Object information determination method, apparatus, device, storage medium, and product

Abstract

The application discloses an object information determination method, device, equipment, storage medium and product, and belongs to the computer technical field.The method comprises the steps of obtaining attribute allocation information corresponding to a target object; determining a multimedia object corresponding to the target object based on the attribute allocation information; obtaining attribute feature information corresponding to the multimedia object; and updating the attribute allocation information according to the attribute feature information to obtain updated attribute allocation information.The technical scheme provided in the application embodiment can determine the associated multimedia object through the attribute allocation information capable of representing the attribute identification allocation situation of the target object, can update the attribute allocation information according to the attribute feature information corresponding to the multimedia object, and the updated attribute allocation information is more in line with the real attributes of the target object, thereby improving the accuracy and authenticity of the attribute allocation information, and improving the information quantity and accuracy of the feature information of the target object without using personalized features.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computers, and in particular relates to an object information determination method and device, equipment, a storage medium and a product. BACKGROUND

[0002] With the rapid development of digital technology and the Internet, various forms of information content services are welcomed by the public, the number of multimedia objects in the Internet has increased rapidly, and recalling multimedia objects has become particularly important in information content services.

[0003] In the related art, when a user enables a personalized feature, the system can obtain personalized feature information of the user, and match the personalized feature information obtained with feature information of a multimedia object, so as to obtain a multimedia object matched with the personalized feature information of the user.

[0004] The related art has high dependence on personalized feature information of a user, and in the case where the user disables the personalized feature, the amount of available feature information is small. SUMMARY

[0005] Embodiments of the present application provide an object information determination method, device, equipment, a storage medium and a product, which can improve the amount and accuracy of feature information of a target object without using personalized features of the target object.

[0006] According to an aspect of an embodiment of the present application, an object information determination method is provided, and the method comprises:

[0007] obtaining attribute allocation information corresponding to a target object, the attribute allocation information being used to represent an attribute identifier allocation situation corresponding to the target object;

[0008] determining a multimedia object corresponding to the target object based on the attribute allocation information;

[0009] obtaining attribute feature information corresponding to the multimedia object, the attribute feature information being used to represent an association degree between the multimedia object and at least one attribute identifier;

[0010] updating the attribute allocation information according to the attribute feature information to obtain updated attribute allocation information.

[0011] According to an aspect of an embodiment of the present application, an object information determination device is provided, and the device comprises:

[0012] an attribute information allocation module configured to obtain attribute allocation information corresponding to a target object, the attribute allocation information being used to represent an attribute identifier allocation situation corresponding to the target object;

[0013] determining a multimedia object corresponding to the target object based on the attribute assignment information;

[0014] attribute feature information corresponding to the multimedia object, the attribute feature information being used to represent a degree of association between the multimedia object and at least one attribute identifier;

[0015] an attribute information updating module configured to update the attribute assignment information according to the attribute feature information to obtain updated attribute assignment information.

[0016] According to an aspect of an embodiment of the present application, a computer device is provided, the computer device comprising a processor and a memory, the memory storing at least one instruction, at least one program, a code set or an instruction set, the at least one instruction, the at least one program, the code set or the instruction set being loaded and executed by the processor to implement the above object information determination method.

[0017] According to an aspect of an embodiment of the present application, a computer readable storage medium is provided, the storage medium storing at least one instruction, at least one program, a code set or an instruction set, the at least one instruction, the at least one program, the code set or the instruction set being loaded and executed by a processor to implement the above object information determination method.

[0018] According to an aspect of an embodiment of the present application, a computer program product is provided, the computer program product comprising computer instructions stored in a computer readable storage medium. A processor of a computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device performs to implement the above object information determination method.

[0019] The technical scheme provided by the embodiment of the present application can bring the following beneficial effects:

[0020] Through the attribute assignment information capable of representing the attribute identifier assignment situation of the target object, the multimedia object associated with the attribute assignment information can be determined, and the attribute assignment information can be updated according to the attribute feature information corresponding to the multimedia object. The updated attribute assignment information is more consistent with the real attribute of the target object, and the accuracy and authenticity of the attribute assignment information are improved. Without using the personalized features of the target object, the information amount and the accuracy of the feature information of the target object can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort based on these drawings.

[0022] Figure 1 is a schematic diagram of an application running environment provided by an embodiment of the present application;

[0023] Figure 2 is a flow of an object information determination method provided by an embodiment of the present application Figure 1 ;

[0024] Figure 3 is a flow of an object information determination method provided by an embodiment of the present application Figure 2 ;

[0025] Figure 4 is a flow of an object information determination method provided by an embodiment of the present application Figure 3 ;

[0026] Figure 4 is a flow of an object information determination method provided by an embodiment of the present application Figure 6 ;

[0027] Figure 5 is a flow of an object information determination method provided by an embodiment of the present application Figure 7 ;

[0028] Figure 8 is a schematic diagram of an approximate nearest neighbor search flow;

[0029] Figure 9 is a flow diagram of a directional model construction;

[0030] Figure 10 is a schematic diagram of a label recall flow;

[0031] Figure 11 is a schematic diagram of a recall quantity data display page;

[0032] Figure 12 is a schematic diagram of a multimedia object recall branch;

[0033] Figure 13 is a block diagram of an object information determination apparatus provided by an embodiment of the present application;

[0034] Figure 1 is a structural block diagram of a computer device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0035] For the purpose, technical solutions and advantages of the present application to be clearer, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0036] Reference is made to Figure 2 which shows a schematic diagram of an application running environment provided by an embodiment of the present application. The application running environment can include a terminal 10 and a server 20.

[0037] The terminal 10 includes but is not limited to a mobile phone, a computer, a smart voice interactive device, a smart home appliance, a vehicle-mounted terminal, a game console, an e-book reader, a multimedia playing device, a wearable device, a flying device, etc. An application client can be installed in the terminal 10. The embodiments of the present application can be applied to various scenarios, including but not limited to cloud technology, artificial intelligence, intelligent transportation, and assisted driving, etc.

[0038] In the embodiments of the present application, the above-mentioned application can be any application capable of displaying multimedia objects. Typically, the application is an information content service type application. Of course, multimedia objects can also be displayed in other types of applications in addition to information content service type applications. For example, video type applications, news type applications, social type applications, interactive entertainment type applications, browser applications, shopping type applications, content sharing type applications, virtual reality (VR) type applications, augmented reality (AR) type applications, etc. The embodiments of the present application do not limit this. In addition, the multimedia objects displayed by different applications will also be different, and the corresponding functions will also be different, which can be pre-configured according to actual needs, and the embodiments of the present application do not limit this. Optionally, the terminal 10 runs the client of the above-mentioned application. Optionally, the above-mentioned multimedia objects include multimedia advertisements.

[0039] The server 20 is used to provide background services for the client of the application in the terminal 10. For example, the server 20 can be a background server of the above-mentioned application. The server 20 can be a standalone physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms, etc. basic cloud computing services. Optionally, the server 20 simultaneously provides background services for the applications in multiple terminals 10.

[0040] Optionally, the terminal 10 and the server 20 can communicate with each other through the network 30. The terminal 10 and the server 20 can be connected directly or indirectly through wired or wireless communication, which is not limited in the present application.

[0041] Before introducing the method embodiments provided in the present application, the application scenarios, related terms or names possibly involved in the method embodiments of the present application are briefly introduced, so as to facilitate the understanding of the present application by those skilled in the art.

