CN113326436A

CN113326436A - Method and device for determining recommended resources, electronic equipment and storage medium

Info

Publication number: CN113326436A
Application number: CN202110669221.8A
Authority: CN
Inventors: 刘新觅; 金锐
Original assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Current assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Priority date: 2021-06-16
Filing date: 2021-06-16
Publication date: 2021-08-31
Anticipated expiration: 2041-06-16
Also published as: CN113326436B

Abstract

The disclosure discloses a method, a device, electronic equipment and a storage medium for determining recommended resources, which are applied to the technical field of artificial intelligence, in particular to the technical field of intelligent recommendation and the technical field of deep learning. The specific implementation scheme of the method for determining the recommended resources is as follows: for each resource in the recalled multiple resources, determining a predicted click rate of each resource for the target object by adopting a click rate prediction model; determining click rate weight of the predicted click rate for the target object based on the first position information of the target object and the attribute information of each resource; determining the click rate of each resource for the target object based on the predicted click rate and the click rate weight; and determining recommended resources for the target object in the plurality of resources based on the click rate of the plurality of resources for the target object.

Description

Method and device for determining recommended resources, electronic equipment and storage medium

Technical Field

The present disclosure relates to the field of artificial intelligence technologies, in particular to the field of intelligent recommendation technologies and the field of deep learning technologies, and more particularly to a method and an apparatus for determining recommended resources, an electronic device, and a storage medium.

Background

With the development of network technology, intelligent recommendation technology is rapidly developed to provide users with information meeting requirements and points of interest. Popular personalized recommendation algorithms include a popular ranking list-based recommendation algorithm, a content-based recommendation algorithm, a social network-based recommendation algorithm and the like. The methods generally adopt a general rough sorting model and a general fine sorting model, and resources ranked at the top several positions are selected from recalled resources and recommended to a user.

Disclosure of Invention

A method, an apparatus, an electronic device, and a storage medium for determining recommended resources are provided, which improve recommendation accuracy and user experience.

According to an aspect of the present disclosure, there is provided a method of determining recommended resources, including: for each resource in the recalled multiple resources, determining a predicted click rate of each resource for the target object by adopting a click rate prediction model; determining click rate weight of the predicted click rate for the target object based on the first position information of the target object and the attribute information of each resource; determining the click rate of each resource for the target object based on the predicted click rate and the click rate weight; and determining recommended resources for the target object in the plurality of resources based on the click rate of the plurality of resources for the target object.

According to another aspect of the present disclosure, there is provided an apparatus for determining recommended resources, including: the click rate prediction module is used for determining the predicted click rate of each resource for the target object by adopting a click rate prediction model aiming at each resource in the recalled resources; the weight determining module is used for determining click rate weight of the predicted click rate aiming at the target object based on the first position information of the target object and the attribute information of each resource; the click rate determining module is used for determining the click rate of each resource aiming at the target object based on the predicted click rate and the click rate weight; and a recommended resource determining module, configured to determine recommended resources for the target object in the multiple resources based on click rates of the multiple resources for the target object.

According to another aspect of the present disclosure, there is provided an electronic device including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of determining recommended resources provided by the present disclosure.

According to another aspect of the present disclosure, there is provided a non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform the method of determining recommended resources provided by the present disclosure.

According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the method of determining recommended resources provided by the present disclosure.

It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present disclosure, nor do they limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description.

Drawings

The drawings are included to provide a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:

FIG. 1 is a schematic diagram of an application scenario of a method and apparatus for determining recommended resources according to an embodiment of the present disclosure;

FIG. 2 is a flow diagram of a method of determining recommended resources according to an embodiment of the present disclosure;

FIG. 3 is a schematic diagram illustrating a principle of determining click rate weights of predicted click rates for a target object according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram illustrating a principle of determining a correlation value between a click rate and a matching position of each resource according to an embodiment of the disclosure;

FIG. 5 is a block diagram of an apparatus for determining recommended resources according to an embodiment of the present disclosure; and

FIG. 6 is a block diagram of an electronic device used to implement a method of determining recommended resources of embodiments of the present disclosure.

