Object screening method and device, electronic equipment, storage medium and program product

Abstract

The invention discloses an object screening method and device, electronic equipment, a storage medium and a program product, and relates to the technical field of the Internet, and the method comprises the steps: obtaining a target seed object corresponding to a first resource providing object, the target seed object is used for representing a resource obtaining object which obtains resources from the first resource providing object and / or other resource providing objects with the same resource type as that provided by the first resource providing object; based on the first feature of the target seed object, determining a feature similarity between second features of other resource acquisition objects in the content platform and the first feature; and screening a similar resource acquisition object from other resource acquisition objects of the content platform based on the feature similarity, and taking the similar resource acquisition object as a target resource acquisition object of the first resource providing object. And the target resource acquisition object is matched based on the features of the target seed object, so that the accuracy of the screened resource acquisition object is improved while the operation cost of the resource providing object is reduced.

Description

Technical Field

[0001] The present disclosure relates to the field of Internet technology, and in particular to an object screening method, device, electronic device, storage medium, and program product. Background Art

[0002] Relevant content platforms can accept resource advertisements placed by resource providers on the platform, so that the relevant resources of the resource providers can be seen by the corresponding resource acquisition targets on the content platform to increase the possibility of resource acquisition; at the same time, the content platform can also help resource providers screen out other potential resource acquisition targets on the platform to enable them to push traffic to the resource providers, or help resource providers match with content creators on the content platform so that the matched content creators can promote the resources provided by the corresponding resource providers, etc.

[0003] In related technologies, in order to find potential resource acquisition objects, resource providers generally select tags that are suitable for resources or that the resource providers themselves are interested in from some fixed tags to express their own suitable or interested resource acquisition objects. This tag-based matching method for potential resource acquisition objects has a low degree of intelligence, increases the operating cost of the resource provider object, and the granularity of the set fixed tags is coarse, which affects the accuracy of the selected resource acquisition objects. Summary of the Invention

[0004] In view of this, the present disclosure provides an object screening method, device, electronic device, storage medium and program product to solve the problem that the existing method of matching potential resource acquisition objects based on tags has a low level of intelligence, increases the operating cost of resource provision objects, and the granularity of the set fixed tags is coarse, which affects the accuracy of the screened resource acquisition objects.

[0005] In a first aspect, the present disclosure provides an object screening method, comprising: obtaining a target seed object corresponding to a first resource providing object, the target seed object being used to characterize a resource acquisition object that has obtained resources from the first resource providing object and / or from other resource providing objects of the same resource type as that provided by the first resource providing object; based on a first feature of the target seed object, determining a feature similarity between a second feature of other resource acquisition objects in a content platform and the first feature; based on the feature similarity, screening out similar resource acquisition objects from other resource acquisition objects in the content platform, and using the similar resource acquisition objects as target resource acquisition objects of the first resource providing object.

[0006] In a second aspect, the present disclosure provides an object screening device, which includes: a first acquisition module, used to obtain a target seed object corresponding to a first resource providing object, wherein the target seed object is used to represent a resource acquisition object that has obtained resources from the first resource providing object and / or from other resource providing objects of the same resource type as the first resource providing object; a first determination module, used to determine the feature similarity between the second feature of other resource acquisition objects in the content platform and the first feature based on the first feature of the target seed object; a screening module, used to screen out similar resource acquisition objects from other resource acquisition objects in the content platform based on the feature similarity, and use the similar resource acquisition objects as the target resource acquisition objects of the first resource providing object.

[0007] In a third aspect, the present disclosure provides an electronic device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the object screening method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0008] In a fourth aspect, the present disclosure provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the object screening method of the first aspect or any corresponding embodiment thereof.

[0009] In a fifth aspect, the present disclosure provides a computer program product, comprising computer instructions for causing a computer to execute the object screening method of the first aspect or any corresponding embodiment thereof.

[0010] The object screening method provided by the embodiment of the present disclosure obtains a target seed object corresponding to a first resource providing object, so that the target seed object that has obtained resources can understand the first feature of the object interested in the resource provided by the first resource providing object, and based on the first feature, finds a similar resource acquisition object corresponding to a second feature similar to the first feature from the content platform, and uses the similar resource acquisition object as a potential target resource acquisition object of the first resource providing object that may also have a resource acquisition request; compared with the method of finding resource acquisition objects by label matching, the solution provided by the present disclosure can directly match potential target resource acquisition objects based on the features of the target seed object that has obtained resources, thereby improving the degree of intelligence, reducing the operating cost of the resource providing object, and combining the object feature matching method to further improve the accuracy of the screened resource acquisition objects. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the specific embodiments of the present disclosure or the technical solutions in the related technologies, the following briefly introduces the drawings required for use in the specific embodiments or related technical descriptions. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0012] Figure 1 is a flow chart of an object screening method according to an embodiment of the present disclosure;

[0013] Figure 2 is a flowchart of another object screening method according to an embodiment of the present disclosure;

[0014] Figure 3A is a schematic diagram corresponding to the object screening method according to an embodiment of the present disclosure;

[0015] Figure 3B is a schematic diagram corresponding to the object screening method according to an embodiment of the present disclosure;

[0016] Figure 4 is a structural block diagram of an object screening device according to an embodiment of the present disclosure;

[0017] Figure 5 Schematic diagram of the hardware structure of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0018] To make the purpose, technical solutions, and advantages of the embodiments of the present disclosure more clear, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present disclosure.

[0019] It is understandable that before using the technical solutions disclosed in the various embodiments of this disclosure, the type, scope of use, usage scenarios, etc. of the personal information involved in this disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0020] For example, in response to a user's active request, a prompt message is sent to the user to clearly inform the user that the operation requested will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the electronic device, application, server, storage medium, or other software or hardware that performs the operations of the disclosed technical solution based on the prompt message.

[0021] As an optional but non-limiting implementation, in response to receiving a user's active request, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. Furthermore, the pop-up window may also contain a selection control for the user to select "agree" or "disagree" to provide personal information to the electronic device.

[0022] It is understandable that the above notification and user authorization process are merely illustrative and do not limit the implementation of the present disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0023] It is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) must comply with the requirements of relevant laws, regulations and relevant provisions.

[0024] The object screening method provided in this embodiment can be applied to various types of related content platforms, such as video content platforms, graphic content platforms, etc. The resource provision objects can be merchants who place product advertisements on the corresponding content platforms, the target seed objects can be platform users who have purchased products from merchants, and the resource acquisition objects can be users on the content platform who are interested in the merchant's products or have consumption needs.

