WO2019128317A1 - 物品推送方法、装置及服务器、计算设备及存储介质 - Google Patents

物品推送方法、装置及服务器、计算设备及存储介质 Download PDF

Info

Publication number
WO2019128317A1
WO2019128317A1 PCT/CN2018/105841 CN2018105841W WO2019128317A1 WO 2019128317 A1 WO2019128317 A1 WO 2019128317A1 CN 2018105841 W CN2018105841 W CN 2018105841W WO 2019128317 A1 WO2019128317 A1 WO 2019128317A1
Authority
WO
WIPO (PCT)
Prior art keywords
item
user
click
score
relevance score
Prior art date
Application number
PCT/CN2018/105841
Other languages
English (en)
French (fr)
Inventor
潘岸腾
Original Assignee
广州优视网络科技有限公司
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 广州优视网络科技有限公司 filed Critical 广州优视网络科技有限公司
Publication of WO2019128317A1 publication Critical patent/WO2019128317A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00Commerce
    • G06Q30/06Buying, selling or leasing transactions
    • G06Q30/0601Electronic shopping [e-shopping]
    • G06Q30/0631Item recommendations

Definitions

  • the present invention relates to the field of data mining processing, and in particular, to an item pushing method and device, a server, a computing device, and a storage medium.
  • the recommendation system is considered to be an effective method to solve these problems. It mines the user's historical behavior, models the user's interest, and predicts the user's future behavior, thus establishing users and content. Relationship. For example, in an application store, there is a type of scenario in which a user clicks on and downloads an application to recommend a batch of other applications, and how to determine the application to be recommended.
  • the current recommendation method is based on an item or a user-based collaborative filtering algorithm, collaborative filtering.
  • the shortcomings of the algorithm include sparseness problem, scalability problem, new user problem, etc. In the current era of information types and expressions, there is a need for a new recommendation algorithm to better recommend items for users.
  • the object of the present invention is to provide an item information pushing method and device and a corresponding server, which proposes a new item pushing algorithm based on user click behavior to improve the accuracy of item pushing and improve user experience.
  • the present invention provides an item pushing method, including:
  • the item other than the first item in the item library as the second item, acquiring a user group composed of a user who clicks on the first item and exposes the second item, according to each user in the user group, the first item and the second item Recording the click behavior of the item, and calculating a relevance score of the second item and the first item;
  • the second item is pushed to the user clicking the first item.
  • the calculating according to the click behavior record of the first item and the second item by each user in the user group, calculating a relevance score of the second item and the first item, including:
  • the average of the initial relevance scores is counted to obtain a relevance score for the second item and the first item.
  • the initial relevance scores of the second item and the first item are respectively calculated according to respective click behavior records of each user in the user group, including:
  • the click behavior record includes a user's click time of the item, a click scene, and a click type record;
  • the relevance score of the second item and the first item is increased by A
  • the correlation score of the second item and the first item is increased by B
  • the second item is The relevance score of the first item is increased by C;
  • An initial correlation score of the second item and the first item is calculated based on one or more of the A, B, C, D, X, and Y.
  • the pushing the second item to the user who clicks on the first item according to the relevance score includes:
  • the corresponding second item is pushed to the user based on the item recommendation list.
  • the item is an application
  • the item library is an application library
  • the present invention also provides an article pushing device, comprising:
  • a first calculation module a user group for acquiring a user who clicks on the first item and exposes the second item, and calculates a second item and the first item according to the click behavior record of the first item and the second item by each user in the user group The relevance score of an item;
  • a second calculation module configured to use the first calculation module to calculate a correlation score between the second item and the first item in the item library by using the other items other than the first item as the second item;
  • Push module for pushing a second item to a user who clicks on the first item according to the relevance score.
  • the performing, by the first computing module includes:
  • the average of the initial relevance scores is counted to obtain a relevance score for the second item and the first item.
  • the initial correlation scores of the second item and the first item are respectively calculated according to the click behavior records of each of the respective users in the user group, including:
  • the click behavior record includes a user's click time of the item, a click scene, and a click type record;
  • the relevance score of the second item and the first item is increased by A
  • the correlation score of the second item and the first item is increased by B
  • the second item is The relevance score of the first item is increased by C;
  • An initial correlation score of the second item and the first item is calculated based on one or more of the A, B, C, D, X, and Y.
  • the pushing module execution comprises:
  • the corresponding second item is pushed to the user according to the item recommendation list.
  • the present invention also provides a server, including:
  • One or more processors are One or more processors;
  • One or more applications wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications configured to execute The article pushing method according to the first aspect.
  • the present invention also provides a computing device, including:
  • a memory having executable code stored thereon that, when executed by the processor, causes the processor to perform the item push method of any of the above.
  • the present invention also provides a non-transitory machine readable storage medium having stored thereon executable code that, when executed by a processor of an electronic device, causes the processor to perform The item pushing method according to any one of the preceding claims.
  • the present invention has the following advantages:
  • the invention provides an item pushing algorithm based on a user click behavior, limiting a sample of data statistics to a user group composed of a user who clicks on the first item and exposes the second item, and then pairs the first item according to each user in the user group. And the click behavior record of the second item, calculating the relevance score of the second item and the first item, eliminating the influence of the irrelevant data, improving the accuracy of the correlation score calculation between the items, and then dividing the score according to the relevance Push items with higher relevance to improve the accuracy of item push.
  • the recommendation algorithm of the present invention records three variables of click time, click scene and click type in the click behavior of the user, records each user behavior through the three-dimensional vector, and then calculates each item in the item library based on the multi-dimensional vector correlation rule.
  • the items are more interesting to the user, improve the accuracy of the item push, and enhance the user experience.
  • FIG. 1 is a schematic flow chart of an embodiment of an article pushing method according to the present invention.
  • FIG. 2 is a schematic flow chart of another embodiment of an article pushing method according to the present invention.
  • FIG. 3 is a schematic flow chart of still another embodiment of an article pushing method according to the present invention.
  • Figure 4 is a schematic view showing an embodiment of the article pushing device of the present invention.
  • FIG. 5 is a schematic structural diagram of an embodiment of a server according to the present invention.
  • FIG. 6 is a schematic structural diagram of a computing device according to an embodiment of the present invention.
  • the present invention provides an item pushing method, as shown in FIG. 1, comprising:
  • S100 Acquire a user group composed of a user who clicks on the first item and exposes the second item, and calculates a correlation between the second item and the first item according to the click behavior record of each user in the user group on the first item and the second item. Score.
  • the degree of association between the second item and the first item is represented by a relevance score of the second item and the first item, and when the correlation score of the second item and the first item is to be calculated, first A sample of the data statistics needs to be determined.
  • a user group composed of users who have clicked on the first item and exposed the second item in the first item is selected as a sample of the data statistics of the embodiment, that is, the user group.
  • the click behavior record of the first item and the second item is used as a sample of the data statistics of the embodiment, and for each user in the user group, the click behavior record of each user for the first item and the second item is acquired as data. Sample the statistics and then perform subsequent operations.
  • the click behavior record is a click action of the user recorded when the user touches the first item and exposes the second item to the user, and each click operation generates a corresponding click behavior record. For example, calculating a correlation score between the second item and the first item according to each click behavior record of each user, and then counting the click behavior records of all users of the user group to calculate the correlation between the second item and the first item. Sex score.
  • S200 The other items other than the first item in the item library are used as the second item, and the correlation scores of the second item and the first item in the item library are respectively calculated according to the calculation method of S100.
  • S100 illustrates a method for calculating a correlation score between the second item and the first item.
  • After calculating a correlation score of the second item and the first item it is necessary to calculate all items in the item library except the first item.
  • Correlation score with the first item at this time, the other items other than the first item in the item library are taken as the second item, and then the user who clicks on the first item and exposes the second item is sequentially obtained according to the method of S100.
  • the user group is composed, and according to the click behavior record of each user in the user group for the first item and the second item, the correlation scores of the second item and the first item are sequentially calculated, thereby obtaining the item library.
  • the first item is included in the item library of the embodiment. It can be understood that a person skilled in the art can store the first item outside the item library according to the solution of the embodiment, and can also calculate the item library. The relevance score of the item to the first item, and therefore the storage relationship of the first item with the item library should not be considered as limiting the inventive solution. For example, there are five items N1, N2, N3, N4, N5 in the item library, and N1 is the first item, and the correlation scores of the second items N2, N3, N4, N5 and the first item N1 are respectively calculated.
  • the subsequent user clicks N1 one or more of the second items N2, N3, N4, and N5 are pushed to the user according to the relevance score; further, the second item N1, N3 is calculated by using N2 as the first item. a correlation score between N4 and N5 and the first item N2.
  • the subsequent user clicks N2 one or more of the second items N1, N3, N4, and N5 are pushed to the user according to the relevance score; and so on. According to the method, the correlation score between the items in the item library is calculated.
  • S300 Push the second item to the user who clicks on the first item according to the relevance score.
  • the application scenario of the embodiment is an application store.
  • the application store after the user clicks on an application (the first application), the application store simultaneously exposes other applications (second application) associated with the application.
  • An embodiment defines a scene in which the first application is clicked and the second application is exposed to be displayed as clicking on the first item and exposing the second item, and recording the user's click behavior in the scene, and then clicking the first
  • the item exposes the user group composed of the user of the second item as a sample of the data statistics, and the second item and the first item can be calculated according to each click behavior record of the first item and the second item by each user in the user group.
  • the item library ie, the application
  • the other item except the first item in the application library on the store is used as the second item, and the correlation score of the other item in the item library and the first item is calculated, thereby obtaining the object All libraries and second article's relevance scores of the first article, i.e., to obtain the degree of association between the applications in the application store, and further when the user clicks a first application, the second application to the user according push relevance score.
  • a correlation score between the second item and the first item is calculated according to a click behavior record of each user in the user group on the first item and the second item. Values, including:
  • S101 Calculate an initial relevance score of the second item and the first item according to respective click behavior records of each user in the user group;
  • S102 Count an average of the initial correlation scores to obtain a correlation score between the second item and the first item.
  • a click behavior record of a user calculates that the relevance score of the second item and the first item is defined as an initial relevance score, and then the initial score obtained by counting the click behavior records of all users in the user group is calculated.
  • the average value of the relevance score is used as the correlation score between the second item and the first item.
  • the correlation score of the second item and the first item is calculated as follows:
  • Score(Item1, Item2) calculates the initial relevance score of the second item and the first item for a click activity record of the user, and then The calculated initial correlation score is calculated as an average value, thereby calculating a correlation score Sim (Item1, Item 2) of the second item and the first item.
  • the initial relevance scores of the second item and the first item are respectively calculated according to the respective click behavior records of each user in the user group, including:
  • the click behavior record includes a user's click time of the item, a click scene, and a click type record;
  • the relevance score of the second item and the first item is increased by A
  • the correlation score of the second item and the first item is increased by B
  • the second item is The relevance score of the first item is increased by C;
  • An initial correlation score of the second item and the first item is calculated based on one or more of the A, B, C, D, X, and Y.