Detailed Description

Exemplary embodiments of the present disclosure are described below with reference to the accompanying drawings, in which various details of the embodiments of the disclosure are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

The present disclosure provides a method for determining recommended resources, which includes a click rate prediction stage, a weight determination stage, a click rate determination stage and a recommended resource determination stage. In the click rate prediction phase, for each resource in the recalled multiple resources, a click rate prediction model is adopted to determine the predicted click rate of each resource for the target object. In the weight determination stage, click rate weights of the predicted click rates for the target objects are determined based on the first position information of the target objects and the attribute information of each resource. In the click rate determination stage, the click rate of each resource for the target object is determined based on the predicted click rate and the click rate weight. In the recommended resource determining stage, recommended resources for the target object in the multiple resources are determined based on click rates of the multiple resources for the target object.

An application scenario of the method and apparatus provided by the present disclosure will be described below with reference to fig. 1.

Fig. 1 is a schematic application scenario diagram of a method and an apparatus for determining recommended resources according to an embodiment of the present disclosure.

As shown in fig. 1, the application scenario 100 includes a terminal device 110, a server 120, and a database 140. Terminal device 110 may be communicatively coupled to server 120 via a network, which may include wired or wireless communication links. The server 120 may also access the database 140, for example, over a network.

A user may interact with server 120 over a network, for example, using terminal device 110 to receive or send messages, etc. Terminal device 110 may be a terminal device having a display screen including, but not limited to, a smart phone, a tablet computer, a laptop portable computer, a desktop computer, and the like. For example, the terminal device 110 may send a request message 130 to the server 120 over the network to request multimedia resources such as audio, video, images, text, or any other resources.

The server 120 may, for example, recall a resource matching the request information 130 from the database 140 in response to the request information 130 sent by the terminal device 110, and feed back the matched resource as recommendation information 150 to the terminal device for presentation by the terminal device.

According to an embodiment of the present disclosure, the server 120 may be a server providing various services, such as a background management server providing support for a website or a client application browsed by a user using a terminal device. The server may be a cloud server, a server of a distributed system, or a server with a combined blockchain.

In an embodiment, after the server 120 recalls the resources from the database, the server 120 may also sort the recalled resources according to a matching degree between the recalled resources and the request information 130, and feed back a plurality of top-ranked resources as the recommendation information 150 to the terminal device, so as to improve the accuracy of the recommendation information 150.

Illustratively, the server 120 may determine the degree of match based on, for example, the click-through prediction model 160. The click-through rate prediction model 160 may be pre-trained by the server 120 or other electronic device communicatively coupled to the server 120.

It should be noted that the method for determining recommended resources provided by the present disclosure may be executed by the server 120. Accordingly, the apparatus for determining recommended resources provided by the present disclosure may be disposed in the server 120.

It should be understood that the number and type of terminal devices, servers, and databases in fig. 1 are merely illustrative. There may be any number and type of terminal devices, servers, and databases, as the implementation requires.

The method for determining recommended resources provided by the present disclosure will be described in detail below with reference to fig. 2 to 4.

FIG. 2 is a flow diagram of a method of determining recommended resources according to an embodiment of the disclosure.

As shown in fig. 2, the method 200 of determining recommended resources of this embodiment may include operations S210 to S240.

In operation S210, for each resource of the recalled plurality of resources, a click through rate prediction model is employed to determine a predicted click through rate of each resource for the target object.

According to the embodiment of the disclosure, the plurality of resources may be, for example, multimedia resources such as video, audio, images, and texts, and may also be car rental service resources, home services resources, home appliance maintenance service resources, and the like, which provide convenience for life. The plurality of resources may be recalled in response to request information sent by the client application at startup in the terminal device, or in response to request information sent by the terminal device according to the query keyword. The types of the plurality of resources correspond to the request information.

