Advertisement delivery method, device, equipment and storage medium

Abstract

The present disclosure provides an advertisement delivery method, device, equipment and storage medium, and relates to the field of e-commerce technology. The method includes: receiving an advertisement playback request; obtaining a candidate advertisement product set according to the advertisement playback request; obtaining the advertisement price attribute information of the first candidate advertisement product in the candidate advertisement product set; obtaining the second candidate advertisement product based on the advertisement price attribute information of the first candidate advertisement product; obtaining the estimated click-through rate of the first candidate advertisement product and the estimated click-through rate of the second candidate advertisement product; obtaining a replacement reference lower limit value according to the estimated click-through rate of the first candidate advertisement product; when the estimated click-through rate of the second candidate advertisement product is higher than the replacement reference lower limit value, replacing the first candidate advertisement product with the second candidate advertisement product, and obtaining an updated candidate advertisement product set for advertisement delivery. The method improves the rationality of advertisement delivery to a certain extent.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to an advertisement delivery method, device, equipment, and readable storage medium. Background Art

[0002] The rapid development of the Internet has changed the world. From portals to search engines, from social networks to e-commerce, the endless stream of online services has made people's lives more convenient. Most of the online services provided by Internet companies (such as portals, e-commerce platforms, short video applications, etc.) are free for users and rely on selling advertising space to advertisers to generate revenue. Internet companies provide valuable online products to users, accumulate a certain number of users and reach a certain level of user activity to generate traffic, and then advertisers place advertisements through the advertising space of the online product and pay according to the advertising effect.

[0003] The number of products that can be displayed in an ad space is limited, and there may be a large number of products to be displayed in the ad space. In the related art, when determining candidate recommended products from a large number of products to be displayed based on multiple factors, it is difficult to reasonably allocate traffic based on the advertising bids of the products, resulting in low rationality of advertising delivery.

[0004] As mentioned above, how to improve the rationality of advertising has become an urgent problem to be solved.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present disclosure is to provide an advertisement delivery method, apparatus, device and readable storage medium, which can improve the rationality of advertisement delivery at least to a certain extent.

[0007] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0008] According to one aspect of the present disclosure, there is provided an advertisement delivery method, comprising: receiving an advertisement playback request; obtaining a set of candidate advertisement products according to the advertisement playback request; obtaining advertisement price attribute information of a first candidate advertisement product in the set of candidate advertisement products; obtaining a second candidate advertisement product based on the advertisement price attribute information of the first candidate advertisement product; obtaining an estimated click-through rate of the first candidate advertisement product and an estimated click-through rate of the second candidate advertisement product; obtaining a replacement reference lower limit value according to the estimated click-through rate of the first candidate advertisement product; and replacing the first candidate advertisement product with the second candidate advertisement product when the estimated click-through rate of the second candidate advertisement product is higher than the replacement reference lower limit value, to obtain an updated set of candidate advertisement products for advertisement delivery.

[0009] According to an embodiment of the present disclosure, the updated set of candidate advertising products includes unreplaced candidate advertising products and replacement candidate advertising products; replacing the first candidate advertising product with the second candidate advertising product includes: replacing the first candidate advertising product with the second candidate advertising product to obtain the replacement candidate advertising product; when the estimated click-through rate of the first candidate advertising product is higher than the estimated click-through rate of the second candidate advertising product, obtaining the estimated click-through rate of the replacement candidate advertising product as the estimated click-through rate of the first candidate advertising product; obtaining the updated set of candidate advertising products for advertising delivery includes: sorting the unreplaced candidate advertising products and the replacement candidate advertising products according to the advertising price attribute information and the estimated click-through rate of the unreplaced candidate advertising products and the advertising price attribute information and the estimated click-through rate of the replacement candidate advertising products, so as to deliver the advertisements in sequence.

[0010] According to an embodiment of the present disclosure, the sorting of the unreplaced candidate advertising products and the replacement candidate advertising products according to the advertising price attribute information and the estimated click-through rate of the unreplaced candidate advertising products and the advertising price attribute information and the estimated click-through rate of the replacement candidate advertising products includes: obtaining the sorting index of the unreplaced candidate advertising products according to the advertising price attribute information and the estimated click-through rate of the unreplaced candidate advertising products; obtaining the sorting index of the replacement candidate advertising products according to the advertising price attribute information and the estimated click-through rate of the replacement candidate advertising products; sorting the unreplaced candidate advertising products and the replacement candidate advertising products in descending order of the sorting index to obtain a sequence of advertising products to be placed; and placing advertisements in sequence includes: placing advertisements in the sequence of advertising products to be placed from front to back.

[0011] According to an embodiment of the present disclosure, the advertisement playback request includes a user identifier; obtaining a set of candidate advertisement products according to the advertisement playback request includes: obtaining the set of candidate advertisement products through a collaborative filtering algorithm based on the user identifier, and arranging the first candidate advertisement products in the set of candidate advertisement products from high to low according to an estimated satisfaction index; obtaining a second candidate advertisement product based on the advertisement price attribute information of the first candidate advertisement product includes: selecting a first predetermined number of first candidate advertisement products from the first candidate advertisement products arranged according to the estimated satisfaction index; obtaining the second candidate advertisement product whose advertisement price attribute information is higher than the advertisement price attribute information of the first predetermined number of first candidate advertisement products.

[0012] According to an embodiment of the present disclosure, the method further includes: obtaining historical advertising bids for goods belonging to a predetermined category; determining a reserved advertising bid for goods in the predetermined category based on the historical advertising bids; obtaining real-time advertising bids for goods belonging to the predetermined category; selecting goods whose real-time advertising bids are higher than the reserved advertising bids to obtain a set of high-bid goods in the predetermined category; the obtaining of a second candidate advertising product based on the advertising price attribute information of the first candidate advertising product includes: obtaining the predetermined category to which the first candidate advertising product belongs; obtaining the second candidate advertising product whose real-time advertising bid is higher than the advertising bid of the first candidate advertising product from the set of high-bid goods in the predetermined category.

[0013] According to an embodiment of the present disclosure, determining the reserved advertising bid for the product in the predetermined category based on the historical advertising bids includes: obtaining the price distribution of the historical advertising bids; and selecting the price of a predetermined quantile of the price distribution as the reserved advertising bid.

[0014] According to an embodiment of the present disclosure, obtaining a replacement reference lower limit value according to the estimated click rate of the first candidate advertisement product includes: multiplying the estimated click rate of the first candidate advertisement product by a predetermined percentage to obtain the replacement reference lower limit value.

[0015] According to another aspect of the present disclosure, there is provided an advertisement delivery device, comprising: a request receiving module for receiving an advertisement playing request; a product filtering module for acquiring a set of candidate advertisement products according to the advertisement playing request; an information acquiring module for acquiring advertisement price attribute information of a first candidate advertisement product in the set of candidate advertisement products; a product supplementing module for acquiring a second candidate advertisement product based on the advertisement price attribute information of the first candidate advertisement product; a click rate estimating module for acquiring an estimated click rate of the first candidate advertisement product and an estimated click rate of the second candidate advertisement product; a replacement preparing module for acquiring a replacement reference lower limit value according to the estimated click rate of the first candidate advertisement product; and a product recommendation module for replacing the first candidate advertisement product with the second candidate advertisement product when the estimated click rate of the second candidate advertisement product is higher than the replacement reference lower limit value, so as to obtain an updated set of candidate advertisement products for advertisement delivery.