[0042] Smart targeting: In the field of advertising, targeting refers to which people a certain advertisement can be delivered to. Naturally, if these targeted groups are handled well, the effect of the advertisement delivery will be good; otherwise, it will bring bad advertisement delivery experience to the advertiser. Since this targeting is so important, the advertiser and the platform party naturally attach great importance to it. The advertiser hopes that the advertisement it delivers can obtain very good targeted groups, but due to the limitation of ability, the advertiser cannot accurately specify what targeting it wants to deliver. If the advertisement system is very dependent on the targeting set by the advertiser, on the one hand, it is not friendly to the advertiser, and on the other hand, it is also difficult to achieve good effect. Therefore, the advertisement platform needs to achieve: neither relying on the accurate targeting selected by the advertiser when delivering the advertisement (only a wide targeting needs to be set, such as boys; or even the advertiser does not need to set the targeting at all), nor achieving good advertisement delivery effect. In order to achieve this purpose, the advertisement platform needs to find suitable targeted groups for each advertisement. The process of the advertisement platform finding suitable targeted groups for each advertisement is called smart targeting.

[0043] Original targeting: The targeting combination condition set by the advertiser independently in addition to the smart targeting function (including automatic expansion and system optimization).

[0044] Non-breakable targeting: Also known as non-ignorable targeting, it is the targeting condition set by the advertiser independently and must be met no matter what.

[0045] Automatic expansion: On the basis of the manual selection of accurate narrow targeting by the advertiser, the system performs intelligent expansion.

[0046] System optimization: On the basis of the manual selection of wide targeting by the advertiser, the system performs intelligent expansion.

[0047] LookAlike (similar audience targeting): On the basis of a small number of high-quality people groups (old customers) given by the advertiser, the system automatically expands a larger scale of similar people groups (potential new customers).

[0048] Advertisement concentration problem: Also known as the head advertisement problem, mainly embodied in that different users tend to retrieve the same small number of advertisement sets, resulting in no exposure opportunity for other advertisements.

[0049] Please refer toFigure 1 Fig. 2 shows a flow of a method for determining object information according to an embodiment of the present application. Figure 1 The method can be applied in a computer device, which refers to an electronic device with data computing and processing capability. For example, the execution subject of each step can be a server 20 in an application running environment as shown in Fig. 1. The method can include the following steps (210-240). Figure 3

[0050] In step 210, attribute allocation information corresponding to a target object is obtained.

[0051] Optionally, the attribute allocation information is used to represent the allocation of attribute identifiers corresponding to the target object. Optionally, the attribute identifiers include attribute identifiers corresponding to multiple attribute dimensions. The attribute dimensions include but are not limited to gender, age, place of origin, education experience, consumption habit, interest category (such as game, e-commerce, entertainment news, etc.), and behavior attribute (such as a URL identifier corresponding to a click and browse behavior, and an application identifier corresponding to an installation and uninstallation behavior).

[0052] Optionally, the attribute allocation information includes gender allocation information. Optionally, the attribute allocation information includes age allocation information.

[0053] In an example embodiment, as shown in Fig. 2, the implementation process of step 210 can include the following steps (211-213), Figure 3 Figure 2 Fig. 2 shows a flow of a method for determining object information according to an embodiment of the present application. Figure 4

[0054] In step 211, location information corresponding to the target object is obtained.

[0055] In step 212, based on the location information, attribute distribution information corresponding to a target region is determined.

[0056] Optionally, the attribute distribution information is used to represent the distribution of attribute identifiers corresponding to the target region.

[0057] Optionally, the attribute distribution information includes gender ratio data corresponding to the target region.

[0058] Optionally, the attribute distribution information includes age distribution data corresponding to at least one age range identifier.

[0059] In step 213, attribute allocation information is generated according to the attribute distribution information.

[0060] In an example embodiment, the attribute distribution information includes gender ratio data corresponding to the target region. Accordingly, as shown in Fig. 2, Figure 4 ​​​As shown, the implementation process of step 213 can include the following steps (2131-2132). Figure 3 Fig. 1 shows a flow of a method for determining object information according to an embodiment of the present application. Figure 4 .

[0061] Step 2131: taking the gender ratio data as initial gender ratio prediction data.

[0062] In a possible implementation, the gender ratio data corresponding to the target region where the target object is located is determined. The gender ratio data can be determined according to the information of other users in the target region. Optionally, the gender ratio data = the number of objects of the first gender / the total number of objects.

[0063] Optionally, the gender ratio data is taken as the initial value of the gender ratio prediction data (rate).

[0064] Step 2132: determining the gender assignment information based on the initial gender ratio prediction data.

[0065] Optionally, the gender prediction data is randomly generated, if the gender prediction data is less than the gender ratio prediction data, the first gender identifier is assigned to the target object, otherwise the second gender identifier is assigned to the target object.

[0066] For example, the rate = 0.55, a number between 0 and 1 is randomly generated, if the number is less than 0.55, the object is set as a male, otherwise the object is set as a female.

[0067] The embodiments of the present application can predict the gender information of the target object by determining the gender ratio data of the target region, and can assign the target object with gender information,

[0068] In an exemplary embodiment, the attribute distribution information includes age distribution data corresponding to at least one age range identifier. Correspondingly, as shown in Figure 3 As shown, the implementation process of step 213 can also include the following steps (2133-2134).

[0069] Step 2133: taking the age distribution data as initial distribution probability data.

[0070] In a possible implementation, the age distribution data corresponding to the target region where the target object is located is determined. The age distribution data can be determined according to the information of other users in the target region.

[0071] In an example, as shown in Table 1, an age distribution table is shown, which includes age distribution data corresponding to each age range.

[0072] Table 1

[0073]

[0074]

[0075] Optionally, the age range is divided into 6 ranges as shown in Table 1 above, and the corresponding distribution probability is the distribution proportion of the corresponding age range in the population.

[0076] Optionally, the age distribution data described above can be used as initial distribution probability data.

[0077] Step 2134, based on the initial distribution probability data, determining an initial age range identifier corresponding to the target object.

[0078] Optionally, the initial age range identifier is used to represent the age distribution information.

[0079] Optionally, according to the distribution probability data, an age range identifier corresponding to the target object is determined.

[0080] Step 220, based on the attribute distribution information, determining a multimedia object corresponding to the target object.

[0081] In an exemplary embodiment, based on the attribute distribution information described above, an attribute feature corresponding to the target object is determined, and the attribute feature is matched with a multimedia feature corresponding to each multimedia object in the multimedia object set, so that a multimedia object corresponding to the target object can be obtained. The attribute feature can be a feature label generated based on one or more attribute identifiers, or can be information representing data corresponding to the attribute distribution information, and the present embodiment does not limit this.

[0082] Optionally, the multimedia object includes a multimedia object recalled based on the attribute distribution information in a target period. Optionally, the multimedia object includes a multimedia object distributed to the target object based on the attribute distribution information in a target period. Optionally, the target period is 7 days.

[0083] Optionally, the multimedia object includes an advertisement, and accordingly, all advertisements exposed to the target object in the target period are matched and recorded according to the above manner.

[0084] Step 230, obtaining attribute feature information corresponding to the multimedia object.

[0085] Optionally, the attribute feature information is used to represent the degree of association between the multimedia object and at least one attribute identifier. Optionally, the attribute feature information includes gender feature information corresponding to the multimedia object. Optionally, the attribute feature information includes age feature information corresponding to the multimedia object.

[0086] The attribute feature information includes preset attribute feature information corresponding to the multimedia object, such as one or more attribute identifiers corresponding to the multimedia object.

[0087] In an example embodiment, as shown in FIG. 2, the implementation of step 230 includes the following steps (231-234). Figure 4

[0088] Step 231: Obtain the display information corresponding to the multimedia object.

[0089] Optionally, the display information is used to represent the display situation of the multimedia object. The display information includes historical operation data corresponding to the multimedia object.

[0090] Step 232: Determine the first object set based on the display information.

[0091] Optionally, the first object set includes at least one first object having an interaction relationship with the multimedia object. Optionally, the interaction relationship can be determined according to a target operation event, which includes but is not limited to a browsing operation event, a click operation event, a stay operation event, a forwarding operation event, a purchase operation event, and the like triggered by the multimedia object.

[0092] Optionally, the target operation event and at least one first object corresponding to the target operation event can be determined based on the historical operation data.

[0093] Optionally, the at least one first object is a historical conversion object corresponding to the multimedia object, such as an account object that has purchased an advertised product.

