送方法及装置
相关申请的交叉引用
本申请要求北京金山安全软件有限公司于2015年11月6日提交的、发明名称为“一种信息推送方法及装置”的、中国专利申请号“201510752379.6”的优先权。
技术领域
本申请涉及互联网技术领域，特别涉及一种信息推送方法及装置。
背景技术
近年来随着网络技术的快速发展，网络用户逐年递增，商家为了能够以较大的力度推广其产品，越来越多的倾向于通过网络投放广告，考虑到商家的这一需求，运营商通常会综合多个商家的需求，向用户推送广告信息。
实际应用中，运营商向用户推送广告时，往往主要考虑商家的需求，例如，商家要求的广告曝光率、商家要求的广告推送频率等等。具体的，现有技术中，进行广告推送时，服务器综合分析已有商家的需求，根据分析结果从当前广告库包含的广告中选择待推送的广告，然后推送上述待推送的广告。
应用上述方式可以成功向用户推送广告信息，然而实际上不同用户对不同广告的感兴趣程度不同，例如，一些用户对化妆品类广告感兴趣，一些用户对汽车类广告感兴趣等等，所以应用现有技术中的方式进行广告推送时针对性不强，用户体验差。
发明内容
本申请实施例公开了一种信息推送方法及装置，以基于用户有针对性的进行信息推送，提高用户体验。
为达到上述目的，本申请实施例公开了一种信息推送方法，所述方法包括：
接收针对目标用户的信息推送请求；
从预设的用户信息数据库中获得所述目标用户对预设的信息库中记录的各条信息感兴趣的概率，其中，所述预设的用户信息数据库用于记录各个用户对所述预设的信息库中记录的各条信息感兴趣的概率；
根据所获得的概率由高到低的顺序，从所述预设的信息数据库中记录的各条信息中确定待推送信息；
推送所述待推送信息。
在本申请的一种具体实现方式中，按照以下方式生成所述预设的用户信息数据库：
获得当前已有的信息分类；
获得当前所有用户对上述各个信息分类感兴趣的概率，生成概率矩阵A；
获得所述预设信息库中记录的各条信息属于上述各个信息分类的概率，生成概率矩阵B；
根据所述概率矩阵A和所述概率矩阵B，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D；
判断所述概率矩阵D是否满足预设的收敛性判断条件，若不满足，按照预设的调整规则调整所述概率矩阵D中的各个元素，并根据调整后的元素更新所述概率矩阵D，返回所述判断所述概率矩阵D是否满足预设的收敛性判断条件的步骤，直至所述概率矩阵D满足所述预设的收敛性判断条件；
根据所述概率矩阵D生成所述预设的用户信息数据库。
在本申请的一种具体实现方式中，所述判断所述概率矩阵D是否满足预设的收敛性判断条件，包括：
根据预设的矩阵分解算法，将所述概率矩阵D分解为概率矩阵A’和概率矩阵B’，其中，所述概率矩阵A’为与所述概率矩阵A对应的矩阵，所述概率矩阵B’为与所述概率矩阵B对应的矩阵；
根据所述概率矩阵A’、所述概率矩阵B’以及所述概率矩阵D，判断所述概率矩阵D是否满足预设的收敛性判断条件。
在本申请的一种具体实现方式中，所述根据所述概率矩阵A’、所述概率矩阵B’以及所述概率矩阵D，判断所述概率矩阵D是否满足预设的收敛性判断条件，包括：
根据以下表达式计算预测概率与真实概率之间的误差argminα,βL(D)，
Figure PCTCN2016083732-appb-000001
其中，
Figure PCTCN2016083732-appb-000002
表示所述概率矩阵A’第i行的元素组成的向量，
Figure PCTCN2016083732-appb-000003
表示所述概率矩阵B’第j行的元素组成的向量，λ表示调整系数，dij表示所述概率矩阵D的元素；
根据所述误差argminα,βL(D)判断所述概率矩阵D是否满足预设的收敛性判断条件。
在本申请的一种具体实现方式中，所述根据所述误差argminα,βL(D)判断所述概率矩阵D是否满足预设的收敛性判断条件，包括：
判断所述误差argminα,βL(D)是否小于预设的第一误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件；或
在所述概率矩阵D为根据调整后的元素更新后的概率矩阵的情况下，判断所述误 差argminα,βL(D)与误差argminα,βL(D)’之间的绝对差值是否小于预设的第二误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件，其中，所述误差argminα,βL(D)’表示基于更新之前的所述概率矩阵D计算得到的误差。
在本申请的一种具体实现方式中，所述信息推送方法还包括：
在判断得所述概率矩阵D满足预设的收敛性判断条件的情况下，根据所述概率矩阵A‘和所述概率矩阵B’更新所述概率矩阵A和所述概率矩阵B。
在本申请的一种具体实现方式中，所述根据所述概率矩阵A和所述概率矩阵B，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D，包括：
根据以下表达式，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D，
Figure PCTCN2016083732-appb-000004
其中，dij表示所述概率矩阵D的元素，αik表示所述概率矩阵A第i行第k列的元素，βjk表示所述概率矩阵B第j行第k列的元素，k表示所获得的信息分类的数量。
为达到上述目的，本申请实施例公开了一种信息推送装置，所述装置包括：
推送请求接收模块，用于接收针对目标用户的信息推送请求；
概率获得模块，用于从预设的用户信息数据库中获得所述目标用户对预设的信息库中记录的各条信息感兴趣的概率，其中，所述预设的用户信息数据库用于记录各个用户对所述预设的信息库中记录的各条信息感兴趣的概率；
待推送信息确定模块，用于根据所获得的概率由高到低的顺序，从所述预设的信息数据库中记录的各条信息中确定待推送信息；
信息推送模块，用于推送所述待推送信息。
在本申请的一种具体实现方式中，所述信息推送装置还包括：
数据库生成模块，用于生成所述预设的用户信息数据库；
其中，所述数据库生成模块，包括：
信息分类获得子模块，用于获得当前已有的信息分类；
第一概率矩阵获得子模块，用于获得当前所有用户对上述各个信息分类感兴趣的概率，生成概率矩阵A；