[0016] According to an embodiment of the present disclosure, the updated set of candidate advertising products includes unreplaced candidate advertising products and replacement candidate advertising products; the product recommendation module includes: a replacement execution module, used to replace the candidate advertising product with the second candidate advertising product to obtain the replacement candidate advertising product; the replacement execution module is also used to obtain the estimated click-through rate of the replacement candidate advertising product as the estimated click-through rate of the candidate advertising product when the estimated click-through rate of the candidate advertising product is higher than the estimated click-through rate of the second candidate advertising product; a product sorting module is used to sort the unreplaced candidate advertising products and the replacement candidate advertising products according to the advertising price attribute information and the estimated click-through rate of the unreplaced candidate advertising products and the advertising price attribute information and the estimated click-through rate of the replacement candidate advertising products, so as to deliver advertisements in sequence.

[0017] According to an embodiment of the present disclosure, the product ranking module includes: a user recommendation degree calculation module, which is used to obtain the ranking index of the unreplaced candidate advertising products based on the advertising price attribute information and the estimated click-through rate of the unreplaced candidate advertising products; the user recommendation degree calculation module is also used to obtain the ranking index of the replacement candidate advertising products based on the advertising price attribute information and the estimated click-through rate of the replacement candidate advertising products; a module for obtaining a sequence of products to be recommended, which is used to sort the unreplaced candidate advertising products and the replacement candidate advertising products in order of the ranking index from large to small, and obtain a sequence of advertising products to be placed; the product recommendation module is also used to place advertisements in the sequence of advertising products to be placed from front to back.

[0018] According to an embodiment of the present disclosure, the advertisement playback request includes a user identifier; the product filtering module includes: a front algorithm module, which is used to obtain the candidate advertising product set through a collaborative filtering algorithm based on the user identifier, and the first candidate advertising products in the candidate advertising product set are arranged from high to low according to the estimated satisfaction index; the product complement module is also used to select the first candidate advertising products that are ranked in the front predetermined number among the first candidate advertising products arranged according to the estimated satisfaction index; and obtain the second candidate advertising products whose advertising price attribute information is higher than the advertising price attribute information of the first candidate advertising products that are ranked in the front predetermined number.

[0019] According to an embodiment of the present disclosure, the device further includes: a historical data acquisition module, which is used to acquire historical advertising bids of products belonging to a predetermined category; a reserve price determination module, which is used to determine the reserve advertising bids of products in the predetermined category based on the historical advertising bids; a real-time data acquisition module, which is used to acquire real-time advertising bids of products belonging to the predetermined category; a high-bid product set acquisition module, which is used to select products whose real-time advertising bids are higher than the reserve advertising bids, and obtain a high-bid product set of the predetermined category; the product supplement module includes: a category query module, which acquires the predetermined category to which the first candidate advertising product belongs; and a supplement product acquisition module, which is used to acquire the second candidate advertising product whose real-time advertising bid is higher than the advertising bid of the first candidate advertising product from the high-bid product set of the predetermined category.

[0020] According to an embodiment of the present disclosure, the reserve price determination module is further used to obtain the price distribution of the historical advertisement bids; and select the price of a predetermined quantile of the price distribution as the reserve advertisement bid.

[0021] According to an embodiment of the present disclosure, the click rate estimation module is further configured to obtain the replacement reference lower limit value by multiplying the estimated click rate of the first candidate advertisement product by a predetermined percentage.

[0022] According to another aspect of the present disclosure, a device is provided, comprising: a memory, a processor, and executable instructions stored in the memory and executable in the processor, wherein the processor implements any of the above methods when executing the executable instructions.

[0023] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which computer-executable instructions are stored, and when the executable instructions are executed by a processor, any of the above methods is implemented.

[0024] The product recommendation method provided by the embodiment of the present disclosure obtains a set of candidate advertising products according to a received advertisement playback request, obtains a second candidate advertising product based on the advertisement price attribute information of the first candidate advertising product in the set of candidate advertising products, obtains a replacement reference lower limit value according to the estimated click-through rate of the first candidate advertising product, and then replaces the first candidate advertising product with the second candidate advertising product when the estimated click-through rate of the second candidate advertising product is higher than the replacement reference lower limit value, and obtains an updated set of candidate advertising products for advertisement delivery, thereby achieving the replacement of the first candidate advertising product in the set of candidate advertising products with the second candidate advertising product obtained based on the advertisement bid, thereby improving the rationality of advertisement delivery to a certain extent.

[0025] It is to be understood that the foregoing general description and the following detailed description are exemplary only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The above and other objects, features and advantages of the present disclosure will become more apparent by describing in detail example embodiments thereof with reference to the attached drawings.

[0027] Figure 1 A schematic diagram showing a system structure in an embodiment of the present disclosure.

[0028] Figure 2 A flow chart of an advertisement delivery method in an embodiment of the present disclosure is shown.

[0029] Figure 3 Shows Figure 2 FIG. 5 is a schematic diagram of the processing process of step S212 in one embodiment.

[0030] Figure 4A The present invention is a flowchart of another advertisement delivery method according to an exemplary embodiment.

[0031] Figure 4B A schematic diagram of a selection process of a second candidate advertisement product is shown.

[0032] Figure 4C A schematic diagram of a replacement strategy for a second candidate advertisement product is shown.

[0033] Figure 5 is based on Figure 2 FIG4 is a schematic diagram of an advertisement delivery process.

[0034] Figure 6 The present invention is a flowchart of a method for obtaining a set of high-bid commodities according to an exemplary embodiment.

[0035] Figure 7 is based on Figure 6 A schematic diagram of a process for obtaining a set of high-bid commodities is shown.

[0036] Figure 8 The invention is a block diagram of an advertisement delivery device according to an exemplary embodiment.

[0037] Fig. 9 The invention is a block diagram of an advertisement delivery device according to an exemplary embodiment.

[0038] Fig.10 A schematic structural diagram of an electronic device in an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0039] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the disclosure will be more comprehensive and complete and will fully convey the concepts of the example embodiments to those skilled in the art. The accompanying drawings are only schematic illustrations of the disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted.

[0040] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, devices, steps, etc. may be adopted. In other cases, well-known structures, methods, devices, implementations or operations are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0041] In the description of the present disclosure, "plurality" means at least two, such as two, three, etc., unless otherwise clearly and specifically limited. The symbol " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0042] In this disclosure, unless otherwise clearly specified and limited, terms such as "connection" should be understood in a broad sense, for example, it can be an electrical connection or can communicate with each other; it can be directly connected or indirectly connected through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in this disclosure can be understood according to specific circumstances.

[0043] As mentioned above, Internet companies can earn revenue through Internet advertising. Based on the transaction method between Internet companies and advertisers, Internet advertising mainly includes the following types of advertising.

[0044] Contract advertising: Contract Internet advertising has evolved from traditional advertising. Advertisers can use the Internet platform as a media similar to television, newspapers, etc. to place banner ads on web pages. The transaction method is similar to the traditional advertising method. The Internet company and the advertiser negotiate to determine the time for the advertiser to monopolize a certain position on the page and the price the advertiser pays the Internet company.