  • the click behavior record includes a user's click time of the item, a click scene, and a click type record, where the click time records the time when the user clicks on the item, and the click scene records the user click.
  • the click type records the type of click behavior of the second item that has been exposed by the user.
  • each click behavior is recorded as:
  • Vc is the user's click behavior vector for the item, where SourceItem is the user's click behavior scene for the item; Time is the time point at which the user clicks on the item; ActionType indicates the user's item The type of click behavior.
  • the initial relevance score of the second item and the first item is recorded as Score(Item1, Item2), and the calculation rule is as follows:
  • a time interval between the time and the click time of the second item the time interval being a minimum time interval between the user's click time of the first item and the click time of the second item, and assigning different relevance points for different time intervals a value, when the time interval is less than or equal to the first preset value, the relevance score of the second item and the first item is increased by A, and when the time interval is less than or equal to the second preset value, the second item
  • the correlation score with the first item is increased by B.
  • the correlation score of the second item and the first item is increased by C; for example, Time1 and Time2 are within the same hour.
  • the S300 includes:
  • S301 Sort the second item of the item library according to the relevance score
  • each first item has a corresponding second item
  • the second item in the item library is sorted according to the size of the relevance score, in the sort
  • the items in the recommendation list are sorted according to the relevance scores from large to small, and the correlation score is larger, indicating the second item and the first item.
  • the correlation score is larger, indicating the second item and the first item.
  • the first item N1 is sorted according to the correlation score with N1 to obtain N3, N5, N2, and N4, and three second items N3, N5, and N2 are selected to form an item recommendation list of N1.
  • the present invention provides an item information pushing device, as shown in FIG. 4, comprising:
  • the first calculation module 100 is configured to acquire a user group composed of a user who clicks on the first item and exposes the second item, and calculates a second item according to the click behavior record of the first item and the second item by each user in the user group. The relevance score of the first item;
  • the second calculating module 200 is configured to use, as the second item, other items in the item library other than the first item, and calculate, by using the first calculating module, a correlation score of the second item in the item library and the first item;
  • the pushing module 300 is configured to push the second item to the user who clicks on the first item according to the relevance score.
  • the degree of association between the second item and the first item is represented by a relevance score of the second item and the first item, and when the correlation score of the second item and the first item is to be calculated, first A sample of the data statistics needs to be determined.
  • a user group composed of users who have clicked on the first item and exposed the second item in the first item is selected as a sample of the data statistics of the embodiment, that is, the user group.
  • the click behavior record of the first item and the second item is used as a sample of the data statistics of the embodiment.
  • the first calculation module 100 acquires the first item and the second item of each user.
  • the click behavior record is a sample of the data statistics, and the click behavior record is a click behavior of the user recorded when the user touches the first item and the second item is exposed to the user, and each click behavior is generated.
  • the first calculation module 100 calculates a second object according to each click behavior record of each user. And a relevance score first item, then click behavior statistics for all users of the user group's records to calculate the relevance score of the second item to the first item.
  • the first calculating module 100 calculates a correlation score between the second item and the first item. After calculating the correlation score of the second item and the first item, the item inventory is calculated. The score of the correlation between all the items outside the first item and the first item.
  • the second calculating module 200 uses the items other than the first item in the item library as the second item, and then sequentially obtains the click on the first item. And exposing the user group composed of the user of the second item, and then calculating the relevance scores of the second item and the first item according to the click behavior record of each user in the user group. In this way, the correlation score of the second item other than the first item in the item library and the first item is obtained.
  • the pushing module 300 can calculate the score of the correlation between the different items and the first item. Determining which item in the item library is more relevant to the first item, and subsequently pushing the second item having higher relevance to the first item to the user according to the relevance score when the user clicks on the first item.
  • the performing, by the first calculating module 100 includes:
  • the average of the initial relevance scores is counted to obtain a relevance score for the second item and the first item.
  • a click behavior record of a user calculates that the relevance score of the second item and the first item is defined as an initial relevance score, and then the initial score obtained by counting the click behavior records of all users in the user group is calculated.
  • the average value of the relevance score is used as the correlation score between the second item and the first item.
  • the correlation score of the second item and the first item is calculated as follows:
  • Score(Item1, Item2) calculates the initial relevance score of the second item and the first item for a click activity record of the user, and then The calculated initial correlation score is calculated as an average value, thereby calculating a correlation score Sim (Item1, Item 2) of the second item and the first item.
  • the initial relevance scores of the second item and the first item are respectively calculated according to the respective click behavior records of each user in the user group, including:
  • the click behavior record includes a user's click time of the item, a click scene, and a click type record;
  • the correlation score of the second item and the first item is increased by A.
  • the correlation score of the second item and the first item is increased by B, and when the time interval is less than or equal to the third preset value, the second item is The relevance score of the first item is increased by C;
  • An initial correlation score of the second item and the first item is calculated based on one or more of the A, B, C, D, X, and Y.
  • the click behavior record includes a user's click time of the item, a click scene, and a click type record, where the click time records the time when the user clicks on the item, and the click scene records the user click.
  • the click type records the type of click behavior of the second item that has been exposed by the user.
  • each click behavior is recorded as:
  • Vc is the user's click behavior vector for the item, where SourceItem is the user's click behavior scene for the item; Time is the time point at which the user clicks on the item; ActionType indicates the user's item The type of click behavior.
  • the initial relevance score of the second item and the first item is recorded as Score(Item1, Item2), and the calculation rule is as follows:
  • a time interval between the time and the click time of the second item the time interval being a minimum time interval between the user's click time of the first item and the click time of the second item, and assigning different relevance points for different time intervals a value, when the time interval is less than or equal to the first preset value, the relevance score of the second item and the first item is increased by A, and when the time interval is less than or equal to the second preset value, the second item
  • the correlation score with the first item is increased by B.
  • the correlation score of the second item and the first item is increased by C; for example, Time1 and Time2 are within the same hour.
  • the pushing module 300 performs:
  • the corresponding second item is pushed to the user based on the item recommendation list.
  • each first item has a corresponding second item
  • the second item in the item library is sorted according to the size of the relevance score, in the sort
  • the items in the recommendation list are sorted according to the relevance scores from large to small, and the correlation score is larger, indicating the second item and the first item. The higher the degree of association, when the subsequent user clicks on the first item, the corresponding second item is pushed to the user according to the item recommendation list.
  • the present invention also provides a server comprising: one or more processors, a memory, and one or more applications, wherein the one or more applications are stored in the memory and It is configured to be executed by the one or more processors configured to perform the item push method of the above-described embodiments.
  • FIG. 5 is a schematic structural diagram of a server according to the present invention, including a processor 503, a memory 505, an input unit 507, and a display unit 509. It will be understood by those skilled in the art that the structural device illustrated in FIG. 5 does not constitute a limitation to all servers, and may include more or less components than those illustrated, or some components may be combined.
  • the memory 505 can be used to store an application 501 and various functional modules, and the processor 503 runs an application 501 stored in the memory 505 to perform various functional applications and data processing of the device.
  • Memory 505 can be internal or external, or both internal and external.
  • the internal memory may include a read only memory (ROM), a programmable ROM (PROM), an electrically programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a flash memory, or a random access memory.
  • ROM read only memory
  • PROM programmable ROM
  • EPROM electrically programmable ROM
  • EEPROM electrically erasable programmable ROM
  • flash memory or a random access memory.
  • the external storage may include a hard disk, a floppy disk, a ZIP disk, a USB disk, a magnetic tape, and the like.
  • the memories disclosed herein include, but are not limited to, these types of memories.
  • the memory 505 disclosed herein is by way of example only, and not limitation.
  • the input unit 507 is for receiving an input of a signal and receiving a keyword input by the user.
  • the input unit 507 can include a touch panel as well as other input devices.
  • the touch panel can collect touch operations on or near the user (such as the user using any suitable object or accessory such as a finger or a stylus on the touch panel or near the touch panel), and according to a preset
  • the program drives the corresponding connection device; other input devices may include, but are not limited to, one or more of a physical keyboard, function keys (such as play control buttons, switch buttons, etc.), trackballs, mice, joysticks, and the like.
  • the display unit 509 can be used to display information input by the user or information provided to the user as well as various menus of the computer device.
  • the display unit 509 can take the form of a liquid crystal display, an organic light emitting diode, or the like.
  • the processor 503 is a control center of the computer device that connects various parts of the entire computer using various interfaces and lines, executes or executes software programs and/or modules stored in the memory 503, and calls data stored in the memory to execute Various functions and processing data.
  • the server includes one or more processors 503, and one or more memories 505, one or more applications 501, wherein the one or more applications 501 are stored in the memory 505 and
  • the configuration is performed by the one or more processors 503 configured to perform the item push method described in the above embodiments.
  • FIG. 6 is a block diagram showing the structure of a computing device that can be used to implement the above described item pushing method in accordance with an embodiment of the present invention.
  • computing device 600 includes a memory 610 and a processor 620.
  • the processor 620 can be a multi-core processor or multiple processors.
  • processor 620 can include a general purpose main processor and one or more special coprocessors, such as a graphics processing unit (GPU), a digital signal processor (DSP), and the like.
  • the processor 620 can be implemented using a customized circuit, such as an Application Specific Integrated Circuit (ASIC) or Field Programmable Gate Array (FPGA).
  • ASIC Application Specific Integrated Circuit
  • FPGA Field Programmable Gate Array
  • Memory 610 can include various types of storage units, such as system memory, read only memory (ROM), and persistent storage.
  • the ROM can store static data or instructions required by the processor 620 or other modules of the computer.
  • the persistent storage device can be a readable and writable storage device.
  • the persistent storage device may be a non-volatile storage device that does not lose stored instructions and data even after the computer is powered off.
  • the persistent storage device employs a mass storage device (eg, magnetic or optical disk, flash memory) as the permanent storage device.
  • the persistent storage device can be a removable storage device (eg, a floppy disk, an optical drive).
  • the system memory can be a readable and writable storage device or a volatile read/write storage device, such as dynamic random access memory.
  • System memory can store instructions and data that some or all of the processors need at runtime.
  • memory 610 may comprise any combination of computer readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash memory, programmable read only memory), and magnetic disks and/or optical disks may also be employed.
  • memory 610 can include removable storage devices that are readable and/or writable, such as a compact disc (CD), a read-only digital versatile disc (eg, a DVD-ROM, a dual layer DVD-ROM), Read-only Blu-ray discs, ultra-density discs, flash cards (such as SD cards, min SD cards, Micro-SD cards, etc.), magnetic floppy disks, and so on.
  • the computer readable storage medium does not include a carrier wave and an instantaneous electronic signal transmitted by wireless or wire.
  • the executable code is stored on the memory 610, and when the executable code is processed by the processor 620, the processor 620 can be caused to execute the item push method described above.
  • the method according to the invention may also be embodied as a computer program or computer program product comprising computer program code instructions for performing the various steps defined above in the above method of the invention.
  • the present invention may be embodied as a non-transitory machine readable storage medium (or computer readable storage medium, or machine readable storage medium) having stored thereon executable code (or computer program, or computer instruction code)
  • executable code or computer program, or computer instruction code
  • a processor of an electronic device or computing device, server, 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