第二概率矩阵获得子模块，用于获得所述预设信息库中记录的各条信息属于上述各个信息分类的概率，生成概率矩阵B；
第三概率矩阵获得子模块，用于根据所述概率矩阵A和所述概率矩阵B，预测各 个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D；
收敛性判断子模块，用于判断所述概率矩阵D是否满足预设的收敛性判断条件；
概率矩阵更新子模块，用于在所述收敛性判断子模块的判断结果为否的情况下，按照预设的调整规则调整所述概率矩阵D中的各个元素，并根据调整后的元素更新所述概率矩阵D，触发所述收敛性判断子模块进行判断，直至所述概率矩阵D满足所述预设的收敛性判断条件；
数据库生成子模块，用于根据所述概率矩阵D生成所述预设的用户信息数据库。
在本申请的一种具体实现方式中，所述收敛性判断子模块，包括：
矩阵分解单元，用于根据预设的矩阵分解算法，将所述概率矩阵D分解为概率矩阵A’和概率矩阵B’，其中，所述概率矩阵A’为与所述概率矩阵A对应的矩阵，所述概率矩阵B’为与所述概率矩阵B对应的矩阵；
收敛性判断单元，用于根据所述概率矩阵A’、所述概率矩阵B’以及所述概率矩阵D，判断所述概率矩阵D是否满足预设的收敛性判断条件。
在本申请的一种具体实现方式中，所述收敛性判断单元，包括：
误差计算子单元，用于根据以下表达式计算预测概率与真实概率之间的误差argminα,βL(D)，
Figure PCTCN2016083732-appb-000005
其中，
Figure PCTCN2016083732-appb-000006
表示所述概率矩阵A’第i行的元素组成的向量，
Figure PCTCN2016083732-appb-000007
表示所述概率矩阵B’第j行的元素组成的向量，λ表示调整系数，dij表示所述概率矩阵D的元素；
收敛性判断子单元，用于根据所述误差argminα,βL(D)判断所述概率矩阵D是否满足预设的收敛性判断条件。
在本申请的一种具体实现方式中，所述收敛性判断子单元，
具体用于判断所述误差argminα,βL(D)是否小于预设的第一误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件；或
具体用于在所述概率矩阵D为根据调整后的元素更新后的概率矩阵的情况下，判断所述误差argminα,βL(D)与误差argminα,βL(D)’之间的绝对差值是否小于预设的第二误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件，其中，所述误差argminα,βL(D)’表示基于更新之前的所述概率矩阵D计算得到的误差。
在本申请的一种具体实现方式中，所述数据库生成模块，还包括：
矩阵更新子模块，用于在判断得所述概率矩阵D满足预设的收敛性判断条件的情况下，根据所述概率矩阵A‘和所述概率矩阵B’更新所述概率矩阵A和所述概率矩阵 B。
在本申请的一种具体实现方式中，所述第三概率矩阵获得子模块，具体用于根据以下表达式，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D，
Figure PCTCN2016083732-appb-000008
其中，dij表示所述概率矩阵D的元素，αik表示所述概率矩阵A第i行第k列的元素，βjk表示所述概率矩阵B第j行第k列的元素，k表示所获得的信息分类的数量。
为达到上述目的，本申请实施例公开了一种信息推送装置，包括：壳体、处理器、存储器、电路板和电源电路，其中，电路板安置在壳体围成的空间内部，处理器和存储器设置在电路板上；电源电路，用于为信息的推送装置的各个电路或器件供电；存储器用于存储可执行程序代码；处理器通过读取存储器中存储的可执行程序代码来运行与可执行程序代码对应的程序，以用于执行本发明上述实施例所述的信息推送方法。
为达到上述目的，本申请实施例公开了一种计算机程序，当其在处理器上运行时，执行本发明上述实施例所述的信息推送方法。
为达到上述目的，本申请实施例公开了一种计算机可读存储介质，所述计算机存储介质存储有一个或者多个模块，当所述一个或者多个模块被终端执行时，使得终端：执行本发明上述实施例所述的信息推送方法。
由以上可见，本申请实施例提供的方案中，根据当前所有用户对已有的信息分类感兴趣的概率以及预设的信息库中记录的每一条信息属于各个信息分类的概率，预测各个用户对预设的信息库中记录的每一条信息感兴趣的概率，并根据预测结果确定待推送信息，进而进行信息推送。可见，应用本申请实施例提供的方案确定待推送信息时，考虑了用户对各条信息感兴趣的概率，因此，能够基于用户有针对性的进行信息推送，提高了用户体验。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的一种信息推送方法的流程示意图；
图2为本申请实施例提供的一种数据库生成方法的流程示意图；
图3为本申请实施例提供的一种信息推送装置的结构示意图；
图4为本申请实施例提供的一种数据库生成装置的结构示意图；
图5为本申请实施例提供的一种信息推送装置的结构示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
图1为本申请实施例提供的一种信息推送方法的流程示意图，该方法包括：
S101：接收针对目标用户的信息推送请求。
上述信息推送请求中至少需携带目标用户的标识，当然，该请求中还可以携带其他信息，本申请并不对此进行限定。
需要说明的是，本申请实施例中所涉及的信息可以是广告信息、新闻信息、天气预报信息等等，本申请并不对此进行限定。
S102：从预设的用户信息数据库中获得目标用户对预设的信息库中记录的各条信息感兴趣的概率。
其中，预设的用户信息数据库用于记录各个用户对预设的信息库中记录的各条信息感兴趣的概率。
具体的，预先生成上述预设的用户信息数据库时可以考虑当前所有用户对已有信息分类感兴趣的概率、预设的信息库中各条信息属于各个信息分类的概率等信息，具体生成过程可以参见图2所示实施例，这里暂不详述。
可以理解的，由于客户需求等不断发生变化，所以预设的信息库中记录的信息也是不断变化的，另外，用户的兴趣爱好也并不是一成不变的，基于上述几种原因，需不断更新上述预设的用户信息数据库，例如，可以是按照固定的时间间隔上述预设的用户信息数据库，如，一天更新一次、一周更新一次等等。