[0025] Taking the video content platform as an example, by placing product advertisements with product purchase links on the video content platform, the merchant's related products can be seen by users of the video content platform to increase the possibility of the products being purchased; at the same time, the video content platform can also help merchants screen out potential users on the platform who may have purchasing needs for the merchant's products to promote traffic to the merchants, or help merchants match with video content creators on the video content platform. By encouraging merchants to cooperate with video creators, the matched video content creators can promote the merchant's products to their video viewing users, etc.

[0026] In related technologies, in order to find potential users who may have consumption needs for merchants on video content platforms, merchants generally select tags that are suitable for products or that merchants are interested in from some fixed tags (such as the user's geographic location, user identity information, etc.) to express their own suitable or interested user groups. This tag-based matching method for potential consumer user groups has a low degree of intelligence, increases the merchant's operating costs, and the fixed tags set have a coarse granularity, which affects the accuracy of the screened consumer user groups.

[0027] According to an embodiment of the present disclosure, an object screening method embodiment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0028] For ease of understanding, in the following embodiments, the content platform is a video content platform, the resource providers are merchants that have placed advertisements on the video content platform, the target seed objects are platform users who have purchased corresponding products from the merchants, and the target resource acquisition objects can be other users of the video content platform who may have purchasing needs. All merchant data or user data used in the following embodiments is obtained after authorization.

[0029] In this embodiment, an object screening method is provided, which can be used on any electronic device capable of executing the method, such as a server. Figure 1 is a flow chart of an object screening method according to an embodiment of the present disclosure, such as Figure 1 As shown, the process includes the following steps:

[0030] Step S101: Acquire a target seed object corresponding to a first resource providing object. The target seed object is used to represent a resource acquisition object that has acquired resources from the first resource providing object and / or from other resource providing objects of the same resource type as the first resource providing object.

[0031] For example, other resource providers with the same resource type as the first resource provider can be selected from resource providers belonging to the same industry as the first resource provider. For example, for the commodity "mobile phone", the first resource provider is the merchant that sells mobile phones, and other resource providers with the same resource type are other merchants that have sold mobile phones on the content platform, or merchants that have sold mobile phones on similar content platforms, to ensure that merchants of new resource types on the content platform can also be matched with the target seed object; in order to select more similar other resource providers, the selected other merchants can be further screened based on parameters such as product model.

[0032] The target seed object corresponding to the first resource provider object can be obtained by obtaining the historical sales records of the first resource provider object and / or other resource providers of the same type. The corresponding historical sales records can be maintained in the platform background after authorization and can be directly retrieved from the platform background when in use. The resource acquisition object can be directly obtained from the first resource provider object as the target seed object, or the resource acquisition object obtained from other resource providers of the same type can be directly used as the target seed object, or the resource acquisition object obtained from both can be used as the target seed object at the same time. The embodiment of the present application does not limit this. Those skilled in the art can match the corresponding target seed object to the current first resource provider object as needed.

[0033] In the embodiment of the present application, merchants can dynamically send object screening demand instructions to the video content platform according to certain sales volume index requirements based on the sales volume data of their own products on the video content platform. The video content platform can perform corresponding object screening operations upon receiving the corresponding instructions; or the video content platform can regularly help merchants screen potential consumers at certain intervals or regularly match merchants with recommended information on video content creators they can cooperate with. The embodiment of the present application does not limit this.

[0034] Step S102: Based on the first feature of the target seed object, determine the feature similarity between the second feature and the first feature of other resource acquisition objects in the content platform.

[0035] Exemplarily, the first feature of the target seed object and the second feature of other resource acquisition objects can be the portrait data feature corresponding to the portrait data of the user authorized by the user (such as the user's identity identification information, etc.) and the consumption data feature corresponding to the historical consumption data of the corresponding user (such as what type of goods have been purchased, the time of purchase, etc.). The method for calculating the feature similarity between the first feature and the second feature is not limited. The first feature and the second feature can be converted into vectors, and the feature similarity calculated based on the distance between the vectors, such as using the Euclidean distance calculation method to calculate the feature similarity between the second feature and the first feature. The embodiment of the present application does not limit the feature similarity calculation method. Those skilled in the art can select the corresponding feature similarity calculation method according to actual needs.

[0036] Step S103 : Based on feature similarity, similar resource acquisition objects are screened out from other resource acquisition objects on the content platform, and the similar resource acquisition objects are used as target resource acquisition objects of the first resource providing object.

[0037] Exemplarily, based on the feature similarity calculation result between the first feature and the second feature, other resource acquisition objects corresponding to the second feature whose similarity with the first feature is greater than the target threshold can be used as similar resource acquisition objects similar to the target seed object, and the similar resource acquisition object can be used as the target resource acquisition object of the first resource providing object; that is, the potential consumer users of the merchant's products on the video content platform can be obtained through the similarity calculation result, and based on the obtained potential consumer users of the merchant, the video content platform can push the merchant's products to the potential consumer users.

[0038] In some optional embodiments, the method further includes: performing a similarity matching operation on the object characteristics of the target resource acquisition object and the object characteristics of the content acquisition object corresponding to the content creator in the content platform; and screening out the content creator that matches the first resource provision object based on the similarity matching operation results.

[0039] Exemplarily, for the content acquisition object corresponding to the content creator in the content platform, taking the video content platform as an example, the content acquisition object corresponding to the video content creator can be the user who follows the video content creator on the video content platform, that is, the video content released by the video content creator on the video content platform can be viewed by the user audience of the video content creator through push or active search. The object characteristics of the audience user corresponding to the video content creator on the content platform are obtained. The user characteristics of the potential user and the user characteristics of the audience user contain the same feature types corresponding to the first feature and the second feature mentioned above, which will not be repeated here. The user characteristics of the potential user and the user characteristics of the audience user are calculated for similarity, and the potential user and the audience user are matched for similarity based on the similarity calculation result. The embodiment of the present application does not limit the similarity calculation method. Based on the similarity matching result, the content creator corresponding to the audience user whose similarity meets the preset requirements can be selected as the content creator matched with the merchant. By screening out video content creators who match merchants, video content creators can promote merchant products in their video content so that they can be seen by the video content creator's audience users in a targeted manner, thereby increasing the possibility of merchant products being purchased.