According to the embodiment of the disclosure, the resources with the keyword as the label can be recalled from the database according to the keyword in the request information, and a plurality of recalled resources can be obtained. Alternatively, a plurality of resources with higher degrees of popularity may be recalled from the database according to the degrees of popularity of the resources in the database. Or, a batch of resources can be roughly screened from massive information in the database according to the historical behavior data of the target object, and the resources are used as a plurality of recalled resources. It is to be understood that the above-mentioned method of recalling a plurality of resources, the timing of recalling a plurality of resources, and the type of a plurality of resources are only examples to facilitate understanding of the present disclosure, and the present disclosure is not limited thereto.

After a plurality of resources are recalled, a click through rate prediction model can be used to determine the predicted click through rate of each resource. The click rate estimation model may include, for example, a logistic regression model, a deep neural network model, or a model constructed based on an Ensemble Learning (Ensemble Learning) method. The model constructed based on the ensemble learning method can comprise a Boosting model, a Bagging model and the like, and the ensemble learning method obtains the click rate estimation model by training a plurality of classifiers and combining the classifiers, so that the click rate estimation model has higher precision. The embodiment can encode the basic attribute, the interest attribute, the environment information and the like of the target object, encode the historical click rate, the resource content and the like of the resource, use the information obtained after encoding as the input of the click rate estimation model, and output the information after the click rate estimation model processing to obtain the predicted click rate of each resource for the target object.

In operation S220, a click rate weight of the predicted click rate for the target object is determined based on the first location information of the target object and the attribute information of each resource.

According to an embodiment of the present disclosure, a similarity between the first location information and the attribute information of each resource may be determined, and the click rate weight may be determined according to the similarity. For example, a first encoding of the first location information and a second encoding of the attribute information may be obtained using a one-hot encoding. The similarity is then taken as the pearson correlation coefficient, cosine similarity, or Jacard similarity coefficient between the first code and the second code. The similarity and the click rate weight may be in a direct proportional relationship, for example, such that the click rate weight is derived based on the similarity.

According to an embodiment of the present disclosure, the attribute information of each resource may include second location information for which each resource is directed. The embodiment may determine matching information between the first location information and the second location information. Based on the matching information, a click-through rate weight of the predicted click-through rate for the target object may be determined. For example, the match information may include a match value, with the match value as a click-through rate weight. Wherein the match value may be determined based on a granularity of a match location between the first location information and the second location information. For example, if the first location information and the second location information each include provincial, city, county, district, and business district location information, when only the provincial location information matches the second location information in the first location information, the matching value is a smaller value. When the provincial level, the city level, the county level, the district level and the business district level position information in the first position information are matched with the second position information, the matching value is a larger value. I.e., the higher the level of the matching location, the smaller the matching value. For example, when two pieces of location information are the same, it may be determined that the two pieces of location information match.

It is to be understood that the above-mentioned method for determining click rate weight is only used as an example to facilitate understanding of the present disclosure, and the present disclosure may also employ the principle of determining click rate weight described below to determine click rate weight, for example, and will not be described in detail herein.

In operation S230, a click rate of each resource for the target object is determined based on the predicted click rate and the click rate weight.

In operation S240, recommended resources for the target object among the plurality of resources are determined based on click rates of the plurality of resources for the target object.

According to the embodiment of the disclosure, the product of the click rate weight and the predicted click rate can be used as the click rate of each resource for the target object. After the click rates of the plurality of resources for the target object are obtained, the plurality of resources may be ranked from high to low according to the click rates. And taking a plurality of resources ranked earlier as recommended resources for the target object.

It can be appreciated that, after a plurality of resources are recalled, the determined click rate can fully consider the geographical relevance of the resources and the target object by determining the click rate weight based on the first position information and the resource attribute information of the target object and determining the click rate of each resource based on the click rate weight. Therefore, the recommended resources determined based on the click rate comprise the resources around the target object, and the click rate of the resources and the user satisfaction are improved conveniently. Furthermore, since the click rate weight is determined based on the matching information between the location information of the resource and the location information of the target object, the relevance of the determined click rate and the geographic location matching relationship can be improved, and the probability of describing the resource of the information around the target object as the recommended resource can be improved.

FIG. 3 is a schematic diagram illustrating a principle of determining click-through rate weights of predicted click-through rates for a target object according to an embodiment of the disclosure.