[0045] Guaranteed targeted advertising: With the development of Internet advertising, the exclusive page method has gradually developed into an on-demand distribution method, for example: placing razor ads for male users and facial mask ads for female users. When placing this type of advertising, traffic can be labeled, such as gender, region, interests, etc. Advertisers purchase corresponding advertising space according to the tags. On the one hand, the media's traffic is sold in a refined manner, and on the other hand, advertisers can also buy traffic for the target population of their own products. At this time, the price mechanism of advertising can be negotiated pricing. Advertisers purchase traffic by selecting tags. Different tags have different traffic sizes. Advertisers will put forward traffic requirements. If the requirements are not met, the Internet company will compensate the advertiser according to the negotiated price.

[0046] Real-Time Bid (RTB): For guaranteed advertising, Internet companies need to solve problems such as traffic estimation and allocation, which is a heavy workload. The emergence of real-time bidding has reduced the burden on Internet companies. Advertisers set targeted traffic and set prices for the targeted traffic. For a certain targeted traffic, advertisers who meet the predetermined conditions bid, and the Internet company determines the ownership of the targeted traffic based on factors such as the advertiser's bid. The sales method of RTB advertising, that is, the bid categories of advertisers, mainly include: Cost Per Mille (CPM), Cost Per Click (CPC), and Cost Per Action (CPA), etc. The "action" in Cost Per Action can be understood as the advertiser's product transaction, order placement, application (App) installation, App activation, etc., which can also be called conversion.

[0047] When Internet companies select recommended products from a large number of advertised products, they may consider the advertiser's advertising bid factor, and in order to maintain stable traffic, they may also consider indicators that reflect the user's experience with the advertised products. In some cases, advertisers' advertising bids may change, such as increasing the advertising bid to increase product exposure when new products are launched, or lowering the advertising bid to save advertising costs when the advertising budget is reduced. In the related art, when determining candidate recommended products based on advertising bids and user experience indicators, changes in advertising bids may have little impact on the determination of candidate recommended products. Therefore, the present disclosure provides an advertisement delivery method, which obtains a set of candidate advertisement products according to a received advertisement playback request, obtains a second candidate advertisement product based on the advertisement price attribute information of the first candidate advertisement product in the candidate advertisement product set, obtains a replacement reference lower limit value according to the estimated click-through rate of the first candidate advertisement product, and then replaces the first candidate advertisement product with the second candidate advertisement product when the estimated click-through rate of the second candidate advertisement product is higher than the replacement reference lower limit value, and obtains an updated set of candidate advertisement products for advertisement delivery, thereby achieving the replacement of the first candidate advertisement product in the candidate advertisement product set with the second candidate advertisement product obtained based on the advertisement bid, thereby improving the influence of the advertisement bid on the determination of the candidate advertisement delivery set to a certain extent, and improving the rationality of advertisement delivery.

[0048] Figure 1 An exemplary system architecture 10 to which the advertisement delivery method or advertisement delivery apparatus of the present disclosure may be applied is shown.

[0049] like Figure 1 As shown, the system architecture 10 may include a terminal device 102, a network 104 and a server 106. The terminal device 102 may be any electronic device having a display screen and supporting web browsing, including but not limited to a smart phone, a tablet computer, a laptop computer, a desktop computer, etc. The network 104 is used to provide a medium for a communication link between the terminal device 102 and the server 106. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc. The server 106 may be a server, a server cluster, etc. that provides various services.

[0050] The user can use the terminal device 102 to interact with the server 106 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 102, such as shopping applications, social applications, etc. For example, when a user uses a shopping application on the terminal device 102, the shopping application can obtain the user's historical shopping information from the server 106 through the network 104, and then deliver advertisements based on the user's historical shopping information.

[0051] The server 106 may be, for example, a backend management server that supports shopping websites browsed by the user using the terminal device 102. The backend management server may analyze and process data such as the user browsing products and clicking advertisements on the website of the terminal device 102 received through the network 104, and feed back information such as advertisement placement to the terminal device.

[0052] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0053] Figure 2 FIG. 1 is a flow chart showing an advertisement delivery method according to an exemplary embodiment. Figure 2 The method shown can be applied to the server side of the above system, for example, and can also be applied to the terminal device of the above system.

[0054] refer to Figure 2 , the method 20 provided in the embodiment of the present disclosure may include the following steps.

[0055] In step S202, an advertisement playing request is received.

[0056] In step S204, a candidate advertising product set is obtained according to the advertisement playing request. When obtaining the candidate advertising products, the candidate advertising products to be recommended to the user can be screened from the massive products in the database according to the recommendation object (also called user, such as the user of the online product of an Internet company).

[0057] In some embodiments, for example, the advertisement delivery server may obtain a candidate advertising product set from a database server through a pre-recommendation algorithm based on the user's historical behavior. The pre-recommendation algorithm may be used to mine products that the user may be interested in from a collection of hundreds of millions of products based on a collaborative filtering method based on the user's historical click, add-to-cart, purchase, query, and other behaviors.

[0058] In other embodiments, for example, the advertisement delivery client may also obtain the candidate advertisement product set from the database server through a pre-recommendation algorithm according to the user's historical behavior.

[0059] In other embodiments, for example, a deep learning model may also be used for personalized recommendations, by training a neural network through a large number of users' historical clicks, add-to-cart, purchases, queries, and other behaviors, so that the neural network can output products that the user may be interested in based on the input user's historical behavior characteristics.

[0060] In step S206, the advertisement price attribute information of the first candidate advertisement product in the candidate advertisement product set is obtained. When the advertiser places an advertisement for the candidate advertisement product in the advertisement delivery system of the Internet company, the advertisement bid and other price attribute information of the product will be set, and the advertisement delivery system will save the advertisement bid; the advertiser can also modify the advertisement bid as needed, and obtain the advertisement bid of the candidate advertisement product through the advertisement delivery system.

[0061] In step S208, a second candidate advertisement product is obtained based on the advertisement price attribute information of the first candidate advertisement product. The candidate advertisement product set includes candidate advertisement products obtained by preliminary screening from a large number of products. A second candidate advertisement product (hereinafter referred to as a "complementary product") with a higher or lower advertisement bid than the candidate advertisement product can be selected from the advertisement product database according to actual needs.

[0062] In some embodiments, for example, if it is necessary to obtain complementary products with higher advertising bids than the candidate advertising products, a high-bid product set can be generated at a predetermined actual interval through an offline system, and complementary products with higher advertising bids than the candidate advertising products can be selected from the high-bid product set to reduce the amount of search calculations and improve system performance. The method for obtaining the high-bid product set can refer to Figure 6 The architecture of the offline system for generating a set of high-bid items can be found in Figure 7 , which will not be elaborated here.

[0063] In step S210, the estimated click-through rate of the first candidate advertising product and the estimated click-through rate of the second candidate advertising product are obtained. Regardless of whether the Internet company adopts CPM, CPC or CPA billing, the final revenue from traffic monetization can be uniformly measured by eCPM (effective Cost Per Mile, revenue per thousand impressions). CPC billing has become the mainstream in current Internet advertising. When CPC billing is adopted, the Internet will charge advertisers only when users trigger the behavior of clicking on advertisements. Therefore, the advertising revenue of Internet companies is not only related to the advertiser's bid (ie CPC), but also to the click-through rate (CTR) of the advertisement. The expected income E for each impression is:

[0064] E=CTR×CPC (1)

[0065] In the formula, CTR can be used to represent user satisfaction, i.e., a satisfaction index. The higher the CTR, the more satisfied the user is and the better the user experience is. The click-through rate (CTR) estimation method of the advertised product can be used to estimate the click-through rate of the product, and the estimated click-through rate (PCTR) of each product can be obtained. A CTR estimation method based on a deep learning model can be used. For example, user behaviors (such as user browsing, adding to cart, purchasing, querying, and collecting), user portrait features (such as age, gender, education, and address), and product information (category, brand, and price, etc.) are input into a deep neural network. The network outputs the click-through rate of each user for each product, and the PCTR of each user for each product is obtained.