[0094] Step 233: Obtain attribute information corresponding to the at least one first object.

[0095] Optionally, the attribute information includes attribute identifiers corresponding to the at least one first object in one or more attribute dimensions.

[0096] Optionally, the attribute information includes gender information corresponding to the at least one first object.

[0097] Optionally, the attribute information includes age information corresponding to the at least one first object.

[0098] Step 234: Determine attribute feature information based on the attribute information.

[0099] Optionally, the attribute feature information includes feature information corresponding to the multimedia object in one or more attribute dimensions. Optionally, the attribute feature information includes gender feature information corresponding to the multimedia object. Optionally, the attribute feature information includes age feature information corresponding to the multimedia object.

[0100] ​In an example embodiment, the attribute distribution information comprises gender distribution information, the attribute information comprises gender information corresponding to the at least one first object, and the attribute feature information comprises gender feature information corresponding to the multimedia object. Accordingly, as shown in Figure 5 The implementation of step 234 can include the following steps (2341-2342).

[0101] In step 2341, gender proportion information corresponding to the first object set is determined based on the gender information.

[0102] Optionally, the gender proportion information comprises a first proportion corresponding to a first gender identifier and a second proportion corresponding to a second gender identifier.

[0103] In step 2342, the gender feature information is generated according to the gender proportion information.

[0104] In an example embodiment, the gender proportion information comprises a first proportion corresponding to a first gender identifier and a second proportion corresponding to a second gender identifier. Accordingly, as shown in Figure 5 The implementation of step 2342 can include the following steps (2342a-2342d), Figure 4 a flowchart of an object information determination method provided by an embodiment of the present application is shown Figure 4 .

[0105] In step 2342a, the first gender feature identifier is obtained in the case where the first proportion is greater than the second proportion.

[0106] Optionally, the first gender feature identifier is used to represent the case where the first proportion is greater than the second proportion.

[0107] The number of multimedia objects corresponding to the target object can be multiple, and each multimedia object in the multiple multimedia objects corresponds to a corresponding first object set. Therefore, according to the gender proportion in the first object set, it can be determined whether the gender feature identifier corresponding to the multimedia object is the first gender feature identifier or the second gender feature identifier.

[0108] The case where the first proportion is greater than the second proportion indicates that the objects of the first gender are more interested in the multimedia object, i.e., representing that the association degree between the multimedia object and the first gender identifier is higher than the association degree between the multimedia object and the second gender identifier.

[0109] In step 2342b, the gender feature identifier data is generated based on the first gender feature identifier.

[0110] The first gender feature identifier is added to the bit position of the multimedia object corresponding thereto, to generate a gender feature identifier sequence. The gender feature identifier data can be the gender feature identifier sequence, which includes the gender feature identifier corresponding to the multimedia object. Whether the first gender feature identifier or the second gender feature identifier is determined according to the relationship between the first proportion and the second proportion.

[0111] In step 2342c, the second gender feature identifier is obtained when the first proportion is less than the second proportion.

[0112] Optionally, the second gender feature identifier is used to represent the case that the first proportion is less than the second proportion.

[0113] The case that the first proportion is less than the second proportion indicates that the object of the second gender is more interested in the multimedia object, i.e., the association degree between the multimedia object and the first gender identifier is lower than the association degree between the multimedia object and the second gender identifier.

[0114] In step 2342d, the gender feature identifier data is generated based on the second gender feature identifier.

[0115] The gender feature identifier data is used to represent the gender feature information.

[0116] The process of generating the gender feature identifier data is described below through specific examples for better understanding. Please refer to Table 2 below, which exemplarily shows the advertisement objects corresponding to two account objects and the gender feature identifier data corresponding to the advertisement objects.

[0117] Table 2

[0118]

[0119] The account object 1 is assigned a gender of male, the advertisement exposure list exposed to the account object 1 within 7 days is list 1, and the gender feature identifier sequence corresponding to the list 1 is sequence 2. The sequence 2 includes the gender feature identifier corresponding to each advertisement in the list 1. If the historical conversion population of an advertisement is dominated by males, the gender feature identifier corresponding to the advertisement is “yes”; if the historical conversion population of an advertisement is dominated by females, the gender feature identifier corresponding to the advertisement is “no”.

[0120] In exemplary embodiments, the attribute assignment information includes age assignment information, the attribute information includes age information corresponding to the at least one first object, and the attribute feature information includes age feature information corresponding to the multimedia object. Accordingly, as shown in Figure 4 The implementation process of the step 234 can further include the following step 2343.

[0121] In step 2343, age distribution information corresponding to the at least one first object is determined based on the age information.

[0122] Optionally, age characteristic information includes age distribution information.

[0123] Optionally, the age distribution information includes probability distribution data corresponding to at least one age group identifier.

[0124] Optionally, based on the age data corresponding to each first object, the age distribution information of at least one first object in each age group can be determined, and then the distribution probability data corresponding to the corresponding age group identifier can be determined based on the object distribution ratio data corresponding to each age group.

[0125] In one possible application scenario, the system obtains all advertisements exposed to the target audience and then retrieves the historical conversion audience of these advertisements. Assume there are 10 advertisements exposed to the target audience, and these 10 advertisements have 200 conversion users. An ad conversion user is a user who sees the ad and makes a conversion. This allows us to statistically analyze the age distribution of these 200 users, resulting in an age distribution table similar to Table 1 above.

[0126] Step 240: Update the attribute allocation information based on the attribute feature information to obtain the updated attribute allocation information.

[0127] In an exemplary embodiment, such as Figure 5 As shown, after step 2342 above, the implementation process of step 240 above may include the following step 241.

[0128] Step 241: Update the gender allocation information based on gender characteristic information to obtain the updated gender allocation information.

[0129] In an exemplary embodiment, such as Figure 6 As shown, after step 2342d above, the implementation process of step 241 above may include the following steps (2411 to 2414).

[0130] Step 2411: Based on the gender feature identification data, determine the number of first identifiers corresponding to the first gender identifier and the number of second identifiers corresponding to the second gender identifier.

[0131] Optionally, the first identifier quantity is used to represent the number of first gender identifiers in the gender feature identifier data, and the second identifier quantity is used to represent the number of second gender identifiers in the gender feature identifier data.

[0132] Step 2412: Obtain gender ratio prediction data.

[0133] Gender ratio prediction data is used to characterize the degree of matching between the target object and the primary gender identifier.

[0134] Step 2413, calibrating the gender ratio prediction data according to the first identification quantity and the second identification quantity, to obtain calibrated gender ratio prediction data.

[0135] In an example embodiment, as shown in FIG. 24, the implementation process of step 2413 can include the following steps (2413a-2413d), Figure 6 Figure 5 FIG. 24 shows a flowchart of an object information determination method provided by an embodiment of the present application. Figure 4 .

[0136] Step 2413a, obtaining a first weight and a second weight.

[0137] Optionally, the first weight is less than the second weight.

[0138] Step 2413b, in the case where the gender distribution information includes a first gender identification, generating first gender parameter data based on the first identification quantity and the first weight.

[0139] Optionally, the first gender parameter data is used to represent the degree of association between the target object and the first gender identification.

[0140] Optionally, the first identification quantity is multiplied by the first weight to obtain the first gender parameter data.

[0141] Step 2413c, generating second gender parameter data based on the second identification quantity and the second weight.

[0142] Optionally, the second gender parameter data is used to represent the degree of association between the target object and the second gender identification.

[0143] Optionally, the second identification quantity is multiplied by the second weight to obtain the second gender parameter data.

[0144] Step 2413d, in the case where the gender distribution information includes a second gender identification, generating first gender parameter data based on the first identification quantity and the second weight.

[0145] Optionally, the first identification quantity is multiplied by the second weight to obtain the first gender parameter data.

[0146] Step 2413e, generating second gender parameter data based on the second identification quantity and the first weight.

[0147] Optionally, the second identification quantity is multiplied by the first weight to obtain the second gender parameter data.