According to the embodiment of the disclosure, the click rate of the resource to the matching position in the matching information can be considered, the incidence relation between the click rate of the resource and the matching position is determined based on the click rate, and the click rate weight is determined based on the incidence relation and the matching information, so that the accuracy of the determined click rate weight is improved. This is because the level of interest of a target object in a resource is generally subject to the overall preference of the population in the same territory. The click rate weight is determined by considering the click rate of the resource at the matching position, so that the finally determined click rate of the resource can describe the group preference in the region.

Illustratively, the aforementioned matching information may include a matching position in addition to the matching value. For example, if the first location information of the target object is XX business circles in P, L city, a province, and the second location information in the attribute information of the resource is Q, M city, a province, or M city, a province, the matching location is included in the attribute information of the resource.

According to the embodiment of the disclosure, when determining the click rate weight, in addition to determining the matching information by using the method described above, the association value between the click rate and the matching position of each resource may be determined based on the historical click information and the historical display information of each resource. Click rate weights are then determined based on the relevance and match values.

For example, the information clicked by the object at the matched position can be picked out from the historical click information, and the click rate of each resource for the matched position can be obtained. Similarly, the exposure amount of each resource for the matching location can be obtained. And taking the ratio of the click quantity to the display quantity as the historical click rate of each resource for the matching position. A relevance value may be determined based on the historical click rate, e.g., a positive correlation between the historical click rate and the relevance value. For example, the historical click rate of each resource can be used as the correlation value. Alternatively, the historical click rates of the plurality of resources for the matching positions may be normalized, and the value obtained by the normalization may be used as the correlation value. It is to be understood that the above method for determining the association value based on the historical click rate is only used as an example to facilitate understanding of the present disclosure, and the present disclosure is not limited thereto.

According to embodiments of the present disclosure, the historical click rate may be affected by the difference between the actual display amount of the resource and the predetermined display amount. This is because the resource with high popularity has a high click rate, and the display amount of the resource affects the popularity of the resource to some extent. Therefore, when determining the relevance value, the embodiment may further adjust the value of the historical click rate based on the historical display information, so as to improve the accuracy of the determined relevance value.

In one embodiment, as shown in fig. 3, in determining the click-through rate weight, for each resource 320, the embodiment 300 may first determine matching information 330 between the first location information 311 of the target object 310 and the second location information 321 in the attribute information of the resource 320. Subsequently, a confidence 340 of the resource 320 for the matching location may be determined based on historical presentation information 323 in the attribute information of the resource 320. Meanwhile, the actual click rate 350 of each resource for the matching location may be determined based on the historical click information 322 and the historical presentation information 323 in the attribute information of the resource 320. An association value 360 between the click-through rate and the matching location for each resource may then be determined based on the resulting confidence 340 and the actual click-through rate 350. Finally, a click-through-rate weight 370 may be determined based on the relevance value 360 and the match value in the match information 330.

For example, when the confidence is determined, the display information displayed to the object at the matching position in the historical display information may be first picked out, and the actual display amount of each resource for the matching position is obtained. A confidence is then determined based on a ratio of the actual exposure to a predetermined exposure for the matching location. For example, the ratio between the actual exposure and the predetermined exposure may be positively correlated with the confidence. The predetermined display amount may be, for example, an average display amount of a plurality of resources, or may be a preset arbitrary value. In one embodiment, the predetermined presentation amount is associated with a matching location. For example, the higher the level of the matching location, the higher the predetermined exposure.

Illustratively, aging information for each resource, which is part of the attribute information, may also be considered in determining the confidence level. This is because, in order to reduce the amount of calculation, the historical display information is usually display information within a predetermined period of time, and the age information of the resource affects the display amount to some extent. After the actual display amount is obtained, the confidence of each resource for the matching position can be determined based on the actual display amount, the predetermined display amount and the aging information, so that the accuracy of the determined confidence is improved, and the accuracy of the determined association value and the click rate weight is improved. The aging information may be represented by, for example, a time interval between the current time and the release time of each resource. The age information may be inversely related to the confidence level.