[0066] In step S212, a replacement reference lower limit value is obtained according to the estimated click-through rate of the first candidate advertisement product. In some embodiments, for example, the estimated click-through rate of the candidate advertisement product may be multiplied by a predetermined percentage to obtain the replacement reference lower limit value; in another example, the estimated click-through rate of the candidate advertisement product may be subtracted from a predetermined value to obtain the replacement reference lower limit value. The replacement reference lower limit value of the click-through rate is obtained so that the second candidate advertisement product for replacement can be selected later, so that the estimated click-through rate of the second candidate advertisement product will not be too low, so as to avoid recommending the second candidate advertisement product with too low an estimated click-through rate, resulting in a poor user experience.

[0067] In step S214, when the estimated click rate of the second candidate advertisement product is higher than the replacement reference lower limit, the first candidate advertisement product is replaced with the second candidate advertisement product to obtain an updated candidate advertisement product set for advertisement delivery. The estimated click rate of the supplementary product is higher than the lower limit of the click rate of the candidate recommendation product, and the estimated click rate of the supplementary product is not too low, indicating that the user acceptance of the supplementary product is not too poor. At this time, the corresponding candidate advertisement product in the candidate advertisement product set can be replaced with the supplementary product, so that the supplementary product obtained according to the advertisement bid is added to the candidate advertisement product.

[0068] In some embodiments, for example, after obtaining an updated set of candidate advertising products, the products may be sorted according to their estimated click-through rates and advertising price attribute information, for example, the ads may be sorted in descending order according to the PCTR×CPC result; then, according to the ad space, a preset number of products ranked in the front may be selected for recommendation. When placing advertisements, the top products may be displayed in a prominent position or at the front of the search results according to the ranking, for example, the first-ranked product may be displayed in the center of the advertising area, and the second- and third-ranked products may be displayed above and below the first-ranked product, respectively, and so on.

[0069] According to an embodiment of the present disclosure, an advertisement delivery method is provided, wherein a second candidate advertisement product is obtained based on advertisement price attribute information of a first candidate advertisement product in a set of candidate advertisement products, a replacement reference lower limit value is obtained based on an estimated click-through rate of the first candidate advertisement product, and when the estimated click-through rate of the second candidate advertisement product is higher than the replacement reference lower limit value, the first candidate advertisement product in the set of candidate advertisement products is replaced by the second candidate advertisement product, and an updated set of candidate advertisement products is obtained for advertisement delivery, thereby replacing the first candidate advertisement product in the set of candidate advertisement products with the second candidate advertisement product obtained based on the advertisement price attribute information, thereby increasing the degree of influence of the advertisement price attribute information on the determination of the set of candidate advertisement products, and thereby improving the accuracy of advertisement delivery to a certain extent.

[0070] Figure 3 Shows Figure 2 The processing process of step S214 in one embodiment is shown in FIG. Figure 3 As shown, in the embodiment of the present disclosure, the above step S214 may further include the following steps.

[0071] In step S2141, when the estimated click rate of the second candidate advertisement product is not higher than the replacement reference lower limit, the non-replaced candidate advertisement product is obtained as the candidate advertisement product. Multiple candidate advertisement products can be selected from the candidate advertisement product set for the second candidate advertisement, and the corresponding multiple second candidate advertisement products are obtained. Then, according to the replacement reference lower limit, it is determined whether to replace. When the estimated click rate of the second candidate advertisement product is not higher than the corresponding replacement reference lower limit, it means that the expected user satisfaction of the second candidate advertisement product is too low, and it is not added to the candidate advertisement product set, and the corresponding first candidate advertisement product is retained.

[0072] In step S2142, when the estimated click rate of the second candidate advertisement product is higher than the replacement reference lower limit, the first candidate advertisement product is replaced with the second candidate advertisement product to obtain the replacement candidate advertisement product. When the estimated click rate of the complementary product is higher than the corresponding replacement reference lower limit, it means that the expected user satisfaction of the complementary product is close to that of the corresponding candidate advertisement product. If the recommendation of the complementary product has little impact on user satisfaction, it can be used to replace the corresponding candidate advertisement product.

[0073] In step S2143, when the estimated click-through rate of the first candidate advertisement product is higher than the estimated click-through rate of the second candidate advertisement product, the estimated click-through rate of the replacement candidate advertisement product is obtained as the estimated click-through rate of the first candidate advertisement product. Since the replacement candidate advertisement product (complementary product) is obtained based on the original candidate advertisement product, it can be considered that the user's satisfaction with the two should be similar (the supplementary product may have fewer historical clicks, purchases, etc. due to reasons such as being new products, resulting in a lower estimated click-through rate obtained using the deep learning model), so the estimated click-through rate of the candidate advertisement product can be used as the estimated click-through rate of the replacement candidate advertisement product, so that when the candidate advertisement products are subsequently sorted, the influence of the advertisement bid of the replacement candidate advertisement product on its ranking is increased.

[0074] In step S2124, when the estimated click rate of the candidate advertisement product is not higher than the estimated click rate of the second candidate advertisement product, the estimated click rate of the replacement candidate advertisement product is obtained as the estimated click rate of the second candidate advertisement product. If the estimated click rate of the supplementary product is higher than the estimated click rate of the original candidate advertisement product, the estimated click rate of the supplementary product itself can be used, which increases the influence of the advertisement bid of the replacement candidate advertisement product on its ranking without affecting user satisfaction.

[0075] In step S2125, the unreplaced candidate advertising products and the replacement candidate advertising products are sorted according to the advertising price attribute information and estimated click-through rate of the unreplaced candidate advertising products and the advertising price attribute information and estimated click-through rate of the replacement candidate advertising products so as to place advertisements in order. The sorting index of the unreplaced candidate advertising products can be obtained according to the advertising price attribute information and estimated click-through rate of the unreplaced candidate advertising products, and the sorting index of the replacement candidate advertising products can be obtained according to the advertising price attribute information and estimated click-through rate of the replacement candidate advertising products. The unreplaced candidate advertising products and the replacement candidate advertising products are sorted in descending order according to the sorting index to obtain a sequence of products to be recommended, and then the products in the sequence of products to be recommended are recommended from front to back.

[0076] In some embodiments, for example, under CPC billing, when selecting advertisements to display to users, the Internet platform may select the advertisements based on the PCTR. t The ads are sorted in descending order based on the results of ×CPC, where t>1 and t is a positive integer, to increase the weight of PCTR and balance the impact of user experience and advertising revenue on ranking.

[0077] Internet platforms according to PCTR tWhen ×CPC ranks candidate ads, PCTR will play a leading role in the ranking results of candidate ads. Although this ranking mechanism ensures user relevance and improves user experience, it weakens the bidding effect of advertisers and affects advertisers' enthusiasm for increasing their ad bids. When advertisers are in urgent need of display volume, they will choose to increase their ad bids. However, when the t value is greater than 1, the ranking is insensitive to the ad bid. Even if the advertiser raises the ad bid, the products it places cannot get more display volume, resulting in a poor experience for advertisers in this ranking mechanism.