[0148] ​The generation process of the first gender parameter data and the second gender parameter data is described below through specific examples. In the initial state, the first gender parameter data corresponding to the first gender parameter P1 is 0, and the second gender parameter data corresponding to the second gender parameter P2 is 0.

[0149] Traverse the multimedia object corresponding to the target object.

[0150] If the gender assignment information corresponding to the target object is the first gender identifier, and the gender feature identifier corresponding to the current multimedia object is the first gender feature identifier, the first gender parameter data P1 is increased by the first weight (for example, P1 is increased by 1, that is, P1=P1+1, and 1 can be used as the first weight);

[0151] If the gender assignment information corresponding to the target object is the first gender identifier, and the gender feature identifier corresponding to the current multimedia object is the second gender feature identifier, the second gender parameter data P2 is increased by the second weight (for example, P2 is increased by 2, that is, P2=P2+2, and 2 can be used as the first weight);

[0152] If the gender assignment information corresponding to the target object is the second gender identifier, and the gender feature identifier corresponding to the current multimedia object is the first gender feature identifier, the first gender parameter data P1 is increased by the second weight (for example, P1 is increased by 2, that is, P1=P1+2);

[0153] If the gender assignment information corresponding to the target object is the second gender identifier, and the gender feature identifier corresponding to the current multimedia object is the second gender feature identifier, the second gender parameter data P2 is increased by the first weight (for example, P2 is increased by 1, that is, P1=P2+1).

[0154] Through the above traversal process, it can be seen that in the case that the gender assignment information includes the first gender identifier, the first identifier quantity is multiplied by the first weight to obtain the first gender parameter data, and the second identifier quantity is multiplied by the second weight to obtain the second gender parameter data.

[0155] In the case that the gender assignment information includes the second gender identifier, the first identifier quantity is multiplied by the second weight to obtain the first gender parameter data, and the second identifier quantity is multiplied by the first weight to obtain the second gender parameter data.

[0156] In step 2413f, the gender proportion prediction data is calibrated based on the first gender parameter data and the second gender parameter data to obtain calibrated gender proportion prediction data.

[0157] In an exemplary embodiment, a parameter sum (P1+P2) corresponding to the first gender parameter data P1 and the second gender parameter data P2 is determined, and a ratio P1_rate=P1 / (P1+P2) between the first gender parameter data P1 and the parameter sum is determined.

[0158] Based on the ratio P1_rate and the predicted sex ratio data rate, update the predicted sex ratio data rate.

[0159] If the above ratio is greater than the first multiple of the predicted gender ratio, such as P1_rate>1.1*rate, then the predicted gender ratio is increased, and the increase is not limited, such as rate=rate+0.02.

[0160] If the above ratio is greater than or equal to the second multiple of the predicted gender ratio and less than or equal to the first multiple of the predicted gender ratio, for example, P1_rate>=0.9*rate&&P1_rate<=1.1*rate, then the predicted gender ratio remains unchanged, i.e., rate=rate.

[0161] If the above ratio is less than a multiple of the predicted sex ratio, such as P1_rate < 0.9 * rate, then the predicted sex ratio should be reduced, and the reduction is not limited, for example, rate = rate - 0.02.

[0162] When performing the next round of operations, the updated rate will be recalculated and the process will be repeated.

[0163] Step 2414: Determine the updated gender allocation information based on the calibrated gender ratio prediction data.

[0164] In an exemplary embodiment, such as Figure 5 As shown, after step 2343 above, the implementation process of step 240 above may include the following step 242.

[0165] Step 242: Update the age distribution information based on the age distribution information to obtain the updated age distribution information.

[0166] In an exemplary embodiment, the age distribution information includes distribution probability data corresponding to at least one age group identifier. Accordingly, such as... Figure 7 As shown, the implementation process of step 242 above may include the following steps (2421 to 2423).

[0167] Step 2421: Obtain the previously determined probability distribution data.

[0168] Step 2422: Based on the distribution probability data, calibrate the previously determined distribution probability data to obtain calibrated distribution probability data.

[0169] Optionally, the distribution probability data corresponding to each age range determined this time is weighted and averaged with the distribution probability data corresponding to each age range determined last time to obtain calibrated distribution probability data.

[0170] For example, the calibrated distribution probability data = 0.1 * new distribution probability data + 0.9 * distribution probability data determined last time. The weighting coefficients 0.1 and 0.9 are predetermined or dynamically adjusted during the calculation process.

[0171] At step 2423, the target age range corresponding to the target object is determined according to the calibrated distribution probability data.

[0172] The target age range is used to represent the updated age distribution information.

[0173] Optionally, the multimedia object corresponding to the target object is re-determined based on the updated attribute distribution information.

[0174] To sum up, the technical scheme provided by the embodiments of the present application can determine the multimedia object associated with the attribute distribution information through the attribute distribution information capable of representing the attribute range distribution of the target object, update the attribute distribution information according to the attribute feature information of the multimedia object, and improve the accuracy and authenticity of the attribute distribution information. The updated attribute distribution information is more consistent with the real attributes of the target object, thereby improving the information quantity and accuracy of the feature information of the target object without using the personalized features of the target object.

[0175] The technical scheme provided by the embodiments of the present application and the beneficial effects thereof will be described below in combination with a specific application field.

[0176] Exemplarily, the multimedia object pushing field is a typical application field corresponding to the embodiments of the present application. In the multimedia object pushing field, the multimedia objects in the multimedia object library can be intelligently targeted. The concept of intelligent targeting has been explained above, and after confirming the function of intelligent targeting, it is necessary to consider how to implement intelligent targeting. The essence of intelligent targeting is to find the appropriate crowd for the multimedia object, and the goal is to recall the target multimedia object with high matching degree for the target object when receiving the multimedia object pushing request of the target object. Optionally, the multimedia object is an advertisement, and the recalled target multimedia object is an advertisement matching the interest of the target object.

[0177] The matching degree can be determined in different ways, which is not limited by the embodiments of the present application. Illustratively, the embodiments of the present application provide two different types of matching methods to determine the matching degree between the target object and the multimedia object.

[0178] In a possible implementation, the matching is triggered by an ANN (Approximate Nearest Neighbor) branch. The original intention of the ANN is to simulate the mining of a multimedia object corresponding to a multimedia publishing object crowd package. For example, an advertisement publisher generally determines a possible high conversion crowd through historical data, and has a better delivery benefit. Through the ANN, the historical data can be automatically analyzed, so as to determine a possible high conversion object from the historical data, and the multimedia object has a better delivery effect.

[0179] The ANN is described below in combination with a specific example. Optionally, the ANN can be implemented based on an artificial neural network, that is, an artificial neural network model is established, and the model is used to predict the matching degree between the multimedia object and the account object. For details, refer to Figure 8 which exemplarily shows a schematic diagram of an approximate nearest neighbor search process. The general process is as follows:

[0180] The multi-party data source is obtained. Optionally, the multimedia object is an advertisement. The data source includes consumed account object data corresponding to the multimedia object, delivery data (including converted object data) corresponding to the multimedia object, and platform data (including target objects and feature data corresponding to the multimedia object).

[0181] The multi-party data source (wherein X1, X2, and X3 represent different feature data corresponding to the multimedia object) is input into a targeting model automatically constructed based on a machine learning model, and the targeting model is used to output a matching score between the multimedia object and each account object, and the matching score is used to represent the matching degree.

[0182] Optionally, the targeting model can be generally abstracted as a probability estimation problem P(1|user (account object), context (context information), ad (multimedia object identifier)), that is, under a predefined target type, the probability of occurrence of the target object when seeing the multimedia object in a certain context environment. The target can represent the matching degree between the multimedia object and the target object. Optionally, the target data capable of representing the matching degree includes, but is not limited to, a click rate, a click conversion rate, and an exposure conversion rate corresponding to the multimedia object. Optionally, a plurality of different predefined target types can be modeled respectively to construct different ANN models, form a plurality of matching trigger branches, and then the plurality of branches are combined. For example, a click rate model, a click conversion rate model, and an exposure conversion rate model are constructed, and then the targeting models are used to determine whether the target object and the multimedia object match. Optionally, the target data is positively correlated with the matching degree.