Illustratively, for example, a weight of a ratio between the actual display amount and the predetermined display amount may be obtained based on the aging information, and the weighted ratio between the actual display amount and the predetermined display amount may be used as the confidence. The time interval between the release time of each resource and the current time may be used to represent the time period of the time interval, and the value of the time interval may be negatively correlated with the value of the weight. Or, the value of the aging information can be used as an adjusting factor, the value of the ratio between the actual display amount and the preset display amount is adjusted, and the difference value between the ratio and the adjusting factor is used as confidence. Or, the value of an exponential function with the difference between the ratio and the adjustment factor as a variable may be used as a confidence, and the like, which is not limited in the present disclosure.

For example, after obtaining the actual click rate and the confidence level, the correlation value between the click rate and the matching position of each resource can be determined based on the product of the confidence level and the actual click rate. Wherein the product between the confidence and the actual click-through rate is positively correlated to the correlation value, for example. For example, the product may be used directly as a confidence level, or the correlation value may be determined based on any positive correlation function, which is not limited by the present disclosure.

Illustratively, after obtaining the relevance value and the matching value, for example, the product between the relevance value and the matching value may be used as the click rate weight. Alternatively, the average between the associated value and the matching value may be used as the click rate weight. The method for determining the click rate weight based on the correlation value and the matching value is not limited in this embodiment, as long as both the correlation value and the matching value are positively correlated with the click rate weight.

FIG. 4 is a schematic diagram illustrating a principle of determining a correlation value between a click rate and a matching position of each resource according to an embodiment of the disclosure.

According to the embodiment of the disclosure, when determining the association value between the click rate and the matching position of each resource based on the actual click rate and the confidence, for example, a reference click rate may be further considered in order to improve the situation that the association value is inaccurate due to the low display amount of the resource. The reference click rate can be obtained in advance based on the historical click rate of each resource in the database for the matching position. For example, the reference click rate may be an average value of historical click rates of each resource in the database for the matching location, and the value of the reference click rate is not limited in the present disclosure.

According to an embodiment of the disclosure, as shown in fig. 4, for each resource 410, when determining the correlation value between the click rate and the matching position of each resource, the embodiment 400 may first determine an actual click rate 420 based on the historical click information 411 and the historical display information 412 of the resource 410. And determines confidence 430 based on historical presentation information 412 using the methods described above. A weighted click rate 450 for the resource 410 for the matching location may then be determined based on the resulting actual click rate 420, confidence 430, and predetermined reference click rate 440. Finally, an evaluation value of the click rate of the resource 410 for the matching position is determined based on the ratio between the weighted click rate 450 and the predetermined reference click rate 440, and the evaluation value is taken as a correlation value 460.

For example, the confidence may be used as a weight of the actual click rate, a difference between a predetermined value and the confidence may be used as a weight of the predetermined reference click rate, a weighted sum of the actual click rate and the predetermined reference click rate may be calculated, and the weighted sum may be used as the weighted click rate. Alternatively, the product of the confidence and the actual click rate may also be directly used as the weighted click rate, which is not limited by the present disclosure.

For example, when determining the evaluation value, the evaluation value may be mapped to a certain value range based on a ratio between the weighted click rate 450 and the predetermined reference click rate 440, so as to limit a value range of the click rate weight and control a magnitude of an adjustment effect of the click rate weight on the predicted click rate.

For example, the embodiment may take the smaller of the ratio between the weighted click rate 450 and the predetermined reference click rate 440 and a first predetermined value as the final evaluation value to limit the evaluation value within the first predetermined value. Alternatively, after the smaller value between the ratio and the first predetermined value is obtained, the larger value between the smaller value and the second predetermined value may be taken as the final evaluation value to limit the evaluation value between the second predetermined value and the first predetermined value. Alternatively, the ratio between the weighted click rate 450 and the predetermined reference click rate 440 may be first amplified, and the evaluation value may be determined based on the magnitude relationship between the amplified value and the predetermined value.