S103：根据所获得的概率由高到低的顺序，从预设的信息数据库中记录的各条信息中确定待推送信息。
根据所获得的概率，从预设的信息数据库记录的各条信息中确定待推送信息时，可以按照所获得的概率由高到低的顺序确定待推送信息，这样能够优先向用户推送其感兴趣的信息。
另外，在确定待推送信息时除了需要考虑所获得的概率之外，还可以考虑预设时段内已向用户推送预设的信息库中信息的次数，例如，今天已向目标用户推送过所获 得的概率中概率最高的信息，则可以不将该概率最高的信息确定为待推送信息，以防止短期内向目标用户重复推送同一信息导致用户反感。
当然，确定待推送信息时还可以考虑客户要求的信息曝光率、信息推送频率等因素，本申请并不对此进行限定。
S104：推送所述待推送信息。
下面通过具体实施例详细介绍前面涉及到的预设的用户信息数据库。
图2为本申请实施例提供的一种数据库生成方法的流程示意图，该方法包括：
S201：获得当前已有的信息分类。
具体的，当前已有的信息分类可以是实际应用过程中运营维护人员人为划分的。
S202：获得当前所有用户对上述各个信息分类感兴趣的概率，生成概率矩阵A。
由于这里所获得的概率只是作为后续生成用户信息数据库的初始概率，所以，在本申请的一种具体实现方式中，当前所有用户对上述各个信息分类感兴趣的概率可以是随机生成的。
相对于上述获得当前所有用户对上述各个信息分类感兴趣的概率的方式，在本申请的另一种具体实现方式中，还可以在一段时间内实时统计用户针对各类信息的操作次数，例如，运行策略游戏的次数、安装休闲游戏的次数，点击策略游戏广告的次数等等，然后通过现有的机器学习算法获得每一信息分类对应的机器学习模型，从而获得当前所有用户对上述各个信息分类感兴趣的概率。
需要说明的是，若在获得当前所有用户对上述各个信息分类感兴趣的概率的过程中，无法获得某一用户对某一信息分类感兴趣的概率，可以设置该用户对该信息分类感兴趣的概率为零。
需要说明的是，本申请只是以上述为例进行说明，实际应用中获得当前所有用户对上述各个信息分类感兴趣的概率的方式并不仅限于此。
S203：获得预设信息库中记录的各条信息属于上述各个信息分类的概率，生成概率矩阵B。
预设信息库中记录的各条信息属于哪一个信息分类可以是人为设定的，也可以是根据用户浏览信息过程中反馈的信息设定的，另外，预设信息库中记录的每一条信息可以只属于一个信息分类，也可以属于多个信息分类。
需要说明的是，在获得预设信息库中记录的各条信息属于上述各个信息分类的概率的过程中，有可能会发现某一条记录不属于任何一个信息分类，这种情况下可以设置该记录属于各个信息分类的概率为零。
S204：根据概率矩阵A和概率矩阵B，预测各个用户对预设的信息库中记录的每 一条信息感兴趣的概率，获得概率矩阵D。
在本申请的一种具体实现方式中，根据概率矩阵A和概率矩阵B，预测各个用户对预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D时，可以根据以下表达式，预测各个用户对预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D，
Figure PCTCN2016083732-appb-000009
其中，dij表示概率矩阵D的元素，具体的，dij为概率矩阵D中第i行第j列的元素，αik表示概率矩阵A第i行第k列的元素，βjk表示概率矩阵B第j行第k列的元素，k表示所获得的信息分类的数量。
S205：判断概率矩阵D是否满足预设的收敛性判断条件，若不满足，执行S206若满足，执行S207。
由于概率矩阵D是根据概率矩阵A和概率矩阵B预测得的，所以，可以考虑通过矩阵分解方式判断概率矩阵D是否满足预设的收敛性判断条件。具体的，判断概率矩阵D是否满足预设的收敛性判断条件时，可以根据预设的矩阵分解算法，将概率矩阵D分解为概率矩阵A’和概率矩阵B’，然后根据概率矩阵A’、概率矩阵B’以及概率矩阵D，判断概率矩阵D是否满足预设的收敛性判断条件。
其中，概率矩阵A’为与概率矩阵A对应的矩阵，概率矩阵B’为与概率矩阵B对应的矩阵。
具体的，根据概率矩阵A’、概率矩阵B’以及概率矩阵D，判断概率矩阵D是否满足预设的收敛性判断条件时，可以先根据以下表达式计算预测概率与真实概率之间的误差argminα,βL(D)，然后根据误差argminα,βL(D)判断概率矩阵D是否满足预设的收敛性判断条件。
其中，前面涉及的表达式为：
Figure PCTCN2016083732-appb-000010
Figure PCTCN2016083732-appb-000011
表示概率矩阵A’第i行的元素组成的向量，
Figure PCTCN2016083732-appb-000012
表示概率矩阵B’第j行的元素组成的向量，λ表示调整系数，dij表示概率矩阵D的元素，具体的，dij表示概率矩阵D第i行第j列的元素。
在本申请的一种具体实现方式中，根据误差argminα,βL(D)判断概率矩阵D是否满足预设的收敛性判断条件时可以根据以下几种情况进行判断：
第一种情况：判断误差argminα,βL(D)是否小于预设的第一误差阈值，若小于，判定 概率矩阵D满足预设的收敛性判断条件；
第二种情况：在概率矩阵D为根据调整后的元素更新后的概率矩阵的情况下，判断误差argminα,βL(D)与误差argminα,βL(D)’之间的绝对差值是否小于预设的第二误差阈值，若小于，判定概率矩阵D满足预设的收敛性判断条件，其中，误差argminα,βL(D)’表示基于更新之前的概率矩阵D计算得到的误差。
S206：按照预设的调整规则调整概率矩阵D中的各个元素，并根据调整后的元素更新概率矩阵D，返回S205。
S207：根据概率矩阵D生成预设的用户信息数据库。
需要说明的是，本步骤中涉及的概率矩阵D可以是为经过更新的概率矩阵D，也可以是经过一次或者多次调整后的概率矩阵D。
在本申请的一种可选实现方式中，在判断得概率矩阵D满足预设的收敛性判断条件的情况下，还可以根据概率矩阵A‘和概率矩阵B’更新概率矩阵A和概率矩阵B，这样相当于反向更新了当前各个用户对各个信息分类感兴趣的概率，同时更新了预设的信息库中的信息属于各个信息分类的概率，尤其是对于之前属于任何一个信息分类的信息而言，通过概率矩阵B的更新，能够实现对该信息的分类，另外，更新上述概率矩阵A和上述概率矩阵B能够有利于后续更新预设的用户信息数据库。