[0183] In one example, as shown in Figure 8 FIG. 1, an example is shown to illustrate the construction flowchart of the targeting model. In the offline training phase, on one hand, the object features of the sample object (e.g. sample account object) and the context features corresponding to the context information are input into the object representation model, and the object features fused with the context information can be output; on the other hand, the multimedia object information is input into the multimedia representation model, and the multimedia object features corresponding to the multimedia object can be output. By using these features, label data (such as the click rate, click conversion rate, and exposure conversion rate) and related constraints, the corresponding targeting model can be constructed. The trained targeting model can output the object features fused with the context information corresponding to the target object according to the feature information of the target, such as the object features and context features corresponding to the target object, and match and search the multimedia object feature library to output the target data.

[0184] From the above Figure 8 , the following conclusions can be drawn:

[0185] 1. Generally, the model structure is a double-tower structure, as shown in Figure 9 above, the object features and the context features are in one tower, and the multimedia object information is in another tower.

[0186] 2. The context information only appears on the account object side and does not appear on the multimedia object side. That is, it means that the information on the account object side is related to the context and may be a dynamic feature that changes with the context. The information on the multimedia object side is independent of the context, that is, a static feature.

[0187] 3. The determination method of the context features is not limited, and the context features can be determined by various methods.

[0188] 4. In the offline phase, the model structure and parameters of the account object tower are obtained by training, and the multimedia object features of each multimedia object on the multimedia object side, such as embedding vectors, are obtained.

[0189] 5. In the online prediction phase, the embedding vector based on the context is obtained for the object features and context features of the target object of a request, and using this vector, a plurality of multimedia objects corresponding to the vectors similar to this vector are found from the embedding vectors corresponding to the multimedia objects as the recalled target multimedia objects.

[0190] The above describes the ANN, and the following describes another possible implementation.

[0191] In one possible implementation, the matching is performed through the TAG (label-based recall) trigger branch.

[0192] In the former embodiment, more details of the request can be described by refining and transforming the context features, but there can be a problem of concentration of multimedia objects. That is, many requests get similar multimedia objects recalled, and some multimedia objects do not get a reasonable exposure opportunity. In addition, there is a problem of newly entered multimedia objects. The above model-based training is essentially to find rules from historical data. However, for newly entered multimedia objects, there is less historical data, which leads to relatively inaccurate training and estimation of the model for newly entered multimedia objects. Thus, the newly entered multimedia objects do not get appropriate exposure opportunities.

[0193] The above TAG branch adopts a new matching mode different from the ANN branch, that is, a label recall-based mode. The TAG branch aims to simulate the process of selecting directional labels for multimedia publishing objects. In short, in the label recall-based mode, labels can be assigned to account objects, and then labels can be assigned to multimedia objects, and then the labels of the account objects are matched with the labels of the multimedia objects, and the matched ones are recalled. In an example, as shown in Figure 10 the label categories corresponding to the account objects include behavior direction categories, interest direction categories, intention direction categories, application installation direction categories, and custom crowd categories. Optionally, the labels of the behavior direction categories include category labels and keyword labels; the labels of the interest direction categories include category labels and keyword labels; the labels of the intention direction categories include category labels; the labels of the application installation direction categories include application labels in the installed list; and the labels of the custom crowd categories include labels in the crowd package list. The labels corresponding to the multimedia objects include creative labels, application identifiers, related conversion objects, and account identifiers. The matching model matches according to different labels on both sides, thereby determining the recalled multimedia objects.

[0194] From the above description, it can be known that the quality of the labels of the account objects and the labels of the multimedia objects, and the matching mode can affect the recall effect of the TAG branch. Therefore, the construction of the TAG branch can be considered from the following aspects.

[0195] 1. The behavior interest direction labels corresponding to the account objects.

[0196] In some possible application scenarios, the same system of labels can be assigned to multimedia object advertisements according to the behavior interest direction label system on the account object side. Directly matched multimedia objects can be recalled when a request occurs. For example, a certain multimedia object is labeled with the label of "wuxia game lovers", and then the account object of a certain request is also labeled with this label, and then this advertisement is recalled in this request.

[0197] 2. Automatically bind application installation targeting label.

[0198] In some possible application scenarios, a similar application corresponding to the target application is determined, and if the application identifier corresponding to the target application is included in the application identifier corresponding to the target account object, the multimedia object corresponding to the similar application can be recalled when the target account object requests.

[0199] 3. Automatically bind custom people group targeting label.

[0200] In some possible application scenarios, the account object cluster package corresponding to the publishing object is determined as the label of the multimedia object corresponding to the publishing object.

[0201] 4. Automatically bind converted object high TGI (Target Group Index) targeting label.

[0202] In some possible application scenarios, if only the converted objects of the multimedia object are considered, they are relatively sparse, so a certain rule is used for fallback. Generally, the priority of fallback is as follows: multimedia object (such as an advertisement) -> associated multimedia object (such as a product) -> publishing object (such as an advertiser) -> industry. That is, first check whether the converted objects of the same multimedia object are sufficient, otherwise, fallback to the converted objects of the associated multimedia object, and so on. The high TGI targeting label refers to a relatively large proportion of this targeting label in the converted objects. This is a process of using results to feedback, trying to strengthen the results through the observed relatively good labels.

[0203] As described above, whether matching and recalling is performed through the ANN branch or the TAG branch, the feature information corresponding to the target object is used, and the multimedia object recall process has a high degree of dependence on the feature information corresponding to the target object. For some users who turn off personalized recommendation, if the above two methods are still used to match and recall multimedia objects, the recall amount of multimedia objects will be sharply reduced. For details, refer to Figure 10 , which exemplarily shows a schematic diagram of a recall amount data display page. According to Figure 11 , it can be seen that for the request of the account object whose personalized recommendation is set to off, the average length of the recall queue is less than 500, and for the request of the account object whose personalized recommendation is set to on, the length of the recall queue is about 36000 or more. It can be seen that after the personalized recommendation is turned off, the recall amount of multimedia objects is sharply reduced.

[0204] The reason for the sharp reduction in the recall amount of multimedia objects is analyzed as follows. Refer to RequestFig. 1 is a schematic diagram of a technical architecture of a multimedia recall system. In Fig. 1, the digital labels "1", "2", "3", "4" and "5" represent the positions of the same meaning. As mentioned above, when the personalized recommendation setting is turned off, the recall amount of multimedia objects is sharply reduced, that is, the recall amount at position 5 is sharply reduced. For analysis, Table 3 below lists the recall amounts corresponding to positions 1 to 5 at different time points and under different requests when the personalized recommendation setting is turned on. Table 4 below lists the recall amounts corresponding to positions 1 to 5 at different time points and under different requests when the personalized recommendation setting is turned off.

[0205] Table 3 (personalized recommendation setting turned on)

[0206] Location 1 Location 2 Location 3 Location 4 Location 5 Request 1 Request 2 24067 38901 47309 174909 36890 Request 3 28381 40910 54810 189032 37991 Request 4 23921 35438 42672 149843 32704 Request 5 27419 41341 52873 193671 36308 Request 29304 46937 50900 164793 38927

[0207] Table 4 (personalized recommendation setting turned off)

[0208] Location 1 Location 2 Location 3 Location 4 Location 5 Request 1 Request 2 20389 14834 24091 3978 498 Request 3 19048 13391 22845 7832 530 Request 4 17481 20501 30903 6346 513 Request 5 19762 11945 21351 4831 406 Multimedia object identification 22469 19772 27924 5048 484

[0209] From the above statistical data, it is not difficult to see that the recall amounts of the three main positions 1, 2 and 4 have all decreased to some extent. Among them, the decrease in the recall amount at position 4 plays a decisive role in the decrease in the final recall amount.