[0040] The object screening method provided by this embodiment obtains the target seed object corresponding to the first resource providing object, so that the target seed object that has obtained the resource can understand the first feature of the object interested in the resource provided by the first resource providing object, and based on the first feature, finds a similar resource acquisition object corresponding to a second feature similar to the first feature from the content platform, and uses the similar resource acquisition object as a potential target resource acquisition object of the first resource providing object that may also have a resource acquisition request; compared with the method of finding resource acquisition objects by label matching, the solution provided by the present disclosure can directly match potential target resource acquisition objects based on the features of the target seed object that has obtained the resource, thereby improving the degree of intelligence, reducing the operating cost of the resource providing object, and combining the object feature matching method to further improve the accuracy of the screened resource acquisition objects.

[0041] In this embodiment, an object screening method is provided, where a target seed object is used to represent a resource acquisition object that has acquired resources from a first resource providing object or from other resource providing objects of the same resource type as that provided by the first resource providing object. Figure 2 is a flow chart of an object screening method according to an embodiment of the present disclosure, such as Figure 2 As shown, the process includes the following steps:

[0042] Step S201: Acquire a target seed object corresponding to a first resource providing object.

[0043] Specifically, the above step S201 includes:

[0044] Step S2011: Determine whether the number of resource acquisition objects that have acquired resources from the first resource providing object meets the object screening requirement.

[0045] For example, taking the example of a merchant placing a product advertisement with a purchase link on a video content platform, the resource acquisition object that has obtained resources from the first resource provider object can be obtained from the merchant's historical sales records. In order to further avoid the same user purchasing products multiple times and affecting the accuracy of the determined number of resource acquisition objects, the purchasing users in the historical sales records can be deduplicated, and the number of purchasing users after deduplication can be used as the number of users who have purchased products from the merchant. The object screening requirements for the number of resource acquisition objects that have obtained resources can be determined by those skilled in the art based on actual needs, such as based on the accuracy of the screening results or the object characteristics that can accurately characterize the resource acquisition objects that have obtained resources. In the embodiment of the present application, the object screening requirement can be that the number of resource acquisition objects that have obtained resources is at least greater than 100.

[0046] In step S2012, if the number of resource acquisition objects meets the object screening requirement, the resource acquisition objects that have acquired resources from the first resource providing object are used as target seed objects. That is, when the number of resource acquisition objects determined is greater than 100, the resource acquisition objects that have acquired resources from the first resource providing object are directly used as target seed objects.

[0047] In step S2013, if the quantity does not meet the object screening requirements, a resource acquisition object that has previously acquired resources from other resource provision objects of the same resource type as the first resource provision object is used as the target seed object. For example, if the number of consumer users included in the merchant's own historical sales records does not meet the quantity requirement in the object screening requirements, the corresponding consumer users are acquired from the historical sales records of other merchants that sell the same type of products as the merchant.

[0048] Step S202: Based on the first feature of the target seed object, determine the similarity between the second feature and the first feature of other resource acquisition objects in the content platform; see Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0049] Step S203: Based on the feature similarity, similar resource acquisition objects are screened from other resource acquisition objects on the content platform, and the similar resource acquisition objects are used as the target resource acquisition objects of the first resource providing object. Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.

[0050] The object screening method provided in this embodiment, in the process of obtaining target seed objects, first obtains the target seed objects from the merchant's own historical sales records. When the number of target seed objects corresponding to the merchant itself does not meet the object screening requirements, the target seed objects are then obtained from the historical sales records of other merchants with the same type of goods sold by the merchant; that is, the first preferred source of target seed objects is the merchant itself, which improves the accuracy of subsequent screening of potential consumer users of the merchant. At the same time, for merchants that lack or have no historical sales records, the data in the sales records of other merchants in the same industry can be further combined to screen out potential consumer users of the current merchant, thereby improving the universality of the method of the embodiment of this application and the accuracy of the object screening results.

[0051] In some optional implementations, the above-mentioned step S103 includes: based on feature similarity, using at least one type of similarity screening method to perform similar resource acquisition object screening operations from other resource acquisition objects of the content platform, and obtain similar resource acquisition object screening results corresponding to each type of similarity screening method; preprocessing the obtained similar resource acquisition object screening results to obtain similar resource acquisition objects.

[0052] Exemplarily, based on the obtained feature similarity between each first feature and the second feature, one or more similarity screening methods can be used to screen out similar resource acquisition objects from other resource acquisition objects on the content platform. For example, other resource acquisition objects can be screened according to a screening method in which the feature similarity is greater than a target threshold requirement; or multiple feature similarities can be sorted in descending order, and other resource acquisition objects can be screened according to a screening method in which the feature similarities with the top target number are selected to obtain similar resource acquisition object screening results corresponding to the corresponding type of similarity screening method.

[0053] The preprocessing method can be to merge the screening results of multiple similar resource acquisition objects; or based on the screening results, re-screen the similar resource acquisition objects contained in the screening results, such as clustering according to the portrait feature data of the similar resource acquisition objects, and further combine the clustering results to select the similar resource acquisition objects corresponding to the clustering results that are the same as the target seed user as the similar resource acquisition objects.

[0054] In some optional implementations, the similarity screening methods include multiple types; based on feature similarity, multiple types of similarity screening methods are used to perform similar resource acquisition object screening operations from other resource acquisition objects on the content platform, and similar resource acquisition object screening results corresponding to each similarity screening method are obtained, including:

[0055] Obtain feature similarity between each second feature and the first feature corresponding to any target seed object; based on the obtained multiple feature similarities corresponding to any target seed object, select an initial similar resource acquisition object from other resource acquisition objects whose feature similarity with any target seed object meets the requirements.

[0056] Exemplarily, for the second features B1, B2 and B3 and the first feature A1 of the target seed object TA1, the first feature A2 of the target seed object TA2 and the first feature A3 of the target seed object TA3, the feature similarity C1 of the second feature B1 and the first feature A1, the feature similarity C2 of the second feature B2 and the first feature A1, and the feature similarity C3 of the third feature B3 and the first feature A1, the feature similarity C4 of the second feature B1 and the first feature A2, the feature similarity C5 of the second feature B2 and the first feature A2, and the feature similarity C6 of the second feature B3 and the first feature A2, the feature similarity C7 of the second feature B1 and the first feature A3, the feature similarity C8 of the second feature B2 and the first feature A3, and the feature similarity C9 of the second feature B3 and the first feature A3 are obtained.