According to the embodiment of the present disclosure, when the matching information includes at least two matching positions of at least two levels, for each matching position, the foregoing method may be adopted to determine the evaluation value of the click rate of each resource for each matching position, so as to obtain at least two evaluation values of the click rate of each resource for at least two matching positions, respectively. As such, in determining the association value, the maximum value of the at least two evaluation values may be taken as the association value between the click rate and the matching position of each resource. For example, if the matching position in the matching information includes a province a and M city, the embodiment may obtain an evaluation value of each resource for the province a based on the history click information and the history presentation information of the resource for the province a. Similarly, the evaluation value for M cities for each asset may be used. The evaluation value for a province and the evaluation value for M city, whichever is larger, are taken as the associated values between the click rate and the matching position of each resource.

According to the embodiment of the present disclosure, when the matching information includes at least two matching positions of at least two levels, it is also possible to determine an evaluation value of the click rate of each resource for a matching position of the smallest granularity among the at least two matching positions only by the foregoing method, and to take the evaluation value as an associated value between the click rate of each resource and the matching position.

Based on the method for determining recommended resources, the present disclosure also provides an apparatus for determining recommended resources, which will be described in detail below with reference to fig. 5.

Fig. 5 is a block diagram of an apparatus for determining recommended resources according to an embodiment of the present disclosure.

As shown in fig. 5, the apparatus 500 for determining recommended resources of this embodiment may include a click-through rate prediction module 510, a weight determination module 520, a click-through rate determination module 530, and a recommended resource determination module 540.

Click-through rate prediction module 510 is configured to, for each resource of the recalled plurality of resources, determine a predicted click-through rate for the target object for each resource using a click-through rate prediction model. In an embodiment, the click-through rate predicting module 510 may be configured to perform the operation S210 described above, which is not described herein again.

The weight determination module 520 is configured to determine click rate weights of the predicted click rates for the target objects based on the first location information of the target objects and the attribute information of each resource. In an embodiment, the weight determining module 520 may be configured to perform the operation S220 described above, which is not described herein again.

The click-through rate determination module 530 is configured to determine a click-through rate for each resource with respect to the target object based on the predicted click-through rate and the click-through rate weight. In an embodiment, the click rate determining module 530 may be configured to perform the operation S230 described above, which is not described herein again.

The recommended resource determining module 540 is configured to determine recommended resources for the target object from the multiple resources based on click rates of the multiple resources for the target object. In an embodiment, the recommended resource determining module 540 may be configured to perform the operation S240 described above, which is not described herein again.

According to an embodiment of the present disclosure, the attribute information of each resource includes second location information for which the resource is directed. The weight determination module 520 may include a matching information determination sub-module and a weight determination sub-module. The matching information determination submodule is used for determining matching information between the first position information and the second position information. The weight determination submodule is used for determining click rate weight of the predicted click rate for the target object based on the matching information.

According to the embodiment of the disclosure, the attribute information of each resource further includes historical click information and historical display information of the resource. The matching information includes a matching position and a matching value. The weight determination module 520 may also include an association value determination sub-module. And the association value determination submodule is used for determining an association value between the click rate and the matching position of each resource based on the historical click information and the historical display information. The weight determination submodule is specifically configured to determine, based on the correlation value and the matching value, a click rate weight of the predicted click rate for the target object.

According to an embodiment of the present disclosure, the association value determination sub-module may include a confidence determination unit, a click rate determination unit, and an association value determination unit. The confidence degree determining unit is used for determining the confidence degree of each resource for the matching position based on the historical display information. And the click rate determining unit is used for determining the actual click rate of each resource aiming at the matched position based on the historical click information and the historical display information. And the association value determining unit is used for determining an association value between the click rate and the matching position of each resource based on the actual click rate and the confidence.

According to an embodiment of the present disclosure, the attribute information of each resource further includes age information of the resource. The confidence determination unit may include a presentation amount determination subunit and a confidence determination subunit. And the display amount determining subunit is used for determining the actual display amount of each resource for the matching position based on the historical display information. The confidence degree determination subunit is used for determining the confidence degree of each resource for the matching position based on the actual display amount, the preset display amount associated with the matching position and the aging information.