[0078] According to the method provided by the embodiment of the present disclosure, when the estimated click-through rate of the second candidate advertising product is higher than the replacement reference lower limit, the first candidate advertising product is replaced by the second candidate advertising product, and the larger value of the estimated click-through rate of the first candidate advertising product and the estimated click-through rate of the second candidate advertising product is selected as the estimated click-through rate of the replacement candidate advertising product, and the advertising ranking is determined together with its advertising price attribute information. This can improve the sensitivity of the ranking to the advertising price attribute information without affecting the satisfaction index, thereby enhancing the enthusiasm of advertisers to bid.

[0079] Figure 4A FIG. 1 is a flow chart showing another method for delivering advertisements according to an exemplary embodiment. Figure 4A The method shown can be applied to the server side of the above system, for example, and can also be applied to the terminal device of the above system.

[0080] refer to Figure 4A , the method 40 provided in the embodiment of the present disclosure may include the following steps.

[0081] In step S402, a set of candidate advertising products is obtained through a collaborative filtering algorithm based on the user identifier in the advertisement play request, and the first candidate advertising products in the set of candidate advertising products are arranged from high to low according to the estimated satisfaction index. A pre-recommendation algorithm can be used to mine candidate advertising products (also called parent products) that the user may be interested in from a set of hundreds of millions of products based on a collaborative filtering method according to the user's historical clicks, add-to-cart, purchase, query and other behaviors. For example, the recommendation value of each advertisement in the product set delivered to the user is calculated based on the collaborative filtering method, and then the products are sorted from high to low according to the recommendation value, and a preset number of products in the front are selected as the parent product queue, and the preset number can be, for example, 800, 1000 or 1200, etc.

[0082] In step S404, the advertisement price attribute information of the candidate advertisement products in the candidate recommendation set is obtained. The specific implementation method can refer to step S206, which will not be described in detail here.

[0083] In step S406, the candidate advertising products that are ranked first in the candidate advertising product set are selected. Although the complementary products of each parent product can be obtained in this step, and then the replacement strategy is executed for all parent products in the subsequent steps, the amount of calculation is large, and when finally recommending to the user, only a small number of products will be displayed due to page restrictions and other reasons. Therefore, the complementary products of the first predetermined number of parent products can be obtained to reduce the amount of calculation and improve system performance.

[0084] In step S408, a second candidate advertisement product is obtained whose advertisement bid is higher than the advertisement bid of the candidate advertisement products ranked in the first predetermined number. In the advertisement delivery system, multiple categories of products may be included, each product belongs to a unique category, and each category may include multiple products. For example, the categories may include mobile phones, computers, keyboards, mice, etc. When selecting a supplementary product with a higher advertisement bid, it may be selected from a set of products in the same category as the parent product, so that the estimated click-through rate of the supplementary product and the parent product is more comparable, so as not to affect user satisfaction after replacement. In addition, when selecting supplementary products, it may be ensured that there is a balance in the account of the advertiser of these selected supplementary products.

[0085] In some embodiments, for example, when supplementing products based on the advertising bids of the parent product queue, a preset number (such as 8, 10 or 12, etc.) of parent products are used as keywords. For each parent product, products in the same category and with a higher CPC are found in the category → high-bid product set mapping table as supplementary products. Figure 4B FIG. 1 shows a schematic diagram of a selection process for a complementary product. Figure 4B As shown, for parent product 1, parent product 2, parent product 3, ..., parent product 10 obtained in the data mining step, the algorithm association step is executed, and the complementary product 1, complementary product 2, complementary product 3, ..., complementary product 10 of the same category and with a higher CPC are obtained in the category → high-bid product set mapping table, and are put into the complementary queue to execute the complementary strategy. Among them, a random algorithm can be used to randomly select complementary products with a CPC higher than the parent product in the high-bid product set to avoid excessive exposure of the complementary product with the highest CPC when the complementary product is selected, resulting in excessive consumption of advertising costs.

[0086] In some embodiments, for example, the predetermined category to which the candidate advertisement product belongs may be first obtained, and a complementary product whose real-time advertisement bid is higher than the advertisement bid of the candidate advertisement product may be obtained from the set of high-bid products in the predetermined category. Figure 6 , which will not be elaborated here.

[0087] In step S410, the estimated click-through rate of the candidate advertisement product and the estimated click-through rate of the second candidate advertisement product are obtained based on the user identifier. The user identifier can be obtained according to the user request, and the user's click-through rate for each product (including the parent product and the complementary product) is estimated by the CTR estimation algorithm to obtain the PCTR of each product. A CTR estimation algorithm based on a deep learning model can be used. For example, the user behavior and user portrait features are obtained according to the user identifier, and the user behavior (such as user browsing, adding to cart, purchasing, querying and collecting, etc.), user portrait features (such as age, gender, education and address, etc.) and product information (category, brand and price, etc.) are input into the deep neural network. The network outputs the click-through rate of each user for each product, and the PCTR of each user for each product is obtained. At this time, the system can obtain the latest CPC of each parent product and complementary product for subsequent sorting steps.

[0088] In step S412, the estimated click rate of the candidate advertisement product is multiplied by a predetermined percentage to obtain a click rate lower limit. The predetermined percentage k of the PCTR of the candidate advertisement product can be preset, for example, k can be set to 5%, 10% or 15%, etc.

[0089] In step S414, when the estimated click rate of the second candidate advertisement product is higher than the replacement reference lower limit, the candidate advertisement product is replaced with the second candidate advertisement product to obtain an updated candidate advertisement product set for advertisement delivery. The PCTR value of the complementary product is compared with the PCTR value of its corresponding parent product. If the PCTR value is 补足商品 >PCTR 父商品 ×(1-k), the parent product in the parent product queue is replaced with the corresponding complementary product, and the parent product queue is updated; otherwise, it is not replaced. Under the premise that the parent product and the complementary product belong to the same category, it can be assumed that the user satisfaction with the two should be similar (the complementary product may have fewer historical clicks and purchases due to reasons such as being new, resulting in a lower PCTR obtained by the deep learning model), so the PCTR of the complementary product added to the parent product queue is 补足商品 Replaced by PCTR 父商品 Then all the products in the parent product queue at this time are sorted according to PCTR t The ×CPC values ​​are sorted from large to small, and finally a preset number (such as 5, 10 or 12, etc.) of advertisements are delivered to the user.

[0090] In some embodiments, for example, Figure 4C A schematic diagram of a replacement strategy for replenishing goods is shown. Figure 4CAs shown, when k=10%, the PCTR of complementary product 2=0.28>(0.3*90%=)0.27, so the parent product 2 in the parent product queue is replaced with complementary product 2, and the PCTR of the replaced complementary product 2 is replaced with 3; the PCTR of complementary product 3=0.25>(0.2*90%=)0.18, so the parent product 3 in the parent product queue is replaced with complementary product 3, and 0.25>0.2, so the PCTR of the replaced complementary product 3 is retained as 0.25; the PCTR of complementary product 10=0.1>(0.11*90%=)0.099, so the parent product 10 in the parent product queue is replaced with complementary product 10, and the PCTR of the replaced complementary product 10 is replaced with 0.11.