[0057] For the multiple feature similarities corresponding to the first feature A1 of the target seed object TA1, C1, C2 and C3, determine whether C1, C2 and C3 meet the similarity requirements. If C1 and C2 meet the requirements, then the other resource acquisition objects corresponding to the second feature B1 corresponding to C1 and the other resource acquisition objects corresponding to the second feature B2 corresponding to C2 are used as the initial similar resource acquisition object CS1 of the target seed object TA1; similarly, the initial similar resource acquisition object CS2 corresponding to the target seed object TA2 and the initial similar resource acquisition object CS3 corresponding to the target seed object TA3 can be obtained.

[0058] The feature similarities of the initial similar resource acquisition objects corresponding to all the obtained target seed objects are sorted.

[0059] For example, taking the example in the previous step as an example, if the feature similarities corresponding to the initial similar resource acquisition object CS2 of the target seed object TA2 are C4 and C5, and the feature similarities corresponding to the initial similar resource acquisition object CS3 of the target seed object TA3 are C7 and C8, then the obtained feature similarities C1, C2, C4, C5, C7 and C8 will be sorted; the specific sorting method can be from large to small or from small to large, which is not limited in the embodiment of the present application.

[0060] Based on the sorting results, the initial similar resource acquisition objects corresponding to the first target quantity or the feature similarity whose feature similarity is greater than the first similarity threshold are selected, and the selected initial similar resource acquisition objects are used as the similar resource acquisition object screening results corresponding to one type of similarity screening method among multiple types of similarity screening methods.

[0061] For example, the embodiment of the present application takes the order from large to small as an example. If the sorting result is C1>C2>C4>C5>C7>C8, the initial similar resource acquisition objects corresponding to the first N feature similarities are selected. The embodiment of the present application does not limit the number N. It can be determined according to the number of sorting. For example, if the number of feature similarities participating in the sorting is 1000, the initial similar resource acquisition objects corresponding to the first 100 most similar feature similarities can be selected; or the initial similar resource acquisition objects corresponding to the feature similarities greater than the first similarity threshold are selected based on the sorting result, and the initial similar resource acquisition objects obtained by the similarity screening method corresponding to the above steps are used as the similar resource acquisition object screening result corresponding to one type of similarity screening method among multiple types of similarity screening methods.

[0062] Determine the feature similarity between any other resource acquisition object and each target seed object and sum the obtained multiple feature similarities to obtain the overall feature similarity between any other resource acquisition object and all target seed objects; and sort the obtained overall feature similarities corresponding to each other resource acquisition object.

[0063] For example, taking the second feature B1 of other resource acquisition object QT1, the second feature B2 of other resource acquisition object QT2, the second feature B3 of other resource acquisition object QT3, and the first feature A1 of target seed object TA1, the first feature A2 of target seed object TA2, and the first feature A3 of target seed object TA3 in the previous embodiment as examples, determine the feature similarity D1 between the second feature B1 and the first feature A1, the feature similarity D2 between the second feature B1 and the first feature A2, and the feature similarity D3 between the second feature B1 and the first feature A3, the feature similarity D4 between the second feature B2 and the first feature A1, the feature similarity D5 between the second feature B2 and the first feature A2, the feature similarity D6 between the second feature B2 and the first feature A3, the feature similarity D7 between the second feature B3 and the first feature A1, the feature similarity D8 between the second feature B3 and the first feature A2, and the feature similarity D9 between the second feature B3 and the first feature A3.

[0064] The feature similarities between any other resource acquisition object and each target seed object are summed up to obtain the overall feature similarities between any other resource acquisition object and all target seed objects.

[0065] For example, taking the previous step as an example, the overall feature similarity E1 corresponding to the other resource acquisition object QT1 is the sum of feature similarity D1, feature similarity D2 and feature similarity D3, the overall feature similarity E2 corresponding to the other resource acquisition object QT2 is the sum of feature similarity D4, feature similarity D5 and feature similarity D6, and the overall feature similarity E3 corresponding to the other resource acquisition object QT3 is the sum of feature similarity D7, feature similarity D8 and feature similarity D9; the obtained E1, E2 and E3 are sorted, and the specific sorting method can be from large to small or from small to large, which is not limited in the embodiments of the present application.

[0066] Based on the sorting results, other resource acquisition objects corresponding to the second target number or the overall feature similarity whose overall feature similarity is greater than the second similarity threshold are selected, and the selected other resource acquisition objects are used as the similar resource acquisition object screening results corresponding to another type of similarity screening method among the multiple types of similarity screening methods.

[0067] For example, the embodiment of the present application takes the order from large to small as an example. If the sorting result is E1>E2>E3, then the other resource acquisition objects corresponding to the top K overall feature similarities are selected. The embodiment of the present application does not limit the number K, and those skilled in the art can make adaptive adjustments based on the screened order of magnitude; or select other resource acquisition objects corresponding to the overall feature similarity greater than the second similarity threshold based on the sorting result, and use the other resource acquisition objects screened out by the similarity screening method corresponding to the above steps as the similar resource acquisition object screening results corresponding to another type of similarity screening method among the multiple types of similarity screening methods.

[0068] In some optional implementations, the obtained similar resource acquisition object screening results are preprocessed to obtain similar resource acquisition objects, including: if the obtained multiple similar resource acquisition object screening results contain the same resource acquisition object, then the same resource acquisition objects contained in the multiple similar resource acquisition object screening results are deduplicated, and the similar resource acquisition objects remaining after the deduplication processing are used as similar resource acquisition objects.

[0069] Exemplarily, the deduplication processing method may be to use any similar resource acquisition object screening result as a benchmark, and compare each similar resource acquisition object in the similar resource acquisition object screening result with the similar resource acquisition objects contained in other similar resource acquisition object screening results to deduplicate the resource acquisition objects in other similar resource acquisition object screening results that are identical to the similar resource acquisition object screening result currently serving as the benchmark; and so on, deduplicate the other similar resource acquisition object screening results. Deduplication processing can reduce the problem of repeated processing of the same data affecting the efficiency of object screening.

[0070] In some optional embodiments, the method further includes: if the order of magnitude of the obtained similar resource acquisition objects exceeds a preset order of magnitude requirement, clustering the similar resource acquisition objects to obtain a first clustering result and clustering the target seed objects to obtain a second clustering result.