According to an embodiment of the present disclosure, the association value determining unit may include a weight determining subunit and an association value determining subunit. And the weighted determining subunit is used for determining the weighted click rate of each resource for the matching position based on the actual click rate, the preset reference click rate and the confidence for the matching position. And the association value determining subunit is used for determining an evaluation value of the click rate of each resource for the matching position based on the ratio of the weighted click rate to a predetermined reference click rate to obtain an association value.

According to an embodiment of the present disclosure, the matching information includes at least two matching positions of at least two levels, and the evaluation value includes at least two evaluation values for a click rate of each resource respectively for the at least two matching positions. The above-mentioned associated value determining subunit is configured to determine a largest evaluation value of the at least two evaluation values as an associated value.

It should be noted that, in the technical solution of the present disclosure, the acquisition, storage, application, and the like of the personal information of the related user all conform to the regulations of the relevant laws and regulations, and do not violate the common customs of the public order.

The present disclosure also provides an electronic device, a readable storage medium, and a computer program product according to embodiments of the present disclosure.

FIG. 6 illustrates a schematic block diagram of an example electronic device 600 that may be used to implement the method of determining recommended resources of embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the disclosure described and/or claimed herein.

As shown in fig. 6, the apparatus 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM)602 or a computer program loaded from a storage unit 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device 600 can also be stored. The calculation unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.

A number of components in the device 600 are connected to the I/O interface 605, including: an input unit 606 such as a keyboard, a mouse, or the like; an output unit 607 such as various types of displays, speakers, and the like; a storage unit 608, such as a magnetic disk, optical disk, or the like; and a communication unit 609 such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information/data with other devices via a computer network such as the internet and/or various telecommunication networks.

The computing unit 601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various dedicated Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, and so forth. The calculation unit 601 performs the respective methods and processes described above, such as a method of determining recommended resources. For example, in some embodiments, the method of determining recommended resources may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 600 via the ROM 602 and/or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the above described method of determining recommended resources may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured by any other suitable means (e.g. by means of firmware) to perform the method of determining recommended resources.

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, Field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), system on a chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.

Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, 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 disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), and the Internet.

The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The Server may be a cloud Server, which is also called a cloud computing Server or a cloud host, and is a host product in a cloud computing service system, so as to solve the defects of high management difficulty and weak service extensibility in a traditional physical host and a VPS service ("Virtual Private Server", or simply "VPS"). The server may also be a server of a distributed system, or a server incorporating a blockchain.

It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present disclosure may be executed in parallel, sequentially, or in different orders, as long as the desired results of the technical solutions disclosed in the present disclosure can be achieved, and the present disclosure is not limited herein.

The above detailed description should not be construed as limiting the scope of the disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present disclosure should be included in the scope of protection of the present disclosure.

Claims

1. A method of determining recommended resources, comprising:

for each resource in a plurality of recalled resources, determining a predicted click rate of each resource for a target object by adopting a click rate estimation model;

determining click rate weight of the predicted click rate for the target object based on the first position information of the target object and the attribute information of each resource;

determining the click rate of each resource for the target object based on the predicted click rate and the click rate weight; and

determining recommended resources of the plurality of resources for the target object based on click rates of the plurality of resources for the target object.

2. The method of claim 1, wherein the attribute information of each resource includes second location information for which the resource is directed; determining click rate weights for the predicted click rate for the target object comprises:

determining matching information between the first location information and the second location information; and

based on the matching information, click rate weights of the predicted click rates for the target objects are determined.

3. The method of claim 2, wherein the attribute information of each resource further comprises historical click information and historical show information; the matching information comprises a matching position and a matching value; determining click rate weights for the predicted click rate for the target object further comprises:

determining an association value between the click rate of each resource and the matching position based on the historical click information and the historical presentation information,

wherein the click rate weight of the predicted click rate for the target object is determined based on the relevance value and the match value.

4. The method of claim 3, wherein determining the associated value between the click-through rate of each resource and the matching location comprises:

determining a confidence level of each resource for the matching location based on the historical presentation information;

determining an actual click rate of each resource for the matching position based on the historical click information and the historical display information; and

and determining an association value between the click rate of each resource and the matching position based on the actual click rate and the confidence level.