  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本发明公开了一种物品推送方法、装置及服务器、计算设备及存储介质,所述方法包括:将物品库中除第一物品外的其它物品作为第二物品,获取点击第一物品且曝光第二物品的用户组成的用户群,根据所述用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值;根据所述相关性分值,向点击第一物品的用户推送第二物品。本发明提供了一种基于用户点击行为的物品推送算法,并将数据统计的样本限定于特定的用户群,剔除无关数据的影响,提高物品之间相关性分值计算的准确性,从而提高物品推送的准确率,提升用户体验。

Description

物品推送方法、装置及服务器、计算设备及存储介质 【技术领域】
本发明涉及数据挖掘处理领域,具体涉及一种物品推送方法、装置及服务器、计算设备及存储介质。
【背景技术】
随着互联网的快速发展,我们正处于信息过载的时代,用户面对海量的信息很难找到自己真正感兴趣的内容,而内容提供商也很难把优质的内容准确地推送给感兴趣的用户,推荐***应运而生,推荐***被认为是解决这些问题的有效方法,它对用户的历史行为进行挖掘,对用户兴趣进行建模,并对用户未来的行为进行预测,从而建立了用户和内容的关系。例如在应用商店中,有一类场景是用户点击、下载一个应用时向其推荐一批其它应用,而如何确定所要推荐的应用,目前的推荐方法是基于物品或基于用户的协同过滤算法,协同过滤算法存在的缺点包括稀疏问题、可拓展性问题、新用户问题等,在目前信息种类、表达方式越来越多的时代,亟需一种新的推荐算法来更好地为用户推荐物品。
【发明内容】
本发明的目的在于提供一种物品信息推送方法、装置及相应的服务器,通过提出一种新的基于用户点击行为的物品推送算法以提高物品推送的准确率、提升用户体验。
为实现该目的,本发明采用如下技术方案:
第一方面,本发明提供一种物品推送方法,包括:
将物品库中除第一物品外的其它物品作为第二物品,获取点击第一物品且曝光第二物品的用户组成的用户群,根据所述用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值;
根据所述相关性分值,向点击第一物品的用户推送第二物品。
进一步的,所述根据用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值,包括:
根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值;
统计所述初始相关性分值的平均值,得到第二物品与第一物品的相关性分值。
具体的,所述根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值,包括:
获取用户的点击行为记录,所述点击行为记录包括用户对物品的点击时间、点击场景及点击类型记录;
判断用户对第一物品的点击时间与对第二物品的点击时间的时间间隔,当所述时间间隔小于或等于第一预设值时,第二物品与第一物品的相关性分值增加A,当所述时间间隔小于或等于第二预设值时,第二物品与第一物品的相关性分值增加B,当所述时间间隔小于或等于第三预设值时,第二物品与第一物品的相关性分值增加C;
判断用户对第二物品的点击场景,当所述点击场景为用户点击了第一物品后在第一预设场景下推荐第二物品时,第二物品与第一物品的相关性分值增加D;
判断用户对第二物品的点击类型,当所述点击类型为用户点击了第一物品且点击了第二物品时,第二物品与第一物品的相关性分值乘X,当所述点击类型为用户点击了第一物品且未点击第二物品时,第二物品与第一物品的相关性分值乘Y;
根据所述A、B、C、D、X、Y的一项或多项,计算得到第二物品与第一物品的初始相关性分值。
优选的,所述根据所述相关性分值,向点击第一物品的用户推送第二物品,包括:
根据所述相关性分值对物品库的第二物品进行排序;
选取排序后预设个数的第二物品组成物品推荐列表;
当用户点击第一物品时,根据所述物品推荐列表向用户推送相应的第二 物品。
优选的,所述物品为应用,所述物品库为应用库。
第二方面,本发明还提供一种物品推送装置,包括:
第一计算模块:用于获取点击第一物品且曝光第二物品的用户组成的用户群,根据用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值;
第二计算模块:用于将物品库中除第一物品外的其它物品作为第二物品,利用第一计算模块分别计算物品库中第二物品与第一物品的相关性分值;
推送模块:用于根据所述相关性分值,向点击第一物品的用户推送第二物品。
进一步的,所述第一计算模块执行包括:
根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值;
统计所述初始相关性分值的平均值,得到第二物品与第一物品的相关性分值。
进一步的,所述根据用户群中每一个各自用户的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值,包括:
获取用户的点击行为记录,所述点击行为记录包括用户对物品的点击时间、点击场景及点击类型记录;
判断用户对第一物品的点击时间与对第二物品的点击时间的时间间隔,当所述时间间隔小于或等于第一预设值时,第二物品与第一物品的相关性分值增加A,当所述时间间隔小于或等于第二预设值时,第二物品与第一物品的相关性分值增加B,当所述时间间隔小于或等于第三预设值时,第二物品与第一物品的相关性分值增加C;
判断用户对第二物品的点击场景,当所述点击场景为用户点击了第一物品后在第一预设场景下推荐第二物品时,第二物品与第一物品的相关性分值增加D;
判断用户对第二物品的点击类型,当所述点击类型为用户点击了第一物品且点击了第二物品时,第二物品与第一物品的相关性分值乘X,当所述点击类型为用户点击了第一物品且未点击第二物品时,第二物品与第一物品的 相关性分值乘Y;
根据所述A、B、C、D、X、Y的一项或多项,计算得到第二物品与第一物品的初始相关性分值。
进一步的,所述推送模块执行包括:
根据所述相关性分值对物品库的第二物品进行排序;
选取排序后预设个数的第二物品组成物品推荐列表;
当用户点击第一物品时,根据所述物品推荐列表向用户推送相应的第二物品。
第三方面,本发明还提供一种服务器,包括:
一个或多个处理器;
存储器;
一个或多个应用程序,其中所述一个或多个应用程序被存储在所述存储器中并被配置为由所述一个或多个处理器执行,所述一个或多个应用程序配置用于执行如第一方面所述的物品推送方法。
第四方面,本发明还提供了一种计算设备,包括:
处理器;以及
存储器,其上存储有可执行代码,当所述可执行代码被所述处理器执行时,使所述处理器执行如上任何一项所述的物品推送方法。
第五方面,本发明还提供了一种非暂时性机器可读存储介质,其上存储有可执行代码,当所述可执行代码被电子设备的处理器执行时,使所述处理器执行如上任一项所述的物品推送方法。
与现有技术相比,本发明具备如下优点:
本发明提供了一种基于用户点击行为的物品推送算法,将数据统计的样本限定于点击第一物品且曝光第二物品的用户组成的用户群,然后根据用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值,剔除无关数据的影响,提高物品之间相关性分值计算的准确性,然后根据相关性分值为用户推送关联性更高的物品,从而提高物品推送的准确率。
并且,本发明的推荐算法中记录用户点击行为中的点击时间、点击场景和点击类型三种变量,通过三维向量记录每一次的用户行为,然后基于多维 向量相关性规则计算出物品库中各物品之间的相关性分值,更为细致地计算出物品之间的相关联程度,再根据相关性分值向用户推送相应的物品,物品之间的相关性计算更为细致,为用户所推送的物品则更能让用户感兴趣,提高物品推送的准确率,提升用户体验。
显然,上述有关本发明优点的描述是概括性的,更多的优点描述将体现在后续的实施例揭示中,以及,本领域技术人员也可以本发明所揭示的内容合理地发现本发明的其他诸多优点。
本发明附加的方面和优点将在下面的描述中部分给出,这些将从下面的描述中变得明显,或通过本发明的实践了解到。
【附图说明】
图1为本发明物品推送方法的一实施例流程示意图;
图2为本发明物品推送方法的另一实施例流程示意图;
图3为本发明物品推送方法的又一实施例流程示意图;
图4为本发明物品推送装置的一实施例示意图;
图5为本发明服务器的一实施例结构示意图。
图6为本发明一个实施例的计算设备的结构示意图。
【具体实施方式】
下面结合附图和示例性实施例对本发明作进一步地描述,其中附图中相同的标号全部指的是相同的部件。此外,如果已知技术的详细描述对于示出本发明的特征是不必要的,则将其省略。