[0071] For example, in the embodiments of the present application, the magnitude of the corresponding order of magnitude in the preset order of magnitude requirement is not limited, and those skilled in the art can determine it according to actual needs. For example, if the magnitude of the corresponding order of magnitude of the preset order of magnitude requirement is in the millions, if the magnitude of the obtained similar resource acquisition objects exceeds the millions, the obtained similar resource acquisition objects can be clustered according to a preset clustering algorithm, such as the K-means mean clustering method, to obtain the corresponding first clustering result and the second clustering result by using the clustering method.

[0072] A matching operation is performed on the first clustering result and the second clustering result, and a similar resource acquisition object corresponding to a category whose matching degree meets a preset requirement is selected from the first clustering result as a similar resource acquisition object.

[0073] Exemplarily, vector representation can be performed based on the similar resource acquisition objects contained in the first clustering result and the target seed objects contained in the second clustering result, and the degree of matching can be determined by calculating the distance between different clustering results based on the vector representation result, such as using Euclidean distance or Manhattan distance, etc.; similar resource acquisition objects corresponding to categories whose distances meet preset requirements are selected from the first clustering result as the filtered similar resource acquisition objects.

[0074] In some optional embodiments, step S102 includes: obtaining a first feature vector corresponding to each first feature and a second feature vector corresponding to each second feature; performing vector similarity calculation on each second feature vector and each first feature vector based on a preset vector similarity calculation method and using the calculated vector similarity as the feature similarity of the corresponding first feature and the second feature.

[0075] For example, the method for generating a feature vector in the embodiment of the present application may be any method that can realize feature vector generation, and the embodiment of the present application is not limited to this, such as using a bag-of-words model or a method based on word embedding. The preset vector similarity calculation method may include a cosine similarity method, etc., and the embodiment of the present application is not limited to this. The vector similarity between each second feature vector and each first feature vector is calculated by the preset vector similarity calculation method, and the calculated vector similarity is used as the feature similarity between the corresponding first feature and the second feature.

[0076] In some optional embodiments, the method further comprises:

[0077] Step a1: Acquire resource acquisition object related data and resource portrait data on the content platform. The resource acquisition object related data includes resource acquisition object portrait data and historical resource acquisition related behavior data.

[0078] For example, taking the video content platform as an example, the resource acquisition object portrait data may include the basic portrait data of users on the video content platform, such as user identity identification information; the historical resource acquisition related behavior data may include data on users seeing products on the video content platform and performing acquisition related behaviors, and acquisition related behaviors include users clicking on product purchase links, users clicking to understand product details, users adding products to shopping carts, etc. For products that users have already purchased, it also includes the name of the product purchased by the user and the industry type to which the product belongs; and data on users seeing products on the video content platform but not performing acquisition related behaviors.

[0079] A user's basic profile data can be obtained after the user's authorization by obtaining the user's registration data and / or the video content data that the user is interested in on the video content platform; the user's historical consumption data can be obtained by filtering from a relevant consumption database based on the user's account information after the user's authorization. Resource profile data may include product feature data on the video content platform, including but not limited to product industry information, product price level, and suitable purchasing population. In the embodiment of the present application, product feature data can be obtained from the product description data provided by the merchant when placing product advertisements on the video content platform.

[0080] Step a2: construct a first model training sample including positive samples and negative samples based on the resource acquisition object related data, and construct a second model training sample based on the resource portrait data. The positive samples include samples in which the resource acquisition object sees the resource and performs acquisition-related behaviors, and the negative samples include samples in which the resource acquisition object sees the resource but does not perform acquisition-related behaviors.

[0081] Step a3: input the first model training sample into the first model and train it, and input the second model training sample into the second model and train it, until the similarity between the feature vector of the resource acquisition object generated by the first model and the feature vector of the resource generated by the second model meets the preset requirements, and the first model is used to generate the first feature vector and the second feature vector.

[0082] Exemplarily, the first model may include an input layer and a multi-layer neural network. Samples in which users see products and perform acquisition-related behaviors are used as positive samples, and samples in which users see products but do not perform acquisition-related behaviors are used as negative samples to construct first model training samples and train the first model. Taking the video content platform as an example, the input first model training samples are converted into vector representations through processing of the input layer. The neural network layer is used to perform nonlinear transformation and feature extraction on the input user-related vectors. Through processing of the neural network layer, the model can automatically learn the correlation between the features of the user-related vectors, and adjust the parameters of the neural network according to algorithms such as back propagation, so that the model can better capture the user's behavior model and preferences.

[0083] The product portrait data is used as the training sample of the second model to train the second model. The model structure of the second model is the same as that of the first model. The product portrait data is used as the input of the second model and converted into a vector representation through the embedding layer. During the training process, the model parameters of the second model can be continuously adjusted so that the output product feature vector can more accurately express the attributes and characteristics of the product. The similarity of the feature vectors output by the first model and the second model is calculated and the parameters of the first model and the second model are continuously adjusted until the user feature vector output by the first model and the product vector output by the second model have a similarity relationship that meets the requirements in the feature space. Specifically, the loss function can be used to determine the correlation between the outputs of the two models. The Sigmoid function can map the original output of the model to the (0, 1) interval to represent the probability of the two inputs being related. For example, in the object screening process, it can represent the probability that the user is interested in a certain product. That is, through the association training of the first model and the second model, the model can be prepared to express the recommendation probability between the user and the corresponding product, so that for any new user, the product type that the user may be interested in can be given based on the user's basic portrait data and recommended.

[0084] When the training is completed, the first model can be used to generate the first feature vector and the second feature vector required for screening the target resource acquisition object. It only needs to input the corresponding first feature or second feature into the first model. The second model can be used in other scenarios that require the generation of product feature vectors. The corresponding model structure can be found in Figure 3A shown.

[0085] In some optional embodiments, the method further comprises:

[0086] Obtain resource acquisition behavior data of resource acquisition objects on the content platform and construct a bipartite graph based on the resource acquisition behavior data. The nodes of the bipartite graph are composed of resource acquisition object nodes and resource nodes. The acquisition behavior between the resource acquisition object and the resource constitutes the undirected edge of the bipartite graph. For example, taking the purchase of goods by users on the video content platform as an example, the bipartite graph is constructed with users and goods as nodes and the purchase behavior between users and goods as undirected edges. For details, see Figure 3B shown.