[0091] According to the method provided by the embodiment of the present disclosure, the parent product mined by the algorithm is replaced with a high-bid complementary product under the same category. On the premise of ensuring that the PCTR of the complementary product is not too low to affect user satisfaction, a higher-bid complementary product is used to replace the parent product to display products with higher bids, thereby improving the advertising system's sensitivity to bids when recommending products; and a predetermined number of complementary products of the parent product are obtained to reduce the amount of calculation and improve system performance.

[0092] Figure 5 is based on Figure 2 FIG4 is a schematic diagram of an advertisement delivery process. Advertisements can be placed through the online system (50) to obtain user requests in real time (S502); a pre-recommendation algorithm is used to mine parent products that the user may be interested in from a collection of hundreds of millions of products based on a collaborative filtering method according to the user's historical clicks, add-to-cart, purchases, inquiries, and other behaviors (S504); a preset number of parent products (such as 8, 10 or 12, etc.) are used as keywords, and products in the same category and with higher CPC are found in the category→high-bid product set mapping table (5002) as complementary products to obtain a candidate product queue (S506); then the click rate of each product (including the parent product and the complementary product) is estimated through the CTR estimation algorithm to obtain the PCTR of each product. At this time, the system can obtain the latest CPC of each parent product and complementary product (S508); then a replacement strategy is executed (S510): a PCTR ratio lower limit k is set, for example, k can be set to 5%, 10% or 15%, that is, the PCTR value of the complementary product is compared with the PCTR of its corresponding parent product. If the PCTR 补足商品 >PCTR 父商品 ×(1-k), then replace the parent product in the parent product queue with the corresponding complementary product and update the parent product queue, otherwise do not replace it; then all the products in the parent product queue at this time are calculated according to PCTR tThe ×CPC values ​​are sorted from large to small, and finally a preset number of advertisements ranked first are delivered to the user (S512).

[0093] Figure 6 FIG. 1 is a flow chart showing a method for obtaining a set of high-bid commodities according to an exemplary embodiment. Figure 6 The method shown can be applied to the server side of the above system, for example, and can also be applied to the terminal device of the above system.

[0094] refer to Figure 6 , the method 60 provided in the embodiment of the present disclosure may include the following steps.

[0095] In step S602, historical advertising price attribute information of the goods belonging to the predetermined category is obtained. The mapping relationship from goods to categories can be obtained according to the goods attribute information library, and the historical bid list of the goods can be obtained from the historical advertising transaction information library of the advertising delivery system. The goods category relationship table and the goods historical bid information table are associated to obtain the historical bid information table of each category, and then the historical advertising bids of all goods in the predetermined category are obtained. Among them: when mining the goods → category mapping relationship, select the valid goods that have not been removed from the shelves, extract the category to which each product belongs, each product belongs to a unique category, and each category may include multiple goods, for example, the category may include mobile phones, computers, keyboards, mice, etc.; each product may be displayed and clicked multiple times, and the bid may be different each time it is clicked. Information aggregation is performed according to the goods, and the historical bid information of each product is mined to obtain the historical bid list of the goods.

[0096] In step S604, the reserved advertising price attribute information of the goods in the predetermined category is determined based on the historical advertising price attribute information. The price distribution of the historical advertising bids can be obtained, and the price of the predetermined quantile of the price distribution is selected as the reserved advertising bid. For example, the bid of the 25%, 30% or 35% quantile can be selected from the historical bid information table of each category as the reserved price of the category. The purpose of setting the reserve price is to obtain a bid threshold based on the historical bid level of the category to prevent advertisers from lowering the bid too low. At the same time, the number of goods in the high-bid goods set can be reduced to reduce the workload of finding data in the set.

[0097] In step S606, the real-time advertising price attribute information of the goods belonging to the predetermined category is obtained. In the advertising delivery system, the CPC of the goods that the advertiser can set in real time can be extracted from the real-time advertising information library. By associating the commodity category relationship table and the commodity real-time bidding information table, the corresponding goods under each category and their real-time CPC information can be obtained.

[0098] In step S608, select the products whose real-time advertising price attribute information is higher than the reserved advertising price attribute information to obtain a set of products with high price attribute information in the predetermined category. According to the reserve price of each category, filter out the products whose CPC is higher than the reserve price in each category, and finally obtain a category→high-bid product set mapping table.

[0099] According to the method provided in the embodiment of the present disclosure, based on product attribute information, historical advertising transaction information and real-time advertising information, high-bid products in each category are obtained through data mining and association algorithms, thereby ensuring the advertising bids of recommended products, reducing the workload of finding high-bid products, and improving the performance of the online system.

[0100] Figure 7 is based on Figure 6 1 shows a schematic diagram of a process of obtaining a set of high-bid commodities, and also shows the architecture of an offline system 70 for obtaining a set of high-bid commodities. The offline system 70 can update the set of high-bid commodities at a preset frequency, for example, regenerate the set of high-bid commodities every 0.5 hour, 1 hour, or 1.5 hours. The offline system 70 can include a data acquisition module 702, a data mining module 704, and an algorithm association module 706.

[0101] The data acquisition module 702 is used to acquire original information data, which may include a product attribute information library 7004, a historical advertising transaction information library 7002, and a real-time advertising information library 7006, etc., wherein: the product attribute information library 7004 may include the brand, category, and listing time of the product; the historical advertising transaction information library 7002 may include the clicked advertising product, the user who clicked the advertising product, the CPC at each click, etc.; the real-time advertising information library 7006 may include the CPC set by the advertiser in real time, the real-time account balance of the advertiser, the ID of the advertiser, etc. The data acquisition module 702 may obtain real-time advertising information from the advertising delivery system, and the real-time advertising information in the advertising delivery system is transmitted to the data acquisition module 702 of the offline system after a short delay.

[0102] The data mining module 704 is used to mine the product→category mapping relationship (S714), the historical bid list of the product (S712), and the real-time CPC of the product (S716) from the product attribute information library, the historical advertising transaction information library, and the real-time advertising information library respectively; in the advertising delivery system, advertisers can set the CPC of the product in real time, and the data mining module 704 can extract the real-time CPC of the product from the real-time advertising information library through the data acquisition module 702.

[0103] The algorithm association module 706 can further obtain the reserve price corresponding to each category (S722), the products under each category and their real-time CPC (S724) based on the data of the data mining module 704, and finally obtain a set of high-bid products under each category; according to the reserve price of each category, the products under each category with a CPC higher than the reserve price are screened out through algorithm association, and finally a category→high-bid product set mapping table (5002) is obtained.

[0104] Figure 8 FIG. 1 is a block diagram of an advertisement delivery device according to an exemplary embodiment. Figure 8 The device shown can be applied to the server side of the above system, for example, and can also be applied to the terminal device of the above system.

[0105] refer to Figure 8 The device 80 provided by the embodiment of the present disclosure may include a request receiving module 801, a product filtering module 802, an information obtaining module 804, a product supplement module 806, a click rate estimation module 808, a replacement preparation module 810 and a product recommendation module 812.

[0106] The request receiving module 801 may be used to receive an advertisement playing request.

[0107] The product filtering module 802 may be used to obtain a candidate advertising product set according to an advertisement play request.

[0108] The information acquisition module 804 may be used to acquire the advertisement price attribute information of the first candidate advertisement product in the candidate advertisement product set.

[0109] The product complement module 806 may be used to obtain a second candidate advertisement product based on the advertisement price attribute information of the first candidate advertisement product.