本技术领域技术人员可以理解,除非特意声明,这里使用的单数形式“一”、“一个”、“所述”和“该”也可包括复数形式。应该进一步理解的是,本发明的说明书中使用的措辞“包括”是指存在所述特征、整数、步骤、操作,但是并不排除存在或添加一个或多个其他特征、整数、步骤、操作。
本技术领域技术人员可以理解,除非另外定义,这里使用的所有术语(包括技术术语和科学术语),具有与本发明所属领域中的普通技术人员的一般理解相同的意义。还应该理解的是,诸如通用字典中定义的那些术语,应该被理解为具有与现有技术的上下文中的意义一致的意义,并且除非像这里一样 被特定定义,否则不会用理想化或过于正式的含义来解释。
本领域技术人员应当理解,本发明所称的“应用”、“应用程序”、“应用软件”以及类似表述的概念,是业内技术人员所公知的相同概念,是指由一系列计算机指令及相关数据资源有机构造的适于电子运行的计算机软件。除非特别指定,这种命名本身不受编程语言种类、级别,也不受其赖以运行的操作***或平台所限制。理所当然地,此类概念也不受任何形式的终端所限制。
在一种实施例中,本发明提供一种物品推送方法,如图1所示,包括:
S100:获取点击第一物品且曝光第二物品的用户组成的用户群,根据用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值。
本发明实施例中,第二物品与第一物品的相关联程度由第二物品与第一物品的相关性分值体现,在要计算第二物品与第一物品的相关性分值时,首先需要确定数据统计的样本,本实施例中,选择已点击了第一物品且在第一物品中曝光了第二物品的用户组成的用户群作为本实施例数据统计的样本,即以该用户群的用户对第一物品、第二物品的点击行为记录作为本实施例数据统计的样本,对于用户群中的任意一个用户,获取每一个用户对第一物品、第二物品的点击行为记录作为数据统计的样本,然后执行后续操作。
所述点击行为记录是在当用户点击了第一物品后、向用户曝光了第二物品时所记录的用户的点击行为操作,每一次点击行为操作均会生成一条相应的点击行为记录,本实施例根据每一个用户的每一条点击行为记录计算出第二物品与第一物品的一次相关性分值,接着统计该用户群的所有用户的点击行为记录计算出第二物品与第一物品的相关性分值。
S200:将物品库中除第一物品外的其它物品作为第二物品,按照S100的计算方法,分别计算物品库中第二物品与第一物品的相关性分值。
S100阐述了计算第二物品与第一物品的相关性分值的方法,在计算了一个第二物品与第一物品的相关性分值后,需要计算物品库中除第一物品外的所有物品与第一物品的相关性分值,此时将物品库中除第一物品外的其它物品作为上述的第二物品,然后按照S100的方法,依次获取点击第一物品且曝光第二物品的用户组成的用户群,再根据用户群中每一用户对第一物品及第 二物品的点击行为记录,依次计算该些第二物品与第一物品的相关性分值,以此得到物品库中除第一物品外的第二物品与第一物品的相关性分值。优选的,本实施例的物品库中包含了第一物品,可以理解的是,本领域技术人员可以根据本实施例的方案将第一物品存储于物品库之外,同样可以计算物品库中的物品与第一物品的相关性分值,因此第一物品与物品库的存储关系不应该视为对本发明方案的限定。例如,物品库中有N1、N2、N3、N4、N5五个物品,以N1为第一物品,分别计算第二物品N2、N3、N4、N5与第一物品N1的相关性分值,当后续用户点击N1时,根据相关性分值向用户推送第二物品N2、N3、N4、N5中的一个或多个;进一步的,以N2为第一物品,分别计算第二物品N1、N3、N4、N5与第一物品N2的相关性分值,当后续用户点击N2时,根据相关性分值向用户推送第二物品N1、N3、N4、N5中的一个或多个;以此类推,根据该方法计算得出物品库中各物品之间的相关性分值。
S300:根据所述相关性分值,向点击第一物品的用户推送第二物品。
在计算得到物品库中除第一物品外的第二物品与第一物品的相关性分值后,根据不同物品与第一物品的相关性分值的大小即可确定物品库中哪个物品与第一物品的关联性更高,在后续当用户点击第一物品时,根据所述相关性分值向用户推送与第一物品关联性更高的第二物品。
本实施例的应用场景为应用商店,在应用商店中,在用户点击了某一应用(第一应用)后,应用商店会同时曝光与该应用存在关联关系的其它应用(第二应用),本实施例将点击了第一应用并曝光出现了第二应用的场景定义为点击了第一物品且曝光了第二物品,并记录用户在该场景下的点击行为,然后以该些点击了第一物品曝光了第二物品的用户组成的用户群作为数据统计的样本,根据用户群中每一用户对第一物品、第二物品的每一条点击行为记录便可计算出第二物品与第一物品的一次相关性分值,然后统计用户群中的所有用户对第二物品的每一条点击行为记录计算出第二物品与第一物品的相关性分值;依据该方法,将物品库(即应用商店上的应用库)中除第一物品外的其它物品作为第二物品,计算物品库中其它物品与第一物品的相关性分值,以此得到物品库中所有第二物品与第一物品的相关性分值,即得到应用商店中应用之间的关联程度,进而当用户点击第一应用时,根据相关性分值向用户推送第二应用。
本发明的一种实施例,如图2所示,所述S100中根据用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值,包括:
S101:根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值;
S102:统计所述初始相关性分值的平均值,得到第二物品与第一物品的相关性分值。
本实施例中,一个用户的点击行为记录计算得到第二物品与第一物品的相关性分值定义为初始相关性分值,然后再统计用户群中的所有用户的点击行为记录所得到的初始相关性分值的平均值,作为第二物品与第一物品的相关性分值,具体的,第二物品与第一物品的相关性分值计算如下:
Figure PCTCN2018105841-appb-000001
其中,u表示用户群中的每一个用户,U表示整个用户群,Score(Item1,Item2)为一个用户的点击行为记录计算得到第二物品与第一物品的初始相关性分值,然后再对算出的初始相关性分值计算平均值,以此计算出第二物品与第一物品的相关性分值Sim(Item1,Item2)。
进一步的,上述实施例中,所述根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值,包括:
获取用户的点击行为记录,所述点击行为记录包括用户对物品的点击时间、点击场景及点击类型记录;
判断用户对第一物品的点击时间与对第二物品的点击时间的时间间隔,当所述时间间隔小于或等于第一预设值时,第二物品与第一物品的相关性分值增加A,当所述时间间隔小于或等于第二预设值时,第二物品与第一物品的相关性分值增加B,当所述时间间隔小于或等于第三预设值时,第二物品与第一物品的相关性分值增加C;
判断用户对第二物品的点击场景,当所述点击场景为用户点击了第一物品后在第一预设场景下推荐第二物品时,第二物品与第一物品的相关性分值 增加D;
判断用户对第二物品的点击类型,当所述点击类型为用户点击了第一物品且点击了第二物品时,第二物品与第一物品的相关性分值乘X,当所述点击类型为用户点击了第一物品且未点击第二物品时,第二物品与第一物品的相关性分值乘Y;
根据所述A、B、C、D、X、Y的一项或多项,计算得到第二物品与第一物品的初始相关性分值。
具体的,本实施例中,所述点击行为记录包括用户对物品的点击时间、点击场景及点击类型记录,所述点击时间记录了用户点击了该物品的时刻,所述点击场景记录了用户点击了第一物品后在何种场景下曝光了第二物品,所述点击类型记录了用户对已曝光的第二物品的点击行为类型,本实施例中,每一条点击行为记录为:
User,Item,Vec=(SourceItem,Time,ActionType)
其中User为用户ID;Item为物品ID;Vec为用户对物品的点击行为向量,其中SourceItem为用户对该物品的点击行为场景;Time表示用户对物品的点击行为的时间点;ActionType表示用户对物品的点击行为类型。
同一个用户对第一物品、第二物品的点击行为记录如下:
User1,Item1,Vec1=(SourceItem1,Time1,ActionType1)
User1,Item2,Vec2=(SourceItem2,Time2,ActionType2)。
第二物品与第一物品的初始相关性分值记为Score(Item1,Item2),并且计算规则如下:
首先:初始化Score(Item1,Item2),即为Score(Item1,Item2)赋予一个初始值,一般的初始值为0,然后根据两条点击行为记录中“Time”值判断用户对第一物品的点击时间与对第二物品的点击时间的时间间隔,该时间间隔为用户对第一物品的点击时间与对第二物品的点击时间的最小时间间隔,并为不同的时间间隔赋予不同的相关性分值,当所述时间间隔小于或等于第一预设值时,第二物品与第一物品的相关性分值增加A,当所述时间间隔小于或等于第二预设值时,第二物品与第一物品的相关性分值增加B,当所述时间间隔小于或等于第三预设值时,第二物品与第一物品的相关性分值增加C;例如Time1与Time2在同一小时内,则定义Score(Item1,Item2)=3,即 在相关性初始值上增加3;Time1与Time2相差0-3天,则Score(Item1,Item2)=1;Time1与Time2相差4-7天,则Score(Item1,Item2)=0.5;Time1与Time2相差7-15天,则Score(Item1,Item2)=0.2。
进一步的,判断用户对第二物品的点击场景,为不同的点击场景赋予不同的分值,例如,SourceItem2=1表示Item2是点击了Item1后,“其他用户还下载”的场景下推荐的,对于该种点击场景,第二物品与第一物品的相关性分值增加:Score(Item1,Item2)=Score(Item1,Item2)+3;SourceItem2=2表示Item2是点击了Item1后,“爆款应用”的场景下推荐的,对于该种点击场景,第二物品与第一物品的相关性分值不变:Score(Item1,Item2)=Score(Item1,Item2)+0。
再者,判断用户对第二物品的点击类型,为不同的点击类型赋予不同的分值,例如,ActionType2=1表示在Item1下给用户推荐了Item2,用户点击了Item2,此时记Score(Item1,Item2)=1*Score(Item1,Item2);ActionType2=0表示在Item1下给用户推荐了Item2,用户没有点击Item2,,此时记Score(Item1,Item2)=0*Score(Item1,Item2),该种情况表示用户对item2不感兴趣。由此,根据每一个用户各自的点击行为记录分别计算出第二物品与第一物品的初始相关性分值,后续再对算出的初始相关性分值计算平均值,以此计算出第二物品与第一物品的相关性分值。
本发明的一种实施例中,如图3所示,所述S300包括:
S301:根据所述相关性分值对物品库的第二物品进行排序;
S302:选取排序后预设个数的第二物品组成物品推荐列表;
S303:当用户点击第一物品时,根据所述物品推荐列表向用户推送相应的第二物品。
在计算得到物品库中物品之间的相关性分值后,每一个第一物品都有对应的第二物品,并且按照相关性分值的大小对物品库中的第二物品进行排序,在排序后选取预设个数的第二物品组成物品推荐列表,优选的,推荐列表中的物品按照相关性分值从大到小进行排序,相关性分值越大,说明第二物品与第一物品的相关联程度越高,当后续用户点击第一物品时,根据所述物品推荐列表向用户推送相应的第二物品。例如,第一物品N1,根据与N1的相 关性分值排序后得到N3、N5、N2、N4,选取3个第二物品N3、N5、N2组成N1的物品推荐列表,当用户点击N1时,向用户推送N3、N5、N2。
在另一种实施例中,本发明提供一种物品信息推送装置,如图4所示,包括:
第一计算模块100:用于获取点击第一物品且曝光第二物品的用户组成的用户群,根据用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值;
第二计算模块200:用于将物品库中除第一物品外的其它物品作为第二物品,利用第一计算模块分别计算物品库中第二物品与第一物品的相关性分值;
推送模块300:用于根据所述相关性分值,向点击第一物品的用户推送第二物品。
本发明实施例中,第二物品与第一物品的相关联程度由第二物品与第一物品的相关性分值体现,在要计算第二物品与第一物品的相关性分值时,首先需要确定数据统计的样本,本实施例中,选择已点击了第一物品且在第一物品中曝光了第二物品的用户组成的用户群作为本实施例数据统计的样本,即以该用户群的用户对第一物品、第二物品的点击行为记录作为本实施例数据统计的样本,对于用户群中的任意一个用户,第一计算模块100获取每一个用户对第一物品、第二物品的点击行为记录作为数据统计的样本,所述点击行为记录是在当用户点击了第一物品后、向用户曝光了第二物品时所记录的用户的点击行为操作,每一次点击行为操作均会生成一条相应的点击行为记录,本实施例中,第一计算模块100根据每一个用户的每一条点击行为记录计算出第二物品与第一物品的一次相关性分值,接着统计该用户群的所有用户的点击行为记录计算出第二物品与第一物品的相关性分值。
本实施例中,第一计算模块100计算第二物品与第一物品的相关性分值的方法,在计算了一个第二物品与第一物品的相关性分值后,需要计算物品库中除第一物品外的所有物品与第一物品的相关性分值,此时第二计算模块200将物品库中除第一物品外的其它物品作为上述的第二物品,然后依次获取点击第一物品且曝光第二物品的用户组成的用户群,再根据用户群中每一 用户对第一物品及第二物品的点击行为记录,依次计算该些第二物品与第一物品的相关性分值,以此得到物品库中除第一物品外的第二物品与第一物品的相关性分值。
在第二计算模块200计算得到物品库中除第一物品外的第二物品与第一物品的相关性分值后,推送模块300根据不同物品与第一物品的相关性分值的大小即可确定物品库中哪个物品与第一物品的关联性更高,在后续当用户点击第一物品时,根据所述相关性分值向用户推送与第一物品关联性更高的第二物品。
本发明的一种实施例中,所述第一计算模块100执行包括:
根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值;
统计所述初始相关性分值的平均值,得到第二物品与第一物品的相关性分值。
本实施例中,一个用户的点击行为记录计算得到第二物品与第一物品的相关性分值定义为初始相关性分值,然后再统计用户群中的所有用户的点击行为记录所得到的初始相关性分值的平均值,作为第二物品与第一物品的相关性分值,具体的,第二物品与第一物品的相关性分值计算如下:
Figure PCTCN2018105841-appb-000002
其中,u表示用户群中的每一个用户,U表示整个用户群,Score(Item1,Item2)为一个用户的点击行为记录计算得到第二物品与第一物品的初始相关性分值,然后再对算出的初始相关性分值计算平均值,以此计算出第二物品与第一物品的相关性分值Sim(Item1,Item2)。
进一步的,上述实施例中,所述根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值,包括:
获取用户的点击行为记录,所述点击行为记录包括用户对物品的点击时间、点击场景及点击类型记录;
判断用户对第一物品的点击时间与对第二物品的点击时间的时间间隔, 当所述时间间隔小于或等于第一预设值时,第二物品与第一物品的相关性分值增加A,当所述时间间隔小于或等于第二预设值时,第二物品与第一物品的相关性分值增加B,当所述时间间隔小于或等于第三预设值时,第二物品与第一物品的相关性分值增加C;
判断用户对第二物品的点击场景,当所述点击场景为用户点击了第一物品后在第一预设场景下推荐第二物品时,第二物品与第一物品的相关性分值增加D;
判断用户对第二物品的点击类型,当所述点击类型为用户点击了第一物品且点击了第二物品时,第二物品与第一物品的相关性分值乘X,当所述点击类型为用户点击了第一物品且未点击第二物品时,第二物品与第一物品的相关性分值乘Y;
根据所述A、B、C、D、X、Y的一项或多项,计算得到第二物品与第一物品的初始相关性分值。
具体的,本实施例中,所述点击行为记录包括用户对物品的点击时间、点击场景及点击类型记录,所述点击时间记录了用户点击了该物品的时刻,所述点击场景记录了用户点击了第一物品后在何种场景下曝光了第二物品,所述点击类型记录了用户对已曝光的第二物品的点击行为类型,本实施例中,每一条点击行为记录为:
User,Item,Vec=(SourceItem,Time,ActionType)
其中User为用户ID;Item为物品ID;Vec为用户对物品的点击行为向量,其中SourceItem为用户对该物品的点击行为场景;Time表示用户对物品的点击行为的时间点;ActionType表示用户对物品的点击行为类型。
同一个用户对第一物品、第二物品的点击行为记录如下:
User1,Item1,Vec1=(SourceItem1,Time1,ActionType1)
User1,Item2,Vec2=(SourceItem2,Time2,ActionType2)。
第二物品与第一物品的初始相关性分值记为Score(Item1,Item2),并且计算规则如下:
首先:初始化Score(Item1,Item2),即为Score(Item1,Item2)赋予一个初始值,一般的初始值为0,然后根据两条点击行为记录中“Time”值判断 用户对第一物品的点击时间与对第二物品的点击时间的时间间隔,该时间间隔为用户对第一物品的点击时间与对第二物品的点击时间的最小时间间隔,并为不同的时间间隔赋予不同的相关性分值,当所述时间间隔小于或等于第一预设值时,第二物品与第一物品的相关性分值增加A,当所述时间间隔小于或等于第二预设值时,第二物品与第一物品的相关性分值增加B,当所述时间间隔小于或等于第三预设值时,第二物品与第一物品的相关性分值增加C;例如Time1与Time2在同一小时内,则定义Score(Item1,Item2)=3,即在相关性初始值上增加3;Time1与Time2相差0-3天,则Score(Item1,Item2)=1;Time1与Time2相差4-7天,则Score(Item1,Item2)=0.5;Time1与Time2相差7-15天,则Score(Item1,Item2)=0.2。
进一步的,判断用户对第二物品的点击场景,为不同的点击场景赋予不同的分值,例如,SourceItem2=1表示Item2是点击了Item1后,“其他用户还下载”的场景下推荐的,对于该种点击场景,第二物品与第一物品的相关性分值增加:Score(Item1,Item2)=Score(Item1,Item2)+3;SourceItem2=2表示Item2是点击了Item1后,“爆款应用”的场景下推荐的,对于该种点击场景,第二物品与第一物品的相关性分值不变:Score(Item1,Item2)=Score(Item1,Item2)+0。
再者,判断用户对第二物品的点击类型,为不同的点击类型赋予不同的分值,例如,ActionType2=1表示在Item1下给用户推荐了Item2,用户点击了Item2,此时记Score(Item1,Item2)=1*Score(Item1,Item2);ActionType2=0表示在Item1下给用户推荐了Item2,用户没有点击Item2,,此时记Score(Item1,Item2)=0*Score(Item1,Item2),该种情况表示用户对item2不感兴趣。由此,根据每一个用户各自的点击行为记录分别计算出第二物品与第一物品的初始相关性分值,后续再对算出的初始相关性分值计算平均值,以此计算出第二物品与第一物品的相关性分值。
本发明的一种实施例中,所述推送模块300执行包括:
根据所述相关性分值对物品库的第二物品进行排序;
选取排序后预设个数的第二物品组成物品推荐列表;
当用户点击第一物品时,根据所述物品推荐列表向用户推送相应的第二 物品。
在计算得到物品库中物品之间的相关性分值后,每一个第一物品都有对应的第二物品,并且按照相关性分值的大小对物品库中的第二物品进行排序,在排序后选取预设个数的第二物品组成物品推荐列表,优选的,推荐列表中的物品按照相关性分值从大到小进行排序,相关性分值越大,说明第二物品与第一物品的相关联程度越高,当后续用户点击第一物品时,根据所述物品推荐列表向用户推送相应的第二物品。
在一种实施例中,本发明还提供一种服务器,包括:一个或多个处理器、存储器及一个或多个应用程序,其中所述一个或多个应用程序被存储在所述存储器中并被配置为由所述一个或多个处理器执行,所述一个或多个应用程序配置用于执行上述实施例的物品推送方法。
图5为本发明服务器的结构示意图,包括处理器503、存储器505、输入单元507以及显示单元509等器件。本领域技术人员可以理解,图5示出的结构器件并不构成对所有服务器的限定,可以包括比图示更多或更少的部件,或者组合某些部件。存储器505可用于存储应用程序501以及各功能模块,处理器503运行存储在存储器505的应用程序501,从而执行设备的各种功能应用以及数据处理。存储器505可以是内存储器或外存储器,或者包括内存储器和外存储器两者。内存储器可以包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦写可编程ROM(EEPROM)、快闪存储器、或者随机存储器。外存储器可以包括硬盘、软盘、ZIP盘、U盘、磁带等。本发明所公开的存储器包括但不限于这些类型的存储器。本发明所公开的存储器505只作为例子而非作为限定。
输入单元507用于接收信号的输入,以及接收用户输入的关键字。输入单元507可包括触控面板以及其它输入设备。触控面板可收集用户在其上或附近的触摸操作(比如用户使用手指、触笔等任何适合的物体或附件在触控面板上或在触控面板附近的操作),并根据预先设定的程序驱动相应的连接装置;其它输入设备可以包括但不限于物理键盘、功能键(比如播放控制按键、开关按键等)、轨迹球、鼠标、操作杆等中的一种或多种。显示单元509可用于显示用户输入的信息或提供给用户的信息以及计算机设备的各种菜单。显 示单元509可采用液晶显示器、有机发光二极管等形式。处理器503是计算机设备的控制中心,利用各种接口和线路连接整个电脑的各个部分,通过运行或执行存储在存储器503内的软件程序和/或模块,以及调用存储在存储器内的数据,执行各种功能和处理数据。
在一实施方式中,服务器包括一个或多个处理器503,以及一个或多个存储器505,一个或多个应用程序501,其中所述一个或多个应用程序501被存储在存储器505中并被配置为由所述一个或多个处理器503执行,所述一个或多个应用程序301配置用于执行以上实施例所述的物品推送方法。
图6示出了根据本发明一实施例可用于实现上述物品推送方法的计算设备的结构示意图。
参见图6,计算设备600包括存储器610和处理器620。
处理器620可以是一个多核的处理器,也可以包含多个处理器。在一些实施例中,处理器620可以包含一个通用的主处理器以及一个或多个特殊的协处理器,例如图形处理器(GPU)、数字信号处理器(DSP)等等。在一些实施例中,处理器620可以使用定制的电路实现,例如特定用途集成电路(ASIC,Application Specific Integrated Circuit)或者现场可编程逻辑门阵列(FPGA,Field Programmable Gate Arrays)。
存储器610可以包括各种类型的存储单元,例如***内存、只读存储器(ROM),和永久存储装置。其中,ROM可以存储处理器620或者计算机的其他模块需要的静态数据或者指令。永久存储装置可以是可读写的存储装置。永久存储装置可以是即使计算机断电后也不会失去存储的指令和数据的非易失性存储设备。在一些实施方式中,永久性存储装置采用大容量存储装置(例如磁或光盘、闪存)作为永久存储装置。另外一些实施方式中,永久性存储装置可以是可移除的存储设备(例如软盘、光驱)。***内存可以是可读写存储设备或者易失性可读写存储设备,例如动态随机访问内存。***内存可以存储一些或者所有处理器在运行时需要的指令和数据。此外,存储器610可以包括任意计算机可读存储媒介的组合,包括各种类型的半导体存储芯片(DRAM,SRAM,SDRAM,闪存,可编程只读存储器),磁盘和/或光盘也可以采用。在一些实施方式中,存储器610可以包括可读和/或写的可移除的存储设备,例如激光唱片(CD)、只读数字多功能光盘(例如DVD-ROM,双层DVD-ROM)、 只读蓝光光盘、超密度光盘、闪存卡(例如SD卡、min SD卡、Micro-SD卡等等)、磁性软盘等等。计算机可读存储媒介不包含载波和通过无线或有线传输的瞬间电子信号。
存储器610上存储有可执行代码,当可执行代码被处理器620处理时,可以使处理器620执行上文述及的物品推送方法。
此外,根据本发明的方法还可以实现为一种计算机程序或计算机程序产品,该计算机程序或计算机程序产品包括用于执行本发明的上述方法中限定的上述各步骤的计算机程序代码指令。
或者,本发明还可以实施为一种非暂时性机器可读存储介质(或计算机可读存储介质、或机器可读存储介质),其上存储有可执行代码(或计算机程序、或计算机指令代码),当所述可执行代码(或计算机程序、或计算机指令代码)被电子设备(或计算设备、服务器等)的处理器执行时,使所述处理器执行根据本发明的上述方法的各个步骤。
本领域技术人员还将明白的是,结合这里的公开所描述的各种示例性逻辑块、模块、电路和算法步骤可以被实现为电子硬件、计算机软件或两者的组合。
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实施例的全部或部分步骤可以通过硬件来完成,也可以通过程序来指令相关的硬件完成,该程序可以存储于一计算机可读存储介质中,存储介质可以包括存储器、磁盘或光盘等。
以上所述仅是本发明的部分实施方式,应当指出,对于本技术领域的普通技术人员来说,在不脱离本发明原理的前提下,还可以做出若干改进和润饰,这些改进和润饰也应视为本发明的保护范围。