[0210] The recall amount at position 1 is the number of multimedia objects recalled by the ANN model. The calculation of this model also needs to use object information. If the object information is missing, it will affect the model calculation and the recall.

[0211] The recall amount at position 2 is the number of multimedia objects recalled by the TAG branch. For this part of target objects with missing information, the recall amount will be impaired.

[0212] The recall amount at position 4 is the number of multimedia objects that cannot be broken through directional recall. Unbreakable directional recall is essentially to recall all multimedia objects that meet all unbreakable conditions. For example, if the object information corresponding to this request indicates that the object is a male student, all multimedia objects with "male student" set as unbreakable direction will be recalled, and multimedia objects without gender requirements will also be recalled. When the personalized recommendation setting is turned off, the system does not know the true gender of the target object, and can only recall multimedia objects without gender requirements, and the recall amount is sharply reduced.

[0213] The reason for the sharp decrease in the recall amount at position 4 is explained below through a specific example. Please refer to Table 5 below, which exemplarily shows the unbreakable directional data corresponding to multimedia objects.

[0214] Table 5

[0215] Unbreakable orientation Shanghai; male 1 Male 2 Shanghai 3 Beijing; female 4 Male; like playing games 5 Female; like online shopping 6 Male; between 30 and 40 years old 7 Request

[0216] Here, the unbreakable orientation is briefly described first. The unbreakable orientation is that the orientation information cannot be violated in the delivery of the multimedia object. For example, the unbreakable orientation of the multimedia object identified as 1 is "Shanghai; male", which indicates that the multimedia object 1 can only be delivered to the account object meeting the condition of "Shanghai; male", and the rest cannot.

[0217] Suppose a request is received now, the request object does not set the closed personalized feature, the system can obtain the information that the gender is male, does not like playing games, and the age is 31 years old; in addition, it can be obtained that the request comes from Shanghai. This request can recall the multimedia objects 1, 2, 3 and 7 in Table 5 above.

[0218] Suppose another request is received now, the request object sets the closed personalized feature, the system will not obtain the related information of the request object, including gender, age, and hobbies. At this time, the system can only request that the request comes from Shanghai. This request can recall the multimedia object 3 in Table 5 above.

[0219] As can be seen, the multimedia object recall quantity of the user who sets the closed personalized feature is greatly reduced.

[0220] However, through the technical solution provided by the embodiment of the present application, the attribute information can be compensated by the way of assigning attribute identification, the actual attribute data is simulated, the continuously updated and calibrated attribute assignment information is added to each branch above, and then the multimedia is recalled, so that the recall quantity is correspondingly increased. As shown in Table 6 below, compared with Table 4 above, the length of the recall queue of each position is obviously increased. In the request of the user who closes the personalized feature, the recall quantity loss is effectively reduced, the information quantity of the feature information of the target object can be improved without using the personalized feature of the target object, and the updated and calibrated attribute assignment information can be more consistent with the real attribute, so that the accuracy of the feature information can be improved, and then the recall quality is improved.

[0221] Table 6

[0222] Location 1 Location 2 Location 3 Location 4 Location 5 Request 1 Request 2 24182 15837 27093 98578 20904 Request 3 22048 11492 23802 74910 19560 Request 4 19387 21603 27789 84091 21807 Request 5 25712 14745 28801 100851 18938 Figure 12 27361 23771 30027 117892 19765

[0223] The following is an apparatus embodiment of the present application, which can be used to execute the method embodiments of the present application. For details not disclosed in the apparatus embodiments of the present application, please refer to the method embodiments of the present application.

[0224] Please refer to Figure 13FIG. 12 shows a block diagram of an object information determination apparatus according to an example embodiment of the present application. The apparatus has the functions of the object information determination method described above, which can be implemented by hardware or by execution of corresponding software by hardware. The apparatus can be a computer device or can be arranged in a computer device. The apparatus 1200 can include an attribute information allocation module 1210, a multimedia object determination module 1220, an attribute feature acquisition module 1230, and an attribute information update module 1240.

[0225] The attribute information allocation module 1210 is configured to obtain attribute allocation information corresponding to a target object, the attribute allocation information being used to represent an attribute identifier allocation situation corresponding to the target object.

[0226] The multimedia object determination module 1220 is configured to determine a multimedia object corresponding to the target object based on the attribute allocation information.

[0227] The attribute feature acquisition module 1230 is configured to obtain attribute feature information corresponding to the multimedia object, the attribute feature information being used to represent an association degree between the multimedia object and at least one attribute identifier.

[0228] The attribute information update module 1240 is configured to update the attribute allocation information according to the attribute feature information to obtain updated attribute allocation information.

[0229] In an example embodiment, the attribute feature acquisition module 1230 includes a display information acquisition sub-module, an object set determination sub-module, an object attribute acquisition sub-module, and an attribute feature determination sub-module.

[0230] The display information acquisition sub-module is configured to obtain display information corresponding to the multimedia object, the display information being used to represent a display situation of the multimedia object.

[0231] The object set determination sub-module is configured to determine a first object set based on the display information, the first object set including at least one first object having an interaction relationship with the multimedia object.

[0232] The object attribute acquisition sub-module is configured to obtain attribute information corresponding to the at least one first object.

[0233] The attribute feature determination sub-module is configured to determine the attribute feature information based on the attribute information.

[0234] In an example embodiment, the attribute distribution information comprises gender distribution information, the attribute information comprises gender information corresponding to the at least one first object, the attribute feature information comprises gender feature information corresponding to the multimedia object, and the attribute feature determination submodule comprises a gender proportion determination unit and a gender feature determination unit.

[0235] The gender proportion determination unit is configured to determine gender proportion information corresponding to the first object set based on the gender information.

[0236] The gender feature determination unit is configured to generate the gender feature information according to the gender proportion information.

[0237] The attribute information updating module 1240 comprises a gender information updating submodule.

[0238] The gender information updating submodule is configured to update the gender distribution information according to the gender feature information to obtain updated gender distribution information.

[0239] In an example embodiment, the gender proportion information comprises a first proportion corresponding to a first gender identifier and a second proportion corresponding to a second gender identifier, and the gender feature determination unit comprises a gender identifier acquisition subunit and a gender feature generation subunit.

[0240] The gender identifier acquisition subunit is configured to acquire a first gender feature identifier in a case where the first proportion is greater than the second proportion.

[0241] The gender feature generation subunit is configured to generate gender feature identifier data based on the first gender feature identifier.

[0242] The gender identifier acquisition subunit is further configured to acquire a second gender feature identifier in a case where the first proportion is less than the second proportion.

[0243] The gender feature generation subunit is further configured to generate the gender feature identifier data based on the second gender feature identifier.

[0244] The gender feature identifier data is used to represent the gender feature information, the first gender feature identifier is used to represent the case where the first proportion is greater than the second proportion, and the second gender feature identifier is used to represent the case where the first proportion is less than the second proportion.

[0245] In an example embodiment, the gender information updating submodule comprises an identifier quantity determination unit, a gender data acquisition unit, a gender data calibration unit, and a gender information distribution unit.

[0246] An identifier quantity determination unit is configured to determine a first identifier quantity corresponding to the first gender identifier and a second identifier quantity corresponding to the second gender identifier based on the gender feature identifier data.

[0247] A gender data acquisition unit is configured to acquire gender proportion prediction data, which is used to represent a matching degree between the target object and the first gender identifier.

[0248] A gender data calibration unit is configured to calibrate the gender proportion prediction data to obtain calibrated gender proportion prediction data according to the first identifier quantity and the second identifier quantity.

[0249] A gender information distribution unit is configured to determine the updated gender distribution information based on the calibrated gender proportion prediction data.

[0250] In an example embodiment, the gender data calibration unit comprises a weight acquisition subunit, a gender parameter generation subunit, and a gender data calibration subunit.

[0251] The weight acquisition subunit is configured to acquire a first weight and a second weight, wherein the first weight is smaller than the second weight.

