CN110110233B - Information processing method, device, medium and computing equipment - Google Patents

Abstract

The embodiment of the invention provides an information processing method. The method comprises the following steps: acquiring user information of a user; obtaining a plurality of pieces of information to be recommended according to the user information, wherein the plurality of pieces of information to be recommended comprise first information to be recommended with pre-sequencing information; and acquiring a plurality of predicted click rates corresponding to the plurality of information to be recommended one by adopting a click rate prediction model according to the pre-sequencing information of the first information to be recommended, the user information and the plurality of information to be recommended. According to the method, when the predicted click rate of the first information to be recommended is determined, the pre-sequencing information of the first model to be recommended is considered, so that the accuracy of the obtained predicted click rate can be improved, and the accuracy and the recommendation effect of information recommendation can be improved. In addition, the embodiment of the invention also provides an information processing device, a medium and a computing device.

Description

Information processing method, device, medium and computing equipment

Technical Field

The embodiment of the invention relates to the field of information recommendation, in particular to an information processing method, an information processing device, an information processing medium and a computing device.

Background

This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. The description herein is not admitted to be prior art by inclusion in this section.

The main task of information recommendation is to solve the problem of information overload, i.e. to screen out a small amount of information of interest to the user from a large amount of information. The general information recommendation roughly comprises two stages of recalling and sorting. The recalling stage is to specifically select part of information which is interested by the user from the massive information stored by the server. The sorting stage is to sort the partial information selected in the recall stage.

The recall stage can adopt a rule recall method and a model recall method. The rule recall may be, for example, a selection of information based on manual rules; and the model recall is to adopt a calculation model to select information. Generally, a plurality of pieces of information to be recommended selected by model recall all have pre-ranking information. And the recommendation algorithm model adopted in the ranking stage cannot absorb the pre-ranking information of the plurality of information to be recommended. That is, the pre-ranking information often cannot be considered in the ranking stage, which may compromise the recommendation performance of the recommendation system established based on the recommendation algorithm model.

Disclosure of Invention

Therefore, in the prior art, when the existing information recommendation method is adopted to recommend information to a user, the obtained pre-ranking information of the information to be recommended cannot be considered, so that the defects that the click rate accuracy of the predicted information to be recommended is low and the information recommendation effect is poor exist.

Therefore, an improved information processing method is highly needed to improve the accuracy of the determined predicted click rate and improve the information recommendation effect.

In this context, it is desirable to provide an information processing method, which can consider pre-ranking information of information to be recommended when obtaining a predicted click rate of the information, so as to improve the accuracy of determining the predicted click rate.

In a first aspect of embodiments of the present invention, there is provided an information processing method, including: acquiring user information of a user; obtaining a plurality of pieces of information to be recommended according to the user information, wherein the plurality of pieces of information to be recommended comprise first information to be recommended with pre-sequencing information; and acquiring a plurality of predicted click rates corresponding to the plurality of information to be recommended one by adopting a click rate prediction model according to the pre-sequencing information of the first information to be recommended, the user information and the plurality of information to be recommended.

In an embodiment of the present invention, the information to be recommended includes a plurality of pieces of first information to be recommended, and the obtaining, by using a click-through rate prediction model, a plurality of predicted click-through rates that are in one-to-one correspondence with the plurality of pieces of information to be recommended includes: dividing the plurality of pieces of first information to be recommended into at least one information interval according to the pre-ordering information of the plurality of pieces of first information to be recommended to obtain a plurality of pieces of interval information which are in one-to-one correspondence with the plurality of pieces of first information to be recommended, wherein the plurality of pieces of interval information are used for representing the information intervals to which the plurality of pieces of first information to be recommended belong; according to the plurality of pieces of first information to be recommended and the plurality of pieces of interval information, obtaining a plurality of pieces of first input information which correspond to the plurality of pieces of first information to be recommended one by one, wherein the first input information is obtained by splicing one piece of first information to be recommended and the interval information corresponding to the one piece of first information to be recommended; and inputting the user information, the other information to be recommended in the plurality of information to be recommended except the first information to be recommended and the plurality of first input information into the click rate prediction model, and acquiring a plurality of predicted click rates corresponding to the plurality of information to be recommended one by one.