[0110] The click rate estimation module 808 may be used to obtain an estimated click rate of the first candidate advertisement product and an estimated click rate of the second candidate advertisement product.

[0111] The replacement preparation module 810 may be used to obtain a replacement reference lower limit value according to the estimated click rate of the first candidate advertisement product.

[0112] The product recommendation module 812 may be used to replace the first candidate advertisement product with the second candidate advertisement product when the estimated click rate of the second candidate advertisement product is higher than the replacement reference lower limit, and obtain an updated candidate advertisement product set for advertisement delivery.

[0113] Fig. 9 FIG. 1 is a block diagram of an advertisement delivery device according to an exemplary embodiment. Fig. 9 The device shown can be applied to the server side of the above system, for example, and can also be applied to the terminal device of the above system.

[0114] refer to Fig. 9 The device 90 provided by the embodiment of the present disclosure may include a request receiving module 901, a product filtering module 902, an information obtaining module 904, a historical data obtaining module 9052, a reserve price determining module 9054, a real-time data obtaining module 9056, a high-bid product set obtaining module 9058, a product supplement module 906, a click-through rate estimation module 908, a replacement preparation module 910 and a product recommendation module 912, wherein the product filtering module 902 may include a pre-algorithm module 9022, the product supplement module 906 may include a category query module 9062 and a supplement product obtaining module 9064, the product recommendation module 912 may include a replacement execution module 9122 and a product sorting module 9124, and the product sorting module 9124 may include a user recommendation degree calculation module 91242 and a to-be-recommended product sequence obtaining module 91244.

[0115] The request receiving module 901 may be used to receive an advertisement playing request. The advertisement playing request includes a user identifier.

[0116] The product filtering module 902 may be used to obtain a candidate advertising product set according to an advertisement play request.

[0117] The pre-algorithm module 9022 may be used to obtain a set of candidate advertising products through a collaborative filtering algorithm based on the user identifier, and the first candidate advertising products in the set of candidate advertising products are arranged from high to low according to the estimated satisfaction index.

[0118] The information acquisition module 904 may be used to acquire the advertisement price attribute information of the first candidate advertisement product in the candidate advertisement product set.

[0119] The historical data acquisition module 9052 may be used to acquire historical advertising bids for products belonging to a predetermined category.

[0120] The reserve price determination module 9054 may be used to determine a reserve advertisement bid for a product of a predetermined category based on historical advertisement bids.

[0121] The reserve price determination module 9054 may also be used to obtain the price distribution of historical advertisement bids; and select the price of a predetermined quantile of the price distribution as the reserve advertisement bid.

[0122] The real-time data acquisition module 9056 may be used to acquire real-time advertising bids for products belonging to a predetermined category.

[0123] The high-bid commodity set obtaining module 9058 may be used to select commodities whose real-time advertisement bids are higher than the reserved advertisement bids, and obtain a high-bid commodity set of a predetermined category.

[0124] The product complement module 906 may be configured to obtain a second candidate advertisement product based on the advertisement price attribute information of the first candidate advertisement product.

[0125] The product supplement module 906 is further configured to select a predetermined number of first candidate advertising products from among the first candidate advertising products arranged according to the estimated satisfaction index; and obtain a second candidate advertising product whose advertising price attribute information is higher than the advertising price attribute information of the predetermined number of first candidate advertising products.

[0126] The category query module 9062 may be used to obtain the predetermined category to which the first candidate advertisement product belongs.

[0127] The complementary product acquisition module 9064 may be used to obtain a second candidate advertisement product whose real-time advertisement bid is higher than the advertisement bid of the first candidate advertisement product from a set of high-bid products of a predetermined category.

[0128] The click rate estimation module 908 may be used to obtain an estimated click rate of the first candidate advertisement product and an estimated click rate of the second candidate advertisement product.

[0129] The click rate estimation module 908 may also be configured to multiply the estimated click rate of the first candidate advertisement product by a predetermined percentage to obtain a replacement reference lower limit value.

[0130] The replacement preparation module 910 may be used to obtain a replacement reference lower limit value according to the estimated click rate of the first candidate advertisement product.

[0131] The product recommendation module 912 can be used to replace the first candidate advertising product with the second candidate advertising product when the estimated click rate of the second candidate advertising product is higher than the replacement reference lower limit, and obtain an updated candidate advertising product set for advertising delivery. The updated candidate advertising product set includes unreplaced candidate advertising products and replacement candidate advertising products.

[0132] The replacement execution module 9122 may be used to replace the candidate advertisement product with the second candidate advertisement product to obtain a replacement candidate advertisement product.

[0133] The replacement execution module 9122 may also be used to obtain the estimated click rate of the replacement candidate advertisement product as the estimated click rate of the candidate advertisement product when the estimated click rate of the candidate advertisement product is higher than the estimated click rate of the second candidate advertisement product.

[0134] The product sorting module 9124 can be used to sort the non-replaced candidate advertising products and the replacement candidate advertising products according to the advertising price attribute information and estimated click-through rate of the non-replaced candidate advertising products and the advertising price attribute information and estimated click-through rate of the replacement candidate advertising products, so as to deliver advertisements in sequence.

[0135] The user recommendation degree calculation module 91242 may be used to obtain ranking indicators of the unreplaced candidate advertisement products based on the advertisement price attribute information and estimated click-through rates of the unreplaced candidate advertisement products.

[0136] The user recommendation degree calculation module 91242 may also be used to obtain a ranking index of the replacement candidate advertisement products based on the advertisement price attribute information and the estimated click rate of the replacement candidate advertisement products.

[0137] The module 91244 for obtaining a sequence of products to be recommended may be used to sort the unreplaced candidate advertising products and the replacement candidate advertising products in descending order according to the sorting index to obtain a sequence of advertising products to be placed.

[0138] The product recommendation module 912 may also be used to place advertisements in a sequence of products to be advertised from front to back.

[0139] The specific implementation of each module in the device provided by the embodiment of the present disclosure can refer to the content of the above method, which will not be repeated here.

[0140] Fig.10 The following is a schematic diagram showing the structure of an electronic device in an embodiment of the present disclosure. It should be noted that: Fig.10 The device shown is only an example of a computer system, which should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0141] like Fig.10 As shown, the device 1000 includes a central processing unit (CPU) 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage part 1008 into a random access memory (RAM) 1003. In the RAM 1003, various programs and data required for the operation of the device 1000 are also stored. The CPU 1001, the ROM 1002, and the RAM 1003 are connected to each other via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0142] The following components are connected to the I / O interface 1005: an input section 1006 including a keyboard, a mouse, etc.; an output section 1007 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN card, a modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to the I / O interface 1005 as needed. A removable medium 1011, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1010 as needed, so that a computer program read therefrom is installed into the storage section 1008 as needed.

[0143] 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 computer-readable medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part 1009, and / or installed from a removable medium 1011. When the computer program is executed by a central processing unit (CPU) 1001, the above-mentioned functions defined in the system of the present disclosure are executed.

[0144] It should be noted that the computer-readable medium shown in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0145] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0146] The modules involved in the embodiments described in the present disclosure may be implemented by software or hardware. The modules described may also be set in a processor, for example, it may be described as: a processor includes a product filtering module, an information acquisition module, a product supplement module, a click rate estimation module, a replacement preparation module and a product recommendation module. The names of these modules do not constitute a limitation on the modules themselves in some cases. For example, the product filtering module may also be described as a "module for obtaining candidate advertising products based on a coordinated filtering method."