进一步的，从上述描述可以看出生成预设的用户信息数据库的过程中虽然需要一定量的各条信息属于各个信息分类的概率信息作为初始值，但是对这些作为初始值的概率的数据量以及准确性没有要求，所以，实际应用中对信息进行分类时，即使出现分类数据匮乏以及分类错误的现象，也基本不会影响生成预设的用户信息数据库。
由以上可见，上述各个实施例提供的方案中，根据当前所有用户对已有的信息分类感兴趣的概率以及预设的信息库中记录的每一条信息属于各个信息分类的概率，预测各个用户对预设的信息库中记录的每一条信息感兴趣的概率，并根据预测结果确定待推送信息，进而进行信息推送。可见，应用上述各个实施例提供的方案确定待推送信息时，考虑了用户对各条信息感兴趣的概率，因此，能够基于用户有针对性的进行信息推送，提高了用户体验。
与上述信息推送方法相对应，本申请实施例还提供了一种信息推送装置。
图3为本申请实施例提供的一种信息推送装置的结构示意图，该装置包括：
推送请求接收模块301，用于接收针对目标用户的信息推送请求；
概率获得模块302，用于从预设的用户信息数据库中获得所述目标用户对预设的信息库中记录的各条信息感兴趣的概率，其中，所述预设的用户信息数据库用于记录各 个用户对所述预设的信息库中记录的各条信息感兴趣的概率；
待推送信息确定模块303，用于根据所获得的概率由高到低的顺序，从所述预设的信息数据库中记录的各条信息中确定待推送信息；
信息推送模块304，用于推送所述待推送信息。
在本申请的一种具体实现方式中，上述信息推送装置还可以包括：
数据库生成模块，用于生成所述预设的用户信息数据库。
下面通过具体实施例详细介绍如何生成前面涉及到的预设的用户信息数据库。
图4为本申请实施例提供的一种数据库生成装置的结构示意图，该装置为数据库生成模块的一种具体装置，包括：
信息分类获得子模块401，用于获得当前已有的信息分类；
第一概率矩阵获得子模块402，用于获得当前所有用户对上述各个信息分类感兴趣的概率，生成概率矩阵A；
第二概率矩阵获得子模块403，用于获得所述预设信息库中记录的各条信息属于上述各个信息分类的概率，生成概率矩阵B；
第三概率矩阵获得子模块404，用于根据所述概率矩阵A和所述概率矩阵B，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D；
收敛性判断子模块405，用于判断所述概率矩阵D是否满足预设的收敛性判断条件；
概率矩阵更新子模块406，用于在所述收敛性判断子模块405的判断结果为否的情况下，按照预设的调整规则调整所述概率矩阵D中的各个元素，并根据调整后的元素更新所述概率矩阵D，触发所述收敛性判断子模块405进行判断，直至所述概率矩阵D满足所述预设的收敛性判断条件；
数据库生成子模块407，用于根据所述概率矩阵D生成所述预设的用户信息数据库。
具体的，所述收敛性判断子模块405可以包括：
矩阵分解单元，用于根据预设的矩阵分解算法，将所述概率矩阵D分解为概率矩阵A’和概率矩阵B’，其中，所述概率矩阵A’为与所述概率矩阵A对应的矩阵，所述概率矩阵B’为与所述概率矩阵B对应的矩阵；
收敛性判断单元，用于根据所述概率矩阵A’、所述概率矩阵B’以及所述概率矩阵D，判断所述概率矩阵D是否满足预设的收敛性判断条件。
具体的，所述收敛性判断单元可以包括：
误差计算子单元，用于根据以下表达式计算预测概率与真实概率之间的误差 argminα,βL(D)，
Figure PCTCN2016083732-appb-000013
其中，
Figure PCTCN2016083732-appb-000014
表示所述概率矩阵A’第i行的元素组成的向量，
Figure PCTCN2016083732-appb-000015
表示所述概率矩阵B’第j行的元素组成的向量，λ表示调整系数，dij表示所述概率矩阵D的元素；
收敛性判断子单元，用于根据所述误差argminα,βL(D)判断所述概率矩阵D是否满足预设的收敛性判断条件。
具体的，所述收敛性判断子单元，
可以具体用于判断所述误差argminα,βL(D)是否小于预设的第一误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件；或
可以具体用于在所述概率矩阵D为根据调整后的元素更新后的概率矩阵的情况下，判断所述误差argminα,βL(D)与误差argminα,βL(D)’之间的绝对差值是否小于预设的第二误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件，其中，所述误差argminα,βL(D)’表示基于更新之前的所述概率矩阵D计算得到的误差。
在本申请的一种较佳实现方式中，所述数据库生成模块还可以包括：
矩阵更新子模块，用于在判断得所述概率矩阵D满足预设的收敛性判断条件的情况下，根据所述概率矩阵A‘和所述概率矩阵B’更新所述概率矩阵A和所述概率矩阵B。
具体的，所述第三概率矩阵获得子模块，可以具体用于根据以下表达式，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D，
Figure PCTCN2016083732-appb-000016
其中，dij表示所述概率矩阵D的元素，αik表示所述概率矩阵A第i行第k列的元素，βjk表示所述概率矩阵B第j行第k列的元素，k表示所获得的信息分类的数量。
由以上可见，上述各个实施例提供的方案中，根据当前所有用户对已有的信息分类感兴趣的概率以及预设的信息库中记录的每一条信息属于各个信息分类的概率，预测各个用户对预设的信息库中记录的每一条信息感兴趣的概率，并根据预测结果确定待推送信息，进而进行信息推送。可见，应用上述各个实施例提供的方案确定待推送信息时，考虑了用户对各条信息感兴趣的概率，因此，能够基于用户有针对性的进行信息推送，提高了用户体验。
对于装置实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