[0087] A walk operation of target length is performed with each of all nodes as a starting point to obtain multiple node sequences of target lengths; for example, the embodiment of the present application does not limit the target length, and can be adaptively adjusted according to the number of nodes, such as a walk operation with a target length of 10. For ease of explanation, the embodiment of the present application takes the target length of 5 as an example to illustrate the walk method. Figure 3BTaking "user 1" in the example as the starting point and performing a walk operation with a target length of 5, the node sequence obtained is "user 1-item 1-user 3-item 3-user 1-item 2". By analogy, multiple node sequences with a target length of 5 can be obtained.

[0088] For any node sequence, any two nodes contained in a preset size window constitute positive samples, and any other two nodes in the node sequence constitute negative samples; illustratively, the window size can be selected according to actual needs, such as taking a window size of 3 as an example, for the above node sequence "user 1-item 1-user 3-item 3-user 1-item 2", for the node "user 1-item 1-user 3" corresponding to the window of 3, its positive samples include: (user 1, item 1), (user 1, user 3) and (item 1, user 3), in addition, any other two nodes constitute negative samples; taking this as an example, positive and negative samples corresponding to each node sequence can be obtained, and the negative samples represent that there is no close correlation between the nodes, so as to help the model learn which node pairs are irrelevant, thereby better distinguishing the relationship between different nodes.

[0089] The positive samples and negative samples corresponding to all node sequences are input into the preset vector generation model until the training results meet the requirements to obtain the vector of each node and filter out the feature vector corresponding to the resource acquisition object node for storage operation. The first feature vector and the second feature vector are obtained from the stored results.

[0090] For example, the embodiments of the present application do not limit the preset vector generation model, and those skilled in the art can select a model that can generate vectors according to actual needs, such as a word embedding model (Skip-gram, etc.). The positive samples and negative samples corresponding to all node sequences are input into the preset vector generation model to implement model training, so that the model can better capture the association information between user nodes and commodity nodes based on the input positive samples and negative samples; specifically, the parameters of the model can be adjusted by minimizing the loss function, such as the cross entropy loss function, so that the model can better distinguish between positive samples and negative samples. During the training process, the model will learn the vector representation of each node, and the generated vector can represent the semantic information of the node in the bipartite graph and the association relationship with other nodes. When the training is completed, the feature vector corresponding to each user node is filtered and stored. When there is a need for object screening, the feature vector of the corresponding user is directly obtained from the stored results.

[0091] As a specific application embodiment of the present application, for a video content platform, merchants can place product advertisements on the video content platform. In order to increase the chances of the products being seen by more users of the video content platform, the video content platform can perform similarity matching of the user characteristics of the target user group corresponding to the merchant's products with the user characteristics of the viewing user group of the content creators of the video content platform. For merchants and content creators with relatively similar user characteristics, the video content platform can bring the two together to achieve business cooperation.

[0092] Specifically, the video content platform obtains user data that has purchased goods from the merchant or from other merchants in the same industry as the merchant, and searches for other similar users on the video content platform based on the user feature vector of the user and a preset similar user calculation method to achieve diffusion processing of the merchant's target user group; the user features of the diffused merchant's target user group are matched with the user features of the video viewing audience of the video content creator on the video content platform for similarity, and the video content creators with whom the merchant is recommended to cooperate are determined based on the matching results. Compared with the method of searching through fixed tags, the method provided by the embodiment of the present application for screening video content creators who can achieve business cooperation with merchants on the platform combines user features for matching to make the results of the video content creators found more accurate, making it easier for merchants to screen out video content creators that meet adaptation needs at a lower cost.

[0093] In this embodiment, an object screening device is also provided, which is used to implement the above-mentioned embodiments and preferred embodiments. The details already described will not be repeated here. As used below, the term "module" can mean a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0094] This embodiment provides an object screening device, such as Figure 4 As shown, including:

[0095] A first acquisition module 401 is configured to acquire a target seed object corresponding to a first resource providing object, where the target seed object represents a resource acquisition object that has acquired resources from the first resource providing object and / or from other resource providing objects of the same resource type as the first resource providing object.

[0096] A first determining module 402 is configured to determine, based on the first feature of the target seed object, a feature similarity between a second feature of other resource acquisition objects in the content platform and the first feature;

[0097] The first screening module 403 is configured to screen out similar resource acquisition objects from other resource acquisition objects on the content platform based on feature similarity, and use the similar resource acquisition objects as target resource acquisition objects of the first resource providing object.

[0098] The object screening device provided in this embodiment obtains the target seed object corresponding to the first resource providing object, so that the target seed object that has obtained the resource can understand the first feature of the object interested in the resource provided by the first resource providing object, and based on the first feature, finds a similar resource acquisition object corresponding to a second feature similar to the first feature from the content platform, and uses the similar resource acquisition object as a potential target resource acquisition object of the first resource providing object that may also have a resource acquisition request; compared with the method of finding resource acquisition objects by label matching, the solution provided by the present disclosure can directly match potential target resource acquisition objects based on the features of the target seed object that has obtained the resource, thereby improving the degree of intelligence, reducing the operating cost of the resource providing object, and combining the object feature matching method to further improve the accuracy of the screened resource acquisition objects.

[0099] In some optional embodiments, the target seed object is used to represent a resource acquisition object that has obtained resources from the first resource providing object or from other resource providing objects of the same resource type as the first resource providing object; the first acquisition module 401 includes: a determination submodule, used to determine whether the number of resource acquisition objects that have obtained resources from the first resource providing object meets the object screening requirements; a first acquisition submodule, used to use the resource acquisition objects that have obtained resources from the first resource providing object as the target seed object if the number meets the object screening requirements; and a second acquisition submodule, used to use the resource acquisition objects that have obtained resources from other resource providing objects of the same resource type as the first resource providing object as the target seed object if the number does not meet the object screening requirements.

[0100] In some optional embodiments, the first screening module 403 includes: a third acquisition sub-module, which is used to perform similar resource acquisition object screening operations from other resource acquisition objects of the content platform based on feature similarity using at least one type of similarity screening method to obtain similar resource acquisition object screening results corresponding to each type of similarity screening method; and a preprocessing sub-module, which is used to preprocess the obtained similar resource acquisition object screening results to obtain similar resource acquisition objects.