[0147] As another aspect, the present disclosure also provides a computer-readable medium, which may be included in the device described in the above embodiment; or may exist independently and not be assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device includes: receiving an advertisement play request; obtaining a candidate advertisement product set according to the advertisement play request; obtaining the advertisement price attribute information of the first candidate advertisement product in the candidate advertisement product set; obtaining the second candidate advertisement product based on the advertisement price attribute information of the first candidate advertisement product; obtaining the estimated click-through rate of the first candidate advertisement product and the estimated click-through rate of the second candidate advertisement product; obtaining a replacement reference lower limit value according to the estimated click-through rate of the first candidate advertisement product; when the estimated click-through rate of the second candidate advertisement product is higher than the replacement reference lower limit value, replacing the first candidate advertisement product with the second candidate advertisement product, and obtaining an updated candidate advertisement product set for advertisement delivery.

[0148] The exemplary embodiments of the present disclosure are specifically shown and described above. It should be understood that the present disclosure is not limited to the detailed structures, configurations or implementations described herein; on the contrary, the present disclosure is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended claims.

Claims

1. An advertisement delivery method, characterized in that: include: Receive an advertisement playback request; Acquire a candidate advertising product set according to the advertisement playing request; Obtaining historical advertising bids for products of a predetermined category to which a first candidate advertising product in the candidate advertising product set belongs; Determining a reserved advertising bid for the product in the predetermined category according to the historical advertising bid; Obtaining real-time advertising bids for products belonging to the predetermined category; Selecting commodities whose real-time advertisement bids are higher than the reserved advertisement bids to obtain a set of high-bid commodities in the predetermined category; Obtaining a second candidate advertisement product from the set of high-bid products in the predetermined category, whose real-time advertisement bid is higher than the advertisement bid of the first candidate advertisement product, so that user satisfaction is not affected if the first candidate advertisement product is replaced by the second candidate advertisement product; Obtaining an estimated click-through rate of the first candidate advertisement product and an estimated click-through rate of the second candidate advertisement product; Obtaining a replacement reference lower limit value according to the estimated click rate of the first candidate advertisement product; When the estimated click rate of the second candidate advertisement product is higher than the replacement reference lower limit, the first candidate advertisement product is replaced by the second candidate advertisement product to obtain an updated candidate advertisement product set for advertisement delivery.

2. The method according to claim 1, characterized in that The updated candidate advertising commodity set includes unreplaced candidate advertising commodities and replaced candidate advertising commodities; The step of replacing the first candidate advertisement product with the second candidate advertisement product comprises: replacing the first candidate advertisement product with the second candidate advertisement product to obtain the replacement candidate advertisement product; When the estimated click rate of the first candidate advertisement product is higher than the estimated click rate of the second candidate advertisement product, obtaining the estimated click rate of the replacement candidate advertisement product as the estimated click rate of the first candidate advertisement product; The step of obtaining an updated set of candidate advertisement products for advertisement delivery includes: The unreplaced candidate advertising products and the replacement candidate advertising products are sorted according to the advertising price attribute information and the estimated click rate of the unreplaced candidate advertising products and the advertising price attribute information and the estimated click rate of the replacement candidate advertising products, so as to place advertisements in sequence.

3. The method according to claim 2, characterized in that The sorting of the unreplaced candidate advertisement products and the replacement candidate advertisement products according to the advertisement price attribute information and estimated click-through rate of the unreplaced candidate advertisement products and the advertisement price attribute information and estimated click-through rate of the replacement candidate advertisement products comprises: Obtaining a ranking index of the unreplaced candidate advertisement products according to the advertisement price attribute information and the estimated click rate of the unreplaced candidate advertisement products; Obtaining a ranking index of the replacement candidate advertisement products according to the advertisement price attribute information and the estimated click rate of the replacement candidate advertisement products; Sorting the unreplaced candidate advertising products and the replaced candidate advertising products in descending order according to the sorting index to obtain a sequence of advertising products to be placed; The sequential advertisement delivery includes: The sequence of products to be advertised is placed from front to back for advertisement placement.

4. The method according to claim 1, characterized in that The advertisement playing request includes a user identifier; The step of obtaining a candidate advertising product set according to the advertisement playing request comprises: Based on the user identifier, the candidate advertising product set is obtained by using a collaborative filtering algorithm, and the first candidate advertising products in the candidate advertising product set are arranged from high to low according to the estimated satisfaction index; The second candidate advertisement product whose real-time advertisement bid is higher than the advertisement bid of the first candidate advertisement product is obtained from the high-bid product set of the predetermined category, comprising: Selecting a predetermined number of first candidate advertising products from among the first candidate advertising products arranged according to the estimated satisfaction index; The second candidate advertisement products whose real-time advertisement bids are higher than the advertisement bids of the first candidate advertisement products ranked in the predetermined number are obtained from a set of high-bid products in a predetermined category to which the first candidate advertisement products ranked in the predetermined number are affiliated.

5. The method according to claim 1, characterized in that Also includes: The predetermined category to which the first candidate advertisement product belongs is obtained.

6. The method according to claim 1, characterized in that Determining the reserved advertisement bid of the product in the predetermined category according to the historical advertisement bid includes: Obtaining the price distribution of the historical advertisement bids; A price at a predetermined percentile of the price distribution is selected as the reserved advertisement bid.

7. The method according to claim 1, characterized in that The step of obtaining a replacement reference lower limit value according to the estimated click rate of the first candidate advertisement product includes: The replacement reference lower limit value is obtained by multiplying the estimated click rate of the first candidate advertisement product by a predetermined percentage.

8. An advertisement delivery device, characterized in that: include: A request receiving module, used for receiving an advertisement playing request; A product filtering module, used to obtain a set of candidate advertisement products according to the advertisement playing request; A historical data acquisition module, used to acquire historical advertising bids of products belonging to a predetermined category to which a first candidate advertising product in the candidate advertising product set belongs; A reserve price determination module, used to determine a reserve advertisement bid for the product of the predetermined category according to the historical advertisement bid; A real-time data acquisition module, used to acquire real-time advertising bids for products belonging to the predetermined category; A high-bid commodity set obtaining module, used for selecting commodities whose real-time advertisement bid is higher than the reserved advertisement bid, and obtaining a high-bid commodity set of the predetermined category; A complementary product acquisition module, configured to obtain a second candidate advertisement product whose real-time advertisement bid is higher than the advertisement bid of the first candidate advertisement product from the set of high-bid products of the predetermined category, so that user satisfaction is not affected if the first candidate advertisement product is replaced by the second candidate advertisement product; A click rate estimation module, used to obtain an estimated click rate of the first candidate advertisement product and an estimated click rate of the second candidate advertisement product; A replacement preparation module, configured to obtain a replacement reference lower limit value according to the estimated click rate of the first candidate advertisement product; The product recommendation module is used to replace the first candidate advertising product with the second candidate advertising product when the estimated click rate of the second candidate advertising product is higher than the replacement reference lower limit, and obtain an updated candidate advertising product set for advertising delivery.

9. A device comprising: A memory, a processor, and executable instructions stored in the memory and executable in the processor, wherein the processor implements the method according to any one of claims 1 to 7 when executing the executable instructions.

10. A computer-readable storage medium having computer-executable instructions stored thereon, characterized in that: When the executable instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images