[0252] The gender parameter generation subunit is configured to generate first gender parameter data based on the first identifier quantity and the first weight, and generate second gender parameter data based on the second identifier quantity and the second weight, when the gender distribution information comprises the first gender identifier.

[0253] The gender parameter generation subunit is further configured to generate the first gender parameter data based on the first identifier quantity and the second weight, and generate the second gender parameter data based on the second identifier quantity and the first weight, when the gender distribution information comprises the second gender identifier.

[0254] The gender data calibration subunit is configured to calibrate the gender proportion prediction data based on the first gender parameter data and the second gender parameter data to obtain the calibrated gender proportion prediction data.

[0255] The first gender parameter data is used to represent an association degree between the target object and the first gender identifier, and the second gender parameter data is used to represent an association degree between the target object and the second gender identifier.

[0256] In an example embodiment, the attribute distribution information comprises age distribution information, the attribute information comprises age information corresponding to the at least one first object, the attribute feature information comprises age feature information corresponding to the multimedia object, and the attribute feature determining submodule comprises an age distribution determining unit.

[0257] The age distribution determining unit is configured to determine age distribution information corresponding to the at least one first object based on the age information, and the age feature information comprises the age distribution information.

[0258] The attribute information updating module 1240 comprises an age information updating submodule.

[0259] The age information updating submodule is configured to update the age distribution information according to the age distribution information to obtain updated age distribution information.

[0260] In an example embodiment, the age distribution information comprises distribution probability data corresponding to at least one age range identifier, and the age information updating submodule comprises a distribution data obtaining unit, a distribution data calibrating unit and an age identifier determining unit.

[0261] The distribution data obtaining unit is configured to obtain previously determined distribution probability data.

[0262] The distribution data calibrating unit is configured to calibrate the previously determined distribution probability data based on the distribution probability data to obtain calibrated distribution probability data.

[0263] The age identifier determining unit is configured to determine a target age range identifier corresponding to the target object according to the calibrated distribution probability data, and the target age range identifier is used to represent the updated age distribution information.

[0264] In an example embodiment, the attribute information distribution module 1210 comprises a location obtaining unit, an attribute distribution determining unit and an attribute distribution unit.

[0265] The location obtaining unit is configured to obtain location information corresponding to the target object.

[0266] The attribute distribution determining unit is configured to determine attribute distribution information corresponding to a target region based on the location information, and the attribute distribution information is used to represent attribute identifier distribution of the target region.

[0267] The attribute distribution unit is configured to generate the attribute distribution information according to the attribute distribution information.

[0268] In an example embodiment, the attribute distribution information comprises gender proportion data corresponding to the target region, the attribute allocation information comprises gender allocation information, and the attribute allocation unit comprises a gender data initialization subunit and a gender allocation subunit.

[0269] The gender data initialization subunit is configured to use the gender proportion data as initial gender proportion prediction data.

[0270] The gender allocation subunit is configured to determine the gender allocation information based on the initial gender proportion prediction data.

[0271] In an example embodiment, the attribute distribution information comprises age distribution data corresponding to at least one age range identifier, the attribute allocation information comprises age allocation information, and the attribute allocation unit comprises an age data initialization subunit and an age allocation subunit.

[0272] The age data initialization subunit is configured to use the age distribution data as initial distribution probability data.

[0273] The age allocation subunit is configured to determine an initial age range identifier corresponding to the target object based on the initial distribution probability data, and the initial age range identifier is used to represent the age allocation information.

[0274] In summary, the technical scheme provided by the embodiments of the present application can determine the multimedia objects associated with the attribute allocation information, update the attribute allocation information according to the attribute feature information of the multimedia objects, and improve the accuracy and authenticity of the attribute allocation information. In addition, the information amount and accuracy of the feature information of the target object can be improved without using the personalized features of the target object.

[0275] It should be noted that the device provided in the above embodiments is only used as an example to illustrate the division of the above functional modules, and in actual applications, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the above described functions. In addition, the device and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process is described in detail in the method embodiments, which will not be described here.

[0276] Please refer to ​ which shows a structural block diagram of a computer device provided in an embodiment of the present application. The computer device can be a server for executing the above object information determination method. Specifically:

[0277] The computer device 1300 includes a central processing unit (CPU) 1301, a system memory 1304, including a random access memory (RAM) 1302 and a read-only memory (ROM) 1303, and a system bus 1305 that couples the system memory 1304 to the central processing unit 1301. The computer device 1300 also includes a basic input / output system (I / O) 1306 that helps transfer information between elements of the computer, and a mass storage device 1307 for storing an operating system 1313, application programs 1314, and other program modules 1315.

[0278] The basic input / output system 1306 includes a display 1308 for displaying information and input devices 1309, such as a mouse, keyboard, etc., for inputting information. Both the display 1308 and the input devices 1309 are connected to the central processing unit 1301 through input / output controllers 1310 that are connected to the system bus 1305. The basic input / output system 1306 can also include input / output controllers 1310 for receiving and processing input from a number of other devices, such as a keyboard, mouse, or electronic stylus. Similarly, the input / output controllers 1310 provide output to a display screen, printer, or other type of output device.

[0279] The mass storage device 1307 is connected to the central processing unit 1301 through a mass storage controller (not shown) that is connected to the system bus 1305. The mass storage device 1307 and its associated computer-readable media provide non-volatile storage for the computer device 1300. That is, the mass storage device 1307 can include a computer-readable medium (not shown) such as a hard disk or a CD-ROM (Compact Disc Read-Only Memory) drive.

[0280] Without loss of generality, computer-readable media can include computer storage media and communication media. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes RAM, ROM, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read Only Memory), flash memory or other solid state memory technology, CD-ROM, digital versatile disks (DVD), or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices. Computer storage media does not, however, include communication media. The system memory 1304 and mass storage device(s) 1307 described above can collectively be referred to as memory.

[0281] According to various embodiments of the present application, the computer device 1300 can also operate in a networking environment via the network 1312 with remote computers. The computer device 1300 can connect to the network 1312 through a network interface unit 1311 attached to the system bus 1305, or can connect to another type of network or remote computer system (not shown) via the network interface unit 1311.

[0282] The memory also includes a computer program that is stored in the memory and configured to be executed by one or more processors to implement the above-mentioned object information determination method.

[0283] In an exemplary embodiment, a computer-readable storage medium is also provided, in which at least one instruction, at least one program, a code set or an instruction set is stored, and when executed by a processor, the at least one instruction, the at least one program, the code set or the instruction set implements the above-mentioned object information determination method.

[0284] Optionally, the computer readable storage medium can include: ROM (Read Only Memory), RAM (Random Access Memory), SSD (Solid State Drives), optical disc, etc. Among them, the random access memory can include ReRAM (Resistance Random Access Memory) and DRAM (Dynamic Random Access Memory).

[0285] In an example embodiment, a computer program product or computer program is also provided, which includes computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device executes the above object information determination method.

[0286] It should be understood that "multiple" mentioned herein refers to two or more. "And / or" describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent: A exists alone, A and B exist together, and B exists alone. The character " / " generally represents that the associated objects before and after it are in an "or" relationship. In addition, the step numbers described herein only exemplarily show a possible execution order between steps. In some other embodiments, the above steps can also be executed in a different order, such as two different numbered steps being executed simultaneously, or two different numbered steps being executed in an order opposite to that shown in the figure, and the embodiments of the present application do not limit this.

[0287] In addition, in the specific embodiments of the present application, data related to user information and the like are involved. When the above embodiments of the present application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data need to comply with relevant national and regional laws, regulations and standards.