Claims (12)

  1. 一种物品推送方法,其特征在于,包括:
    将物品库中除第一物品外的其它物品作为第二物品,获取点击第一物品且曝光第二物品的用户组成的用户群;
    根据所述用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值;
    根据所述相关性分值,向点击第一物品的用户推送第二物品。
  2. 根据权利要求1所述的方法,其特征在于,所述根据用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值,包括:
    根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值;
    统计所述初始相关性分值的平均值,得到第二物品与第一物品的相关性分值。
  3. 根据权利要求2所述的方法,其特征在于,所述根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值,包括:
    获取用户的点击行为记录,所述点击行为记录包括用户对物品的点击时间、点击场景及点击类型记录;
    判断用户对第一物品的点击时间与对第二物品的点击时间的时间间隔,当所述时间间隔小于或等于第一预设值时,第二物品与第一物品的相关性分值增加A,当所述时间间隔小于或等于第二预设值时,第二物品与第一物品的相关性分值增加B,当所述时间间隔小于或等于第三预设值时,第二物品与第一物品的相关性分值增加C;
    判断用户对第二物品的点击场景,当所述点击场景为用户点击第一物品后在第一预设场景下推荐第二物品时,第二物品与第一物品的相关性分值增加D;
    判断用户对第二物品的点击类型,当所述点击类型为用户点击第一物品且点击第二物品时,第二物品与第一物品的相关性分值乘X,当所述点击类 型为用户点击第一物品且未点击第二物品时,第二物品与第一物品的相关性分值乘Y;
    根据所述A、B、C、D、X、Y的一项或多项,计算得到第二物品与第一物品的初始相关性分值。
  4. 根据权利要求1所述的方法,其特征在于,所述根据所述相关性分值,向点击第一物品的用户推送第二物品,包括:
    根据所述相关性分值对物品库的第二物品进行排序;
    选取排序后预设个数的第二物品组成物品推荐列表;
    当用户点击第一物品时,根据所述物品推荐列表向用户推送相应的第二物品。
  5. 根据权利要求1所述的方法,其特征在于,所述物品为应用,所述物品库为应用库。
  6. 一种物品推送装置,其特征在于,包括:
    第一计算模块:用于获取点击第一物品且曝光第二物品的用户组成的用户群,根据用户群中每一用户对第一物品及第二物品的点击行为记录,计算第二物品与第一物品的相关性分值;
    第二计算模块:用于将物品库中除第一物品外的其它物品作为第二物品,利用第一计算模块分别计算物品库中第二物品与第一物品的相关性分值;
    推送模块:用于根据所述相关性分值,向点击第一物品的用户推送第二物品。
  7. 根据权利要求6所述的装置,其特征在于,所述第一计算模块执行包括:
    根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值;
    统计所述初始相关性分值的平均值,得到第二物品与第一物品的相关性分值。
  8. 根据权利要求7所述的装置,其特征在于,所述根据用户群中每一个用户各自的点击行为记录,分别计算得到第二物品与第一物品的初始相关性分值,包括:
    获取用户的点击行为记录,所述点击行为记录包括用户对物品的点击时 间、点击场景及点击类型记录;
    判断用户对第一物品的点击时间与对第二物品的点击时间的时间间隔,当所述时间间隔小于或等于第一预设值时,第二物品与第一物品的相关性分值增加A,当所述时间间隔小于或等于第二预设值时,第二物品与第一物品的相关性分值增加B,当所述时间间隔小于或等于第三预设值时,第二物品与第一物品的相关性分值增加C;
    判断用户对第二物品的点击场景,当所述点击场景为用户点击了第一物品后在第一预设场景下推荐第二物品时,第二物品与第一物品的相关性分值增加D;
    判断用户对第二物品的点击类型,当所述点击类型为用户点击了第一物品且点击了第二物品时,第二物品与第一物品的相关性分值乘X,当所述点击类型为用户点击了第一物品且未点击第二物品时,第二物品与第一物品的相关性分值乘Y;
    根据所述A、B、C、D、X、Y的一项或多项,计算得到第二物品与第一物品的初始相关性分值。
  9. 根据权利要求6所述的装置,其特征在于,所述推送模块执行包括:
    根据所述相关性分值对物品库的第二物品进行排序;
    选取排序后预设个数的第二物品组成物品推荐列表;
    当用户点击第一物品时,根据所述物品推荐列表向用户推送相应的第二物品。
  10. 一种服务器,其特征在于,包括:
    一个或多个处理器;
    存储器;
    一个或多个应用程序,其中所述一个或多个应用程序被存储在所述存储器中并被配置为由所述一个或多个处理器执行,所述一个或多个应用程序配置用于执行如权利要求1-5所述的物品推送方法。
  11. 一种计算设备,包括:
    处理器;以及
    存储器,其上存储有可执行代码,当所述可执行代码被所述处理器执行时,使所述处理器执行如权利要求1-5中任何一项所述的物品推送方法。
  12. 一种非暂时性机器可读存储介质,其上存储有可执行代码,当所述可执行代码被电子设备的处理器执行时,使所述处理器执行如权利要求1至5中任一项所述的物品推送方法。
PCT/CN2018/105841 2017-12-29 2018-09-14 物品推送方法、装置及服务器、计算设备及存储介质 WO2019128317A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201711484128.X 2017-12-29
CN201711484128.XA CN108109052A (zh) 2017-12-29 2017-12-29 物品推送方法、装置及服务器