为了实现上述实施例，本发明的实施例还提出一种信息推送装置，包括：壳体501、处理器502、存储器503、电路板504和电源电路505，其中，电路板504安置在壳体501围成的空间内部，处理器502和存储器503设置在电路板504上；电源电路505，用于为电子设备的各个电路或器件供电；存储器503用于存储可执行程序代码；处理器502通过读取存储器503中存储的可执行程序代码来运行与可执行程序代码对应的程序，以用于执行以下步骤：
S101’：接收针对目标用户的信息推送请求。
S102’：从预设的用户信息数据库中获得目标用户对预设的信息库中记录的各条信息感兴趣的概率。
S103’：根据所获得的概率由高到低的顺序，从预设的信息数据库中记录的各条信息中确定待推送信息。
S104’：推送所述待推送信息。
在本发明的实施例中，处理器502还通过读取存储器503中存储的可执行程序代码来运行与可执行程序代码对应的程序，以用于执行以下步骤：
S201’：获得当前已有的信息分类。
S202’：获得当前所有用户对上述各个信息分类感兴趣的概率，生成概率矩阵A。
S203’：获得预设信息库中记录的各条信息属于上述各个信息分类的概率，生成概率矩阵B。
S204’：根据概率矩阵A和概率矩阵B，预测各个用户对预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D。
S205’：判断概率矩阵D是否满足预设的收敛性判断条件，若不满足，执行S206’若满足，执行S207’。
S206’：按照预设的调整规则调整概率矩阵D中的各个元素，并根据调整后的元素更新概率矩阵D，返回S205’。
S207’：根据概率矩阵D生成预设的用户信息数据库。
应当理解上述步骤中的执行细节可参考上述对应的方法实施例，在此不再赘述。
为了实现上述实施例，本发明的实施例还提出一种计算程序，当其在处理器上运行时，执行下述步骤：
接收针对目标用户的信息推送请求；
从预设的用户信息数据库中获得所述目标用户对预设的信息库中记录的各条信息感兴趣的概率，其中，所述预设的用户信息数据库用于记录各个用户对所述预设的信 息库中记录的各条信息感兴趣的概率；
根据所获得的概率由高到低的顺序，从所述预设的信息数据库中记录的各条信息中确定待推送信息；
推送所述待推送信息。
应当理解上述步骤中的执行细节可参考上述对应的方法实施例，在此不再赘述。
为了实现上述实施例，本发明的实施例还提出一种计算机可读存储介质，计算机存储介质存储有一个或者多个模块，当一个或者多个模块被终端执行时，使得终端执行下述步骤：
接收针对目标用户的信息推送请求；
从预设的用户信息数据库中获得所述目标用户对预设的信息库中记录的各条信息感兴趣的概率，其中，所述预设的用户信息数据库用于记录各个用户对所述预设的信息库中记录的各条信息感兴趣的概率；
根据所获得的概率由高到低的顺序，从所述预设的信息数据库中记录的各条信息中确定待推送信息；
推送所述待推送信息。
应当理解上述步骤中的执行细节可参考上述对应的方法实施例，在此不再赘述。
需要说明的是，在本文中，诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来，而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
流程图中或在此以其他方式描述的任何过程或方法描述可以被理解为，表示包括一个或更多个用于实现特定逻辑功能或过程的步骤的可执行指令的代码的模块、片段或部分，并且本发明的优选实施方式的范围包括另外的实现，其中可以不按所示出或讨论的顺序，包括根据所涉及的功能按基本同时的方式或按相反的顺序，来执行功能，这应被本发明的实施例所属技术领域的技术人员所理解。
在流程图中表示或在此以其他方式描述的逻辑和/或步骤，例如，可以被认为是用 于实现逻辑功能的可执行指令的定序列表，可以具体实现在任何计算机可读介质中，以供指令执行系统、装置或设备(如基于计算机的系统、包括处理器的系统或其他可以从指令执行系统、装置或设备取指令并执行指令的系统)使用，或结合这些指令执行系统、装置或设备而使用。就本说明书而言，"计算机可读介质"可以是任何可以包含、存储、通信、传播或传输程序以供指令执行系统、装置或设备或结合这些指令执行系统、装置或设备而使用的装置。计算机可读介质的更具体的示例(非穷尽性列表)包括以下：具有一个或多个布线的电连接部(电子装置)，便携式计算机盘盒(磁装置)，随机存取存储器(RAM)，只读存储器(ROM)，可擦除可编辑只读存储器(EPROM或闪速存储器)，光纤装置，以及便携式光盘只读存储器(CDROM)。另外，计算机可读介质甚至可以是可在其上打印所述程序的纸或其他合适的介质，因为可以例如通过对纸或其他介质进行光学扫描，接着进行编辑、解译或必要时以其他合适方式进行处理来以电子方式获得所述程序，然后将其存储在计算机存储器中。
应当理解，本发明的各部分可以用硬件、软件、固件或它们的组合来实现。在上述实施方式中，多个步骤或方法可以用存储在存储器中且由合适的指令执行系统执行的软件或固件来实现。例如，如果用硬件来实现，和在另一实施方式中一样，可用本领域公知的下列技术中的任一项或他们的组合来实现：具有用于对数据信号实现逻辑功能的逻辑门电路的离散逻辑电路，具有合适的组合逻辑门电路的专用集成电路，可编程门阵列(PGA)，现场可编程门阵列(FPGA)等。
本技术领域的普通技术人员可以理解实现上述实施例方法携带的全部或部分步骤是可以通过程序来指令相关的硬件完成，所述的程序可以存储于一种计算机可读存储介质中，该程序在执行时，包括方法实施例的步骤之一或其组合。
此外，在本发明各个实施例中的各功能单元可以集成在一个处理模块中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个模块中。上述集成的模块既可以采用硬件的形式实现，也可以采用软件功能模块的形式实现。所述集成的模块如果以软件功能模块的形式实现并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中。
本领域普通技术人员可以理解实现上述方法实施方式中的全部或部分步骤是可以通过程序来指令相关的硬件来完成，所述的程序可以存储于计算机可读取存储介质中，这里所称得的存储介质，如：ROM/RAM、磁碟、光盘等。
以上所述仅为本申请的较佳实施例而已，并非用于限定本申请的保护范围。凡在本申请的精神和原则之内所作的任何修改、等同替换、改进等，均包含在本申请的保护范围内。