[0288] The above only describes the example embodiments of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. An object information determination method characterized by comprising: The method comprises: obtaining attribute allocation information corresponding to a target object, the attribute allocation information being used to represent an attribute identifier allocation situation corresponding to the target object; determining a multimedia object corresponding to the target object based on the attribute allocation information; obtaining attribute feature information corresponding to the multimedia object, the attribute feature information being used to represent an association degree between the multimedia object and at least one attribute identifier; updating the attribute allocation information according to the attribute feature information to obtain updated attribute allocation information; the attribute allocation information comprises gender allocation information, the attribute feature information comprises gender feature information corresponding to the multimedia object, and the updating of the attribute allocation information according to the attribute feature information to obtain the updated attribute allocation information comprises: determining a first identifier quantity corresponding to a first gender identifier and a second identifier quantity corresponding to a second gender identifier based on gender feature identifier data, the gender feature identifier data being used to represent the gender feature information; obtaining gender ratio prediction data, the gender ratio prediction data being used to represent a matching degree between the target object and the first gender identifier; calibrating the gender ratio prediction data according to the first identifier quantity and the second identifier quantity to obtain calibrated gender ratio prediction data; determining updated gender allocation information based on the calibrated gender ratio prediction data; the calibration of the gender ratio prediction data according to the first identifier quantity and the second identifier quantity to obtain the calibrated gender ratio prediction data comprises: obtaining a first weight and a second weight, the first weight being smaller than the second weight; in a case where the gender allocation information comprises the first gender identifier, generating first gender parameter data based on the first identifier quantity and the first weight, and generating second gender parameter data based on the second identifier quantity and the second weight; in a case where the gender allocation information comprises the second gender identifier, generating the first gender parameter data based on the first identifier quantity and the second weight, and generating the second gender parameter data based on the second identifier quantity and the first weight; calibrating the gender ratio prediction data based on the first gender parameter data and the second gender parameter data to obtain the calibrated gender ratio prediction data; wherein the first gender parameter data is used to represent an association degree between the target object and the first gender identifier, and the second gender parameter data is used to represent an association degree between the target object and the second gender identifier.

2. The method of claim 1, wherein, the obtaining of the attribute feature information corresponding to the multimedia object comprises: obtaining display information corresponding to the multimedia object, the display information being used to represent a display situation of the multimedia object; determining a first object set based on the display information, the first object set comprising at least one first object having an interaction relationship with the multimedia object; obtaining attribute information corresponding to the at least one first object; determining the attribute feature information based on the attribute information.

3. The method of claim 2, wherein, The attribute information includes gender information corresponding to the at least one first object, and the attribute feature information is determined based on the attribute information, including: Based on the gender information, determine the gender proportion information corresponding to the first object set; According to the gender proportion information, the gender feature information is generated.

4. The method of claim 3, wherein, The gender proportion information includes a first proportion corresponding to a first gender identifier and a second proportion corresponding to a second gender identifier, and the gender feature information is generated according to the gender proportion information, including: In the case where the first proportion is greater than the second proportion, a first gender feature identifier is obtained; and gender feature identifier data is generated based on the first gender feature identifier; In the case where the first proportion is less than the second proportion, a second gender feature identifier is obtained; and the gender feature identifier data is generated based on the second gender feature identifier; Wherein, the gender feature identifier data is used to represent the gender feature information, the first gender feature identifier is used to represent the case where the first proportion is greater than the second proportion, and the second gender feature identifier is used to represent the case where the first proportion is less than the second proportion.

5. The method of claim 1, wherein, The multimedia object corresponding to the target object includes a multimedia object recalled based on the attribute allocation information within a target period.

6. The method of claim 1, wherein, The multimedia object includes an advertisement.

7. The method of claim 2, wherein, The attribute allocation information includes age allocation information, the attribute information includes age information corresponding to the at least one first object, and the attribute feature information includes age feature information corresponding to the multimedia object. The attribute feature information is determined based on the attribute information, including: Based on the age information, determine the age distribution information corresponding to the at least one first object, and the age feature information includes the age distribution information; The attribute allocation information is updated according to the attribute feature information to obtain updated attribute allocation information, including: The age allocation information is updated according to the age distribution information to obtain updated age allocation information.

8. The method of claim 7, wherein, The age distribution information includes distribution probability data corresponding to at least one age range identifier, and the age allocation information is updated according to the age distribution information to obtain updated age allocation information, including: Obtain the distribution probability data determined last time; Based on the distribution probability data, the distribution probability data determined last time is calibrated to obtain calibrated distribution probability data; According to the calibrated distribution probability data, determine the target age range identifier corresponding to the target object, and the target age range identifier is used to represent the updated age allocation information.

9. The method of claim 1, wherein, The attribute allocation information corresponding to the target object is obtained, including: Obtain the location information corresponding to the target object; Based on the location information, determine the attribute distribution information corresponding to the target region, and the attribute distribution information is used to represent the attribute identifier distribution of the target region; According to the attribute distribution information, the attribute allocation information is generated.

10. The method of claim 9, wherein, The attribute distribution information includes gender proportion data corresponding to the target region, the attribute allocation information includes gender allocation information, and the attribute allocation information is generated according to the attribute distribution information, including: The gender proportion data is taken as initial gender proportion prediction data; The gender allocation information is determined based on the initial gender proportion prediction data.

11. The method of claim 9, wherein, The attribute distribution information includes age distribution data corresponding to at least one age range identifier, and the attribute allocation information includes age allocation information. The attribute allocation information is generated according to the attribute distribution information, including: The age distribution data is taken as initial distribution probability data; An initial age range identifier corresponding to the target object is determined based on the initial distribution probability data, and the initial age range identifier is used to represent the age allocation information.

12. An object information determining apparatus characterized by comprising: The device includes: An attribute information allocation module configured to obtain attribute allocation information corresponding to a target object, the attribute allocation information being used to represent an attribute identifier allocation situation of the target object; A multimedia object determination module configured to determine a multimedia object corresponding to the target object based on the attribute allocation information; An attribute feature acquisition module configured to obtain attribute feature information corresponding to the multimedia object, the attribute feature information being used to represent an association degree between the multimedia object and at least one attribute identifier; An attribute information updating module configured to update the attribute allocation information according to the attribute feature information to obtain updated attribute allocation information; The attribute allocation information includes gender allocation information, and the attribute feature information includes gender feature information corresponding to the multimedia object. The attribute allocation information is updated according to the attribute feature information to obtain updated attribute allocation information, including: First identifier quantity corresponding to a first gender identifier and second identifier quantity corresponding to a second gender identifier are determined based on gender feature identifier data, the gender feature identifier data being used to represent the gender feature information; Gender proportion prediction data is obtained, the gender proportion prediction data being used to represent a matching degree between the target object and the first gender identifier; The gender proportion prediction data is calibrated according to the first identifier quantity and the second identifier quantity to obtain calibrated gender proportion prediction data; Updated gender allocation information is determined based on the calibrated gender proportion prediction data; The gender proportion prediction data is calibrated according to the first identifier quantity and the second identifier quantity to obtain calibrated gender proportion prediction data, including: First weight and second weight are obtained, the first weight being smaller than the second weight; In a case where the attribute allocation information includes the first gender identifier, first gender parameter data is generated based on the first identifier quantity and the first weight, and second gender parameter data is generated based on the second identifier quantity and the second weight; In a case where the gender distribution information includes the second gender identifier, the first gender parameter data is generated based on the first identifier quantity and the second weight, and the second gender parameter data is generated based on the second identifier quantity and the first weight; The gender proportion prediction data is calibrated based on the first gender parameter data and the second gender parameter data to obtain calibrated gender proportion prediction data; The first gender parameter data is used to represent the degree of association between the target object and the first gender identifier, and the second gender parameter data is used to represent the degree of association between the target object and the second gender identifier.

13. A computer device, comprising: The computer device includes a processor and a memory, and the memory stores at least one instruction, at least one program, a code set or an instruction set, which are loaded and executed by the processor to implement the object information determination method according to any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, The storage medium stores at least one instruction, at least one program, a code set or an instruction set, which are loaded and executed by the processor to implement the object information determination method according to any one of claims 1 to 11.

15. A computer program product, characterised in that, The computer program product includes computer instructions stored in a computer readable storage medium, and the processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to enable the computer device to implement the object information determination method according to any one of claims 1 to 11.

Images