[0101] In some optional embodiments, the similarity screening method includes multiple types; the third acquisition submodule includes: a first acquisition unit, which is used to obtain the feature similarity between each second feature and the first feature corresponding to any target seed object; a first selection unit, which is used to select an initial similar resource acquisition object whose feature similarity with any target seed object meets the requirements from other resource acquisition objects based on the multiple feature similarities corresponding to any target seed object obtained; a first sorting unit, which is used to sort the feature similarities of the initial similar resource acquisition objects corresponding to all the target seed objects obtained; a second selection unit, which is used to select the initial similar resource acquisition objects corresponding to the first target number or the feature similarity whose feature similarity is greater than the first similarity threshold based on the sorting result, and use the selected initial similar resource acquisition objects as multiple types a first determining unit for determining the feature similarity between any other resource acquisition object and each target seed object and summing the obtained multiple feature similarities to obtain the overall feature similarity between any other resource acquisition object and all target seed objects; a second sorting unit for sorting the obtained overall feature similarity corresponding to each other resource acquisition object; a third selecting unit for selecting other resource acquisition objects corresponding to a second target number or an overall feature similarity whose overall feature similarity is greater than a second similarity threshold based on the sorting result, and using the selected other resource acquisition objects as the similar resource acquisition object screening result corresponding to another type of similarity screening method among the multiple types of similarity screening methods.

[0102] In some optional embodiments, the preprocessing submodule includes: a deduplication unit, which is used to deduplicate the same resource acquisition objects contained in the multiple similar resource acquisition object screening results if the same resource acquisition objects are included in the multiple similar resource acquisition object screening results, and use the similar resource acquisition objects remaining after the deduplication processing as similar resource acquisition objects.

[0103] In some optional embodiments, the device also includes: a clustering module, which is used to perform a clustering operation on the similar resource acquisition objects to obtain a first clustering result and to perform a clustering operation on the target seed objects to obtain a second clustering result if the order of magnitude of the similar resource acquisition objects obtained exceeds a preset order of magnitude requirement; a first matching module, which is used to perform a matching operation on the first clustering result and the second clustering result, and select from the first clustering result the similar resource acquisition objects corresponding to the categories whose matching degree meets the preset requirements as the similar resource acquisition objects.

[0104] In some optional embodiments, the device also includes: a second matching module, used to perform similarity matching operations on the object characteristics of the target resource acquisition object and the object characteristics of the content acquisition object corresponding to the content creator in the same content platform; a second screening module, used to screen out content creators that match the first resource provision object based on the similarity matching operation results.

[0105] In some optional embodiments, the first determination module 402 includes: a first acquisition submodule, used to obtain a first feature vector corresponding to each first feature and a second feature vector corresponding to each second feature; a calculation submodule, used to perform vector similarity calculation on each second feature vector and each first feature vector based on a preset vector similarity calculation method and use the calculated vector similarity as the feature similarity of the corresponding first feature and the second feature.

[0106] In some optional embodiments, the device also includes: a second acquisition module, used to acquire resource acquisition object related data and resource portrait data on the content platform, the resource acquisition object related data including resource acquisition object portrait data and historical resource acquisition related behavior data; a first sample construction module, used to construct a first model training sample including positive samples and negative samples based on the resource acquisition object related data and to construct a second model training sample based on the resource portrait data, the positive sample including a sample in which the resource acquisition object sees the resource and performs acquisition related behavior, and the negative sample including a sample in which the resource acquisition object sees the resource but does not perform acquisition related behavior; a first training module, used to input the first model training sample into the first model and perform training and input the second model training sample into the second model and perform training until the similarity between the feature vector of the resource acquisition object generated by the first model and the feature vector of the resource generated by the second model meets the preset requirements, and the first model is used to generate the first feature vector and the second feature vector.

[0107] In some optional embodiments, the device also includes: a third acquisition module, which is used to obtain resource acquisition behavior data of resource acquisition objects on the content platform and construct a bipartite graph based on the resource acquisition behavior data, wherein the nodes of the bipartite graph are composed of resource acquisition object nodes and resource nodes, and the acquisition behavior between the resource acquisition object and the resource constitutes an undirected edge of the bipartite graph; a walk module, which is used to perform a walk operation of a target length with each of all nodes as a starting point to obtain multiple node sequences of target lengths; a second sample construction module, which is used to, for any node sequence, constitute positive samples with any two nodes contained in a preset size window, and constitute negative samples with any other two nodes in the node sequence; a second training module, which is used to input the positive samples and negative samples corresponding to all node sequences into a preset vector generation model until the training results meet the requirements to obtain the vector of each node and filter out the feature vector corresponding to the resource acquisition object node for storage operation, and the first feature vector and the second feature vector are obtained from the stored results.

[0108] The object screening device provided in the embodiments of the present disclosure can execute the object screening method provided in any embodiment of the present disclosure, and has the functional modules and beneficial effects corresponding to the execution method. The further functional description of each of the above modules and units is the same as that of the corresponding embodiments above, and will not be repeated here.

[0109] Figure 5 A schematic structural diagram of an electronic device provided in an embodiment of the present disclosure.

[0110] The following specific reference Figure 5 , which shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiments of the present disclosure. The electronic device may include a processor (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a memory 508 into a random access memory (RAM) 503. Various programs and data required for the operation of the electronic device are also stored in the RAM 503. The processor 501, ROM 502, and RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0111] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although Figure 5An electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all of the devices shown, and more or fewer devices may be implemented or possessed instead.

[0112] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network via the communication device 509, or installed from the memory 508, or installed from the ROM 502. When the computer program is executed by the processor 501, the above-mentioned functions defined in the object screening method of the embodiment of the present disclosure are performed.

[0113] Figure 5 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0114] The embodiments of the present disclosure also provide a computer-readable storage medium. The above-mentioned method according to the embodiments of the present disclosure can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the object screening method shown in the above embodiment is implemented.

[0115] A portion of the present disclosure may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present disclosure through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes but is not limited to a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0116] Although the embodiments of the present disclosure have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present disclosure, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for object screening, characterized in that: The method comprises: Obtaining a target seed object corresponding to the first resource providing object, where the target seed object is used to represent a resource acquisition object that has acquired resources from the first resource providing object and / or from other resource providing objects of the same resource type as that provided by the first resource providing object; Based on the first feature of the target seed object, determining a feature similarity between a second feature of other resource acquisition objects in the content platform and the first feature; Based on the feature similarity, similar resource acquisition objects are screened out from other resource acquisition objects on the content platform, and the similar resource acquisition objects are used as target resource acquisition objects of the first resource providing object.