Publications (1)

Publication Number Publication Date
WO2019128317A1 true WO2019128317A1 (zh) 2019-07-04

Family

ID=62215024

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/105841 WO2019128317A1 (zh) 2017-12-29 2018-09-14 物品推送方法、装置及服务器、计算设备及存储介质

Country Status (2)

Country Link
CN (1) CN108109052A (zh)
WO (1) WO2019128317A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108109052A (zh) * 2017-12-29 2018-06-01 广州优视网络科技有限公司 物品推送方法、装置及服务器
CN112381615B (zh) * 2020-11-27 2022-09-02 华中科技大学 基于用户重复行为模式挖掘的短序列推荐方法
CN113516504A (zh) * 2021-05-20 2021-10-19 深圳马六甲网络科技有限公司 一种商品推荐方法、装置、设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104935981A (zh) * 2015-06-17 2015-09-23 Tcl集团股份有限公司 一种广告信息的推送方法及***
CN105719152A (zh) * 2014-12-05 2016-06-29 深圳富泰宏精密工业有限公司 广告推送***及方法
CN105761101A (zh) * 2016-02-04 2016-07-13 云南今日游情科技有限公司 基于相互选择的移动终端高精准度广告推送***及方法
CN108109052A (zh) * 2017-12-29 2018-06-01 广州优视网络科技有限公司 物品推送方法、装置及服务器

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101887390A (zh) * 2010-06-23 2010-11-17 宇龙计算机通信科技(深圳)有限公司 一种应用软件评级方法及装置
CN103034508B (zh) * 2011-10-10 2015-08-19 腾讯科技(深圳)有限公司 软件推荐方法和***
CN104391999B (zh) * 2014-12-15 2018-02-02 北京国双科技有限公司 信息推荐方法和装置
CN106651542B (zh) * 2016-12-31 2021-06-25 珠海市魅族科技有限公司 一种物品推荐的方法及装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105719152A (zh) * 2014-12-05 2016-06-29 深圳富泰宏精密工业有限公司 广告推送***及方法
CN104935981A (zh) * 2015-06-17 2015-09-23 Tcl集团股份有限公司 一种广告信息的推送方法及***
CN105761101A (zh) * 2016-02-04 2016-07-13 云南今日游情科技有限公司 基于相互选择的移动终端高精准度广告推送***及方法
CN108109052A (zh) * 2017-12-29 2018-06-01 广州优视网络科技有限公司 物品推送方法、装置及服务器

Also Published As

Publication number Publication date
CN108109052A (zh) 2018-06-01

Similar Documents

Publication Publication Date Title
TW201931256A (zh) 營銷資訊的推送方法及裝置
CN107613022A (zh) 内容推送方法、装置及计算机设备
WO2019128317A1 (zh) 物品推送方法、装置及服务器、计算设备及存储介质
AU2020378006B2 (en) Page simulation system
US20160373312A1 (en) Platform application visual analytics system
KR102161137B1 (ko) 게임 데이터 수집을 위한 방법 및 시스템
Layton Learning data mining with python
US20160232548A1 (en) Adaptive pricing analytics
WO2018060783A1 (en) Objective based advertisement placement platform
US10579685B2 (en) Content event insights
CN102314491A (zh) 多核环境下基于海量日志的类似行为模式用户识别方法
Li et al. Measuring scale-up and scale-out Hadoop with remote and local file systems and selecting the best platform
JP6568605B2 (ja) 大規模ソースコードリポジトリにおける自動インポートおよびディペンデンシー
JP2018508865A (ja) アプリケーションイベントの追跡
CN108549674B (zh) 一种推荐方法、装置及存储介质
CN113312554B (zh) 用于评价推荐***的方法及装置、电子设备和介质
US10817519B2 (en) Automatic conversion stage discovery
US20150170067A1 (en) Determining analysis recommendations based on data analysis context
US20150269177A1 (en) Method and system for determining user interest in a file
CN109657153A (zh) 一种用于确定用户的关联财经信息的方法与设备
TWI776102B (zh) 提供躥升音源排行榜的方法及系統
JP2007328699A (ja) 類似キャラクタデータ検索方法、サーバおよびプログラム
US11811862B1 (en) System and method for management of workload distribution
KR102651973B1 (ko) 다변수 부동산 세금 시뮬레이션 방법 및 시스템
US20240005235A1 (en) Method and system for dynamically recommending commands for performing a product data management operation

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: 18894710

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: 18894710

Country of ref document: EP

Kind code of ref document: A1