Claims (17)

  1. 一种信息推送方法，其特征在于，所述方法包括：
    接收针对目标用户的信息推送请求；
    从预设的用户信息数据库中获得所述目标用户对预设的信息库中记录的各条信息感兴趣的概率，其中，所述预设的用户信息数据库用于记录各个用户对所述预设的信息库中记录的各条信息感兴趣的概率；
    根据所获得的概率由高到低的顺序，从所述预设的信息数据库中记录的各条信息中确定待推送信息；
    推送所述待推送信息。
  2. 根据权利要求1所述的方法，其特征在于，按照以下方式生成所述预设的用户信息数据库：
    获得当前已有的信息分类；
    获得当前所有用户对上述各个信息分类感兴趣的概率，生成概率矩阵A；
    获得所述预设信息库中记录的各条信息属于上述各个信息分类的概率，生成概率矩阵B；
    根据所述概率矩阵A和所述概率矩阵B，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D；
    判断所述概率矩阵D是否满足预设的收敛性判断条件，若不满足，按照预设的调整规则调整所述概率矩阵D中的各个元素，并根据调整后的元素更新所述概率矩阵D，返回所述判断所述概率矩阵D是否满足预设的收敛性判断条件的步骤，直至所述概率矩阵D满足所述预设的收敛性判断条件；
    根据所述概率矩阵D生成所述预设的用户信息数据库。
  3. 根据权利要求2所述的方法，其特征在于，所述判断所述概率矩阵D是否满足预设的收敛性判断条件，包括：
    根据预设的矩阵分解算法，将所述概率矩阵D分解为概率矩阵A’和概率矩阵B’，其中，所述概率矩阵A’为与所述概率矩阵A对应的矩阵，所述概率矩阵B’为与所述概率矩阵B对应的矩阵；
    根据所述概率矩阵A’、所述概率矩阵B’以及所述概率矩阵D，判断所述概率矩阵D是否满足预设的收敛性判断条件。
  4. 根据权利要求3所述的方法，其特征在于，所述根据所述概率矩阵A’、所述概率矩阵B’以及所述概率矩阵D，判断所述概率矩阵D是否满足预设的收敛性判断条 件，包括：
    根据以下表达式计算预测概率与真实概率之间的误差argminα，βL(D)，
    Figure PCTCN2016083732-appb-100001
    其中，
    Figure PCTCN2016083732-appb-100002
    表示所述概率矩阵A’第i行的元素组成的向量，
    Figure PCTCN2016083732-appb-100003
    表示所述概率矩阵B’第j行的元素组成的向量，λ表示调整系数，dij表示所述概率矩阵D的元素；
    根据所述误差argminα,βL(D)判断所述概率矩阵D是否满足预设的收敛性判断条件。
  5. 根据权利要求4所述的方法，其特征在于，所述根据所述误差argminα,βL(D)判断所述概率矩阵D是否满足预设的收敛性判断条件，包括：
    判断所述误差argminα,βL(D)是否小于预设的第一误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件；或
    在所述概率矩阵D为根据调整后的元素更新后的概率矩阵的情况下，判断所述误差argminα,βL(D)与误差argminα,βL(D)’之间的绝对差值是否小于预设的第二误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件，其中，所述误差argminα,βL(D)’表示基于更新之前的所述概率矩阵D计算得到的误差。
  6. 根据权利要求3-5中任一项所述的方法，其特征在于，所述方法还包括：
    在判断得所述概率矩阵D满足预设的收敛性判断条件的情况下，根据所述概率矩阵A‘和所述概率矩阵B’更新所述概率矩阵A和所述概率矩阵B。
  7. 根据权利要求2-6中任一项所述的方法，其特征在于，所述根据所述概率矩阵A和所述概率矩阵B，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D，包括：
    根据以下表达式，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D，
    Figure PCTCN2016083732-appb-100004
    其中，dij表示所述概率矩阵D的元素，αik表示所述概率矩阵A第i行第k列的元素，βjk表示所述概率矩阵B第j行第k列的元素，k表示所获得的信息分类的数量。
  8. 一种信息推送装置，其特征在于，所述装置包括：
    推送请求接收模块，用于接收针对目标用户的信息推送请求；
    概率获得模块，用于从预设的用户信息数据库中获得所述目标用户对预设的信息库中记录的各条信息感兴趣的概率，其中，所述预设的用户信息数据库用于记录各个用户对所述预设的信息库中记录的各条信息感兴趣的概率；
    待推送信息确定模块，用于根据所获得的概率由高到低的顺序，从所述预设的信息数据库中记录的各条信息中确定待推送信息；
    信息推送模块，用于推送所述待推送信息。
  9. 根据权利要求8所述的装置，其特征在于，所述装置还包括：