2. The method according to claim 1, characterized in that The target seed object is used to represent a resource acquisition object that has acquired resources from the first resource providing object or from other resource providing objects of the same resource type as that provided by the first resource providing object; The obtaining of the target seed object corresponding to the first resource providing object includes: determining whether the number of resource acquisition objects that have acquired resources from the first resource providing object meets the object screening requirement; If the quantity meets the object screening requirement, the resource acquisition object that has acquired resources from the first resource providing object is used as the target seed object; If the quantity does not meet the object screening requirement, a resource acquisition object that has acquired resources from other resource providing objects of the same resource type as that provided by the first resource providing object is used as the target seed object.

3. The method according to claim 1, characterized in that The screening of similar resource acquisition objects from other resource acquisition objects on the content platform based on the feature similarity includes: Based on the feature similarity, using at least one type of similarity screening method to perform a similar resource acquisition object screening operation from other resource acquisition objects of the content platform, and obtaining similar resource acquisition object screening results corresponding to each type of similarity screening method; The obtained similar resource acquisition object screening result is preprocessed to obtain the similar resource acquisition object.

4. The method according to claim 3, characterized in that The similarity screening method includes multiple types; based on the feature similarity, a similar resource acquisition object screening operation is performed from other resource acquisition objects of the content platform using multiple types of similarity screening methods, and a similar resource acquisition object screening result corresponding to each similarity screening method is obtained, including: Obtaining a feature similarity between each of the second features and the first feature corresponding to any target seed object; Based on the obtained multiple feature similarities corresponding to any target seed object, an initial similar resource acquisition object whose feature similarity with any target seed object meets the requirements is selected from the other resource acquisition objects; Sort the feature similarities of the initial similar resource acquisition objects corresponding to all the target seed objects obtained; Selecting initial similar resource acquisition objects corresponding to a first target number or feature similarity greater than a first similarity threshold based on the sorting result, and using the selected initial similar resource acquisition objects as similar resource acquisition object screening results corresponding to one type of similarity screening method among the multiple types of similarity screening methods; Determine the feature similarity between any other resource acquisition object and each target seed object and sum the obtained multiple feature similarities to obtain the overall feature similarity between any other resource acquisition object and all target seed objects; Sort the overall feature similarity corresponding to each other resource acquisition object obtained; Based on the sorting results, other resource acquisition objects corresponding to the second target number or the overall feature similarity whose overall feature similarity is greater than the second similarity threshold are selected, and the selected other resource acquisition objects are used as the similar resource acquisition object screening results corresponding to another type of similarity screening method among the multiple types of similarity screening methods.

5. The method according to claim 4, characterized in that The preprocessing of the obtained similar resource acquisition object screening result to obtain the similar resource acquisition object includes: If the obtained multiple similar resource acquisition object screening results include the same resource acquisition object, deduplication processing is performed on the same resource acquisition objects included in the multiple similar resource acquisition object screening results, and the similar resource acquisition objects remaining after deduplication processing are used as the similar resource acquisition objects.

6. The method according to claim 1 or 2, characterized in that The method further comprises: If the magnitude of the obtained similar resource acquisition objects exceeds the preset magnitude requirement, clustering the similar resource acquisition objects to obtain a first clustering result and clustering the target seed objects to obtain a second clustering result; A matching operation is performed on the first clustering result and the second clustering result, and a similar resource acquisition object corresponding to a category whose matching degree meets a preset requirement is selected from the first clustering result as the similar resource acquisition object.

7. The method according to claim 1 or 2, characterized in that The method further comprises: Performing a similarity matching operation on the object features of the target resource acquisition object and the object features of the content acquisition object corresponding to the content creator in the same content platform; Based on the similarity matching operation result, content creators matching the first resource providing object are screened out.

8. The method according to claim 1, characterized in that The determining of the similarity between the second feature of other resource acquisition objects in the content platform and the first feature includes: Obtaining a first eigenvector corresponding to each first feature and a second eigenvector corresponding to each second feature; Based on a preset vector similarity calculation method, vector similarity calculation is performed on each of the second feature vectors and each of the first feature vectors, and the calculated vector similarity is used as the feature similarity of the corresponding first feature and the second feature.

9. The method according to claim 8, characterized in that The method further comprises: Acquire resource acquisition object related data and resource portrait data on the content platform, wherein the resource acquisition object related data includes resource acquisition object portrait data and historical resource acquisition related behavior data; Constructing a first model training sample comprising positive samples and negative samples based on the resource acquisition object related data, and constructing a second model training sample based on the resource portrait data, wherein the positive samples include samples in which the resource acquisition object sees the resource and performs acquisition-related behaviors, and the negative samples include samples in which the resource acquisition object sees the resource but does not perform acquisition-related behaviors; The first model training sample is input into the first model and trained, and the second model training sample is input into the second model and trained, until the similarity between the feature vector of the resource acquisition object generated by the first model and the feature vector of the resource generated by the second model meets the preset requirements, and the first model is used to generate the first feature vector and the second feature vector.

10. The method according to claim 8, characterized in that The method further comprises: Acquire resource acquisition behavior data of resource acquisition objects on the content platform and construct a bipartite graph based on the resource acquisition behavior data, wherein the nodes of the bipartite graph are composed of resource acquisition object nodes and resource nodes, and the acquisition behaviors between the resource acquisition objects and the resources constitute undirected edges of the bipartite graph; Taking each of the nodes as the starting point, a walk operation of the target length is performed to obtain multiple node sequences of the target length; For any node sequence, any two nodes contained in the preset size window constitute positive samples, and any other two nodes in the node sequence constitute negative samples; The positive samples and negative samples corresponding to all node sequences are input into the preset vector generation model until the training results meet the requirements to obtain the vector of each node and filter out the feature vector corresponding to the resource acquisition object node for storage operation, and the first feature vector and the second feature vector are obtained from the storage results.

11. An object screening device, characterized in that: The device comprises: A first acquisition module is configured to acquire a target seed object corresponding to a first resource providing object, wherein the target seed object is used to represent a resource acquisition object that has acquired resources from the first resource providing object and / or from other resource providing objects of the same resource type as that provided by the first resource providing object; A first determining module is configured to determine, based on the first feature of the target seed object, a feature similarity between a second feature of other resource acquisition objects in the content platform and the first feature; A screening module is used to screen out similar resource acquisition objects from other resource acquisition objects of the content platform based on the feature similarity, and use the similar resource acquisition objects as target resource acquisition objects of the first resource providing object.

12. An electronic device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the object screening method according to any one of claims 1 to 10 by executing the computer instructions.

13. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the object screening method according to any one of claims 1 to 10.

14. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the object screening method according to any one of claims 1 to 10.

Images