5. The method of claim 4, wherein the attribute information of each resource further includes age information of the resource; determining a confidence level for the each resource for the matching location comprises:

determining an actual display amount of each resource for the matching location based on the historical display information; and

determining a confidence level of the each resource for the matching location based on the actual exposure amount, a predetermined exposure amount associated with the matching location, and the age information.

6. The method of claim 4, wherein the determining a correlation value between the click-through rate of each resource and the matching location based on the actual click-through rate and the confidence comprises:

determining a weighted click rate of each resource for the matching location based on the actual click rate, a predetermined reference click rate for the matching location, and the confidence; and

and determining the evaluation value of the click rate of each resource for the matching position based on the ratio of the weighted click rate to the preset reference click rate to obtain the correlation value.

7. The method according to claim 6, wherein the matching information includes at least two matching positions of at least two levels, the evaluation values include at least two evaluation values of the click rate of each resource for the at least two matching positions, respectively; the determining the correlation value between the click rate of each resource and the matching position further comprises:

determining a maximum evaluation value of the at least two evaluation values as the associated value.

8. An apparatus for determining recommended resources, comprising:

the click rate prediction module is used for determining the predicted click rate of each resource for the target object by adopting a click rate prediction model aiming at each resource in the recalled resources;

a weight determination module, configured to determine, based on the first location information of the target object and the attribute information of each resource, a click rate weight of the predicted click rate for the target object;

the click rate determining module is used for determining the click rate of each resource aiming at the target object based on the predicted click rate and the click rate weight; and

and the recommended resource determining module is used for determining recommended resources aiming at the target object in the plurality of resources based on the click rate of the plurality of resources aiming at the target object.

9. The apparatus of claim 8, wherein the attribute information of each resource includes second location information for which the resource is directed; the weight determination module includes:

a matching information determination submodule for determining matching information between the first location information and the second location information; and

and the weight determination submodule is used for determining the click rate weight of the predicted click rate aiming at the target object based on the matching information.

10. The apparatus of claim 9, wherein the attribute information of each resource further comprises historical click information and historical show information; the matching information comprises a matching position and a matching value; the weight determination module further comprises:

a correlation value determination submodule for determining a correlation value between the click rate of each resource and the matching position based on the historical click information and the historical display information,

wherein the weight determination submodule is configured to determine a click rate weight of the predicted click rate for the target object based on the relevance value and the matching value.

11. The apparatus of claim 10, wherein the correlation value determination submodule comprises:

a confidence determining unit, configured to determine a confidence of each resource with respect to the matching location based on the historical presentation information;

the click rate determining unit is used for determining the actual click rate of each resource aiming at the matching position based on the historical click information and the historical display information; and

and the association value determining unit is used for determining an association value between the click rate of each resource and the matching position based on the actual click rate and the confidence level.

12. The apparatus of claim 11, wherein the attribute information of each resource further includes age information of the resource; the confidence level determination unit includes:

a display amount determining subunit, configured to determine, based on the historical display information, an actual display amount of each resource for the matching location; and

a confidence determining subunit, configured to determine a confidence of each resource for the matching location based on the actual display amount, the predetermined display amount associated with the matching location, and the aging information.

13. The apparatus of claim 11, wherein the association value determining unit comprises:

a weighted determination subunit, configured to determine, based on the actual click rate, a predetermined reference click rate for the matching position, and the confidence, a weighted click rate of each resource for the matching position; and

and the association value determining subunit is configured to determine, based on a ratio between the weighted click rate and the predetermined reference click rate, an evaluation value of the click rate of each resource with respect to the matching position, so as to obtain the association value.

14. The apparatus of claim 13, wherein the matching information comprises at least two matching locations of at least two levels, the evaluation values comprise at least two evaluation values of click through rate of each resource for the at least two matching locations, respectively; the association value determination subunit is specifically configured to:

15. An electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

16. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any of claims 1-7.

17. A computer program product comprising a computer program which, when executed by a processor, implements a method according to any one of claims 1 to 7.