    数据库生成模块，用于生成所述预设的用户信息数据库；
    其中，所述数据库生成模块，包括：
    信息分类获得子模块，用于获得当前已有的信息分类；
    第一概率矩阵获得子模块，用于获得当前所有用户对上述各个信息分类感兴趣的概率，生成概率矩阵A；
    第二概率矩阵获得子模块，用于获得所述预设信息库中记录的各条信息属于上述各个信息分类的概率，生成概率矩阵B；
    第三概率矩阵获得子模块，用于根据所述概率矩阵A和所述概率矩阵B，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D；
    收敛性判断子模块，用于判断所述概率矩阵D是否满足预设的收敛性判断条件；
    概率矩阵更新子模块，用于在所述收敛性判断子模块的判断结果为否的情况下，按照预设的调整规则调整所述概率矩阵D中的各个元素，并根据调整后的元素更新所述概率矩阵D，触发所述收敛性判断子模块进行判断，直至所述概率矩阵D满足所述预设的收敛性判断条件；
    数据库生成子模块，用于根据所述概率矩阵D生成所述预设的用户信息数据库。
  10. 根据权利要求9所述的装置，其特征在于，所述收敛性判断子模块，包括：
    矩阵分解单元，用于根据预设的矩阵分解算法，将所述概率矩阵D分解为概率矩阵A’和概率矩阵B’，其中，所述概率矩阵A’为与所述概率矩阵A对应的矩阵，所述概率矩阵B’为与所述概率矩阵B对应的矩阵；
    收敛性判断单元，用于根据所述概率矩阵A’、所述概率矩阵B’以及所述概率矩阵D，判断所述概率矩阵D是否满足预设的收敛性判断条件。
  11. 根据权利要求10所述的装置，其特征在于，所述收敛性判断单元，包括：
    误差计算子单元，用于根据以下表达式计算预测概率与真实概率之间的误差argminα，βL(D)，
    Figure PCTCN2016083732-appb-100005
    其中，
    Figure PCTCN2016083732-appb-100006
    表示所述概率矩阵A’第i行的元素组成的向量，
    Figure PCTCN2016083732-appb-100007
    表示所述概率矩阵B’第j行的元素组成的向量，λ表示调整系数，dij表示所述概率矩阵D的元素；
    收敛性判断子单元，用于根据所述误差argminα,βL(D)判断所述概率矩阵D是否满足预设的收敛性判断条件。
  12. 根据权利要求11所述的装置，其特征在于，所述收敛性判断子单元，
    具体用于判断所述误差argminα,βL(D)是否小于预设的第一误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件；或
    具体用于在所述概率矩阵D为根据调整后的元素更新后的概率矩阵的情况下，判断所述误差argminα,βL(D)与误差argminα,βL(D)’之间的绝对差值是否小于预设的第二误差阈值，若小于，判定所述概率矩阵D满足预设的收敛性判断条件，其中，所述误差argminα,βL(D)’表示基于更新之前的所述概率矩阵D计算得到的误差。
  13. 根据权利要求10-12中任一项所述的装置，其特征在于，所述数据库生成模块，还包括：
    矩阵更新子模块，用于在判断得所述概率矩阵D满足预设的收敛性判断条件的情况下，根据所述概率矩阵A‘和所述概率矩阵B’更新所述概率矩阵A和所述概率矩阵B。
  14. 根据权利要求9所述的装置，其特征在于，所述第三概率矩阵获得子模块，具体用于根据以下表达式，预测各个用户对所述预设的信息库中记录的每一条信息感兴趣的概率，获得概率矩阵D，
    Figure PCTCN2016083732-appb-100008
    其中，dij表示所述概率矩阵D的元素，αik表示所述概率矩阵A第i行第k列的元素，βjk表示所述概率矩阵B第j行第k列的元素，k表示所获得的信息分类的数量。
  15. 一种信息推送装置，其特征在于，包括：壳体、处理器、存储器、电路板和电源电路，其中，电路板安置在壳体围成的空间内部，处理器和存储器设置在电路板上；电源电路，用于为信息的推送装置的各个电路或器件供电；存储器用于存储可执行程序代码；处理器通过读取存储器中存储的可执行程序代码来运行与可执行程序代码对应的程序，以用于执行权利要求1至7任一项所述的信息推送方法。
  16. 一种计算机程序，其特征在于，当其在处理器上运行时，执行权利要求1至7任一项所述的信息推送方法。
  17. 一种计算机可读存储介质，其特征在于，所述计算机存储介质存储有一个或者多个模块，当所述一个或者多个模块被终端执行时，使得终端装置：执行权利要求1至7任一项所述的信息推送方法。
PCT/CN2016/083732 2015-11-06 2016-05-27 信息推送方法及装置 WO2017075980A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510752379.6A CN105260477A (zh) 2015-11-06 2015-11-06 一种信息推送方法及装置
CN201510752379.6 2015-11-06

Publications (1)

Publication Number Publication Date
WO2017075980A1 true WO2017075980A1 (zh) 2017-05-11

Family

ID=55100167

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/083732 WO2017075980A1 (zh) 2015-11-06 2016-05-27 信息推送方法及装置

Country Status (2)

Country Link
CN (1) CN105260477A (zh)
WO (1) WO2017075980A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110502702A (zh) * 2019-07-09 2019-11-26 阿里巴巴集团控股有限公司 用户行为预测方法以及装置
CN113590926A (zh) * 2020-04-30 2021-11-02 北京爱笔科技有限公司 用户兴趣的识别方法、装置、设备及计算机可读介质

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105260477A (zh) * 2015-11-06 2016-01-20 北京金山安全软件有限公司 一种信息推送方法及装置
CN106503226A (zh) * 2016-10-28 2017-03-15 努比亚技术有限公司 信息推送方法与装置
CN108229994A (zh) * 2016-12-21 2018-06-29 北京金山云网络技术有限公司 一种信息推送方法及装置
CN108694182B (zh) * 2017-04-07 2021-03-02 北京嘀嘀无限科技发展有限公司 活动推送方法、活动推送装置和服务器
US10922717B2 (en) 2017-04-07 2021-02-16 Beijing Didi Infinity Technology And Development Co., Ltd. Systems and methods for activity recommendation
CN109379410B (zh) * 2018-09-21 2019-11-19 北京达佳互联信息技术有限公司 信息推送方法、装置、服务器以及存储介质
CN111259302B (zh) * 2020-01-19 2023-04-07 深圳市雅阅科技有限公司 信息推送方法、装置及电子设备

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101986298A (zh) * 2010-10-28 2011-03-16 浙江大学 用于在线论坛的信息实时推荐方法
CN102436512A (zh) * 2012-01-17 2012-05-02 电子科技大学 一种基于偏好度的网页文本内容管控方法
CN103489117A (zh) * 2012-06-12 2014-01-01 深圳市腾讯计算机系统有限公司 信息投放方法和系统
US20140143324A1 (en) * 2012-11-17 2014-05-22 Samuel Lessin Prompting social networking system users in a newsfeed to provide additional user profile information
CN105260477A (zh) * 2015-11-06 2016-01-20 北京金山安全软件有限公司 一种信息推送方法及装置

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101226557B (zh) * 2008-02-22 2010-07-14 中国科学院软件研究所 一种高效的关联主题模型数据处理方法
CN102346899A (zh) * 2011-10-08 2012-02-08 亿赞普(北京)科技有限公司 一种基于用户行为的广告点击率预测方法和装置
CN104217334A (zh) * 2013-06-05 2014-12-17 北京京东尚科信息技术有限公司 一种产品信息推荐方法、装置和系统

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101986298A (zh) * 2010-10-28 2011-03-16 浙江大学 用于在线论坛的信息实时推荐方法
CN102436512A (zh) * 2012-01-17 2012-05-02 电子科技大学 一种基于偏好度的网页文本内容管控方法
CN103489117A (zh) * 2012-06-12 2014-01-01 深圳市腾讯计算机系统有限公司 信息投放方法和系统
US20140143324A1 (en) * 2012-11-17 2014-05-22 Samuel Lessin Prompting social networking system users in a newsfeed to provide additional user profile information
CN105260477A (zh) * 2015-11-06 2016-01-20 北京金山安全软件有限公司 一种信息推送方法及装置

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110502702A (zh) * 2019-07-09 2019-11-26 阿里巴巴集团控股有限公司 用户行为预测方法以及装置
CN110502702B (zh) * 2019-07-09 2023-03-24 创新先进技术有限公司 用户行为预测方法以及装置
CN113590926A (zh) * 2020-04-30 2021-11-02 北京爱笔科技有限公司 用户兴趣的识别方法、装置、设备及计算机可读介质

Also Published As

Publication number Publication date
CN105260477A (zh) 2016-01-20

Similar Documents

Publication Publication Date Title
WO2017075980A1 (zh) 信息推送方法及装置
US10846757B2 (en) Automated system and method for creating machine-generated advertisements
CN107613022B (zh) 内容推送方法、装置及计算机设备
US10558852B2 (en) Predictive analysis of target behaviors utilizing RNN-based user embeddings
US20190354810A1 (en) Active learning to reduce noise in labels
US10599770B1 (en) Generating author vectors
JP6588572B2 (ja) 情報推薦方法および情報推薦装置
US8719192B2 (en) Transfer of learning for query classification
WO2019174423A1 (zh) 实体情感分析方法及相关装置
US10348550B2 (en) Method and system for processing network media information
US20180197087A1 (en) Systems and methods for retraining a classification model
US9785717B1 (en) Intent based search result interaction
US20170032280A1 (en) Engagement estimator
US9715486B2 (en) Annotation probability distribution based on a factor graph
US11763084B2 (en) Automatic formulation of data science problem statements
US20170315996A1 (en) Focused sentiment classification
US9286379B2 (en) Document quality measurement
US11250219B2 (en) Cognitive natural language generation with style model
US20160063376A1 (en) Obtaining user traits
US11948095B2 (en) Method and system for recommending digital content
US11275994B2 (en) Unstructured key definitions for optimal performance
US20200409948A1 (en) Adaptive Query Optimization Using Machine Learning
US20230045330A1 (en) Multi-term query subsumption for document classification
US20150095275A1 (en) Massive rule-based classification engine
US20210357699A1 (en) Data quality assessment for data analytics

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16861251

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16861251

Country of ref document: EP

Kind code of ref document: A1