In another embodiment of the present invention, dividing the plurality of pieces of first information to be recommended into at least one information interval includes: and dividing the plurality of pieces of first information to be recommended into at least one information interval by adopting an entropy-based discretization method according to the pre-ordering information of the plurality of pieces of first information to be recommended.

In another embodiment of the present invention, the information processing method further includes: obtaining a plurality of sample data, wherein at least one sample data in the plurality of sample data comprises recommended information, interval information corresponding to the recommended information, clicked information of the recommended information and the user information, and the clicked information of the recommended information is used for representing whether the recommended information is clicked by the user; and taking the plurality of sample data as the input of the click rate prediction model, and optimally training the click rate prediction model by adopting a preset optimization algorithm. Wherein the click rate prediction model comprises a logistic regression model, a decision tree model or a gradient lifting tree model.

In another embodiment of the present invention, acquiring a plurality of pieces of information to be pushed according to the user information includes: acquiring the plurality of pieces of first information to be recommended by adopting a recall model according to the user information; and the first input information is obtained by splicing a piece of first information to be recommended, interval information corresponding to the piece of first information to be recommended and source information of the piece of first information to be recommended, wherein the source information is used for representing a recall model adopted for obtaining the first information to be recommended. Wherein the recall model comprises at least one of a matrix decomposition recall model, a collaborative filtering recall model, and a neural network recall model.

In another embodiment of the present invention, the obtaining the plurality of pieces of information to be recommended further includes: and acquiring the second information to be recommended according to a preset recall rule. The predetermined recall rule comprises: at least one of a hotspot recall rule, a regional recall rule, and an incident recall rule.

In still another embodiment of the present invention, before obtaining the plurality of predicted click rates, the information processing method further includes: determining cross information of the user information and the plurality of pieces of information to be recommended according to the user information and the plurality of pieces of information to be recommended; and obtaining the plurality of predicted click rates comprises: and acquiring a plurality of predicted click rates corresponding to the plurality of information to be recommended one by adopting a click rate prediction model according to the pre-sequencing information of the first information to be recommended, the user information, the plurality of information to be recommended and the cross information.

In still another embodiment of the present invention, the information processing method further includes: recommending information to be recommended to the user according to the plurality of predicted click rates, wherein the recommending information comprises: sequencing the plurality of information to be recommended in sequence according to the size of the predicted click rate corresponding to one; and recommending the information to be recommended arranged at a preset position to the user.

In a second aspect of embodiments of the present invention, there is provided an information processing apparatus comprising: the user information acquisition module is used for acquiring user information of a user; the recommendation information acquisition module is used for acquiring a plurality of pieces of information to be recommended according to the user information, wherein the plurality of pieces of information to be recommended comprise first information to be recommended with pre-sequencing information; and the click rate obtaining module is used for obtaining a plurality of predicted click rates which are in one-to-one correspondence with the plurality of information to be recommended by adopting a click rate prediction model according to the pre-sequencing information of the first information to be recommended, the user information and the plurality of information to be recommended.

In an embodiment of the present invention, the information to be recommended includes a plurality of pieces of first information to be recommended, and the click rate obtaining module includes: the information interval dividing submodule is used for dividing the first information to be recommended into at least one information interval according to the pre-ordering information of the first information to be recommended to obtain a plurality of interval information which is in one-to-one correspondence with the first information to be recommended, and the interval information is used for representing the information interval to which the first information to be recommended belongs; the first input information acquisition submodule is used for acquiring a plurality of pieces of first input information which correspond to the plurality of pieces of first information to be recommended one by one according to the plurality of pieces of first information to be recommended and the plurality of pieces of interval information, and the first input information is obtained by splicing one piece of first information to be recommended and the interval information corresponding to the one piece of first information to be recommended; and the predicted click rate obtaining sub-module is used for inputting the user information, the other information to be recommended except the first information to be recommended in the plurality of information to be recommended and the plurality of first input information into the click rate prediction model, and obtaining a plurality of predicted click rates corresponding to the plurality of information to be recommended one by one.

In another embodiment of the present invention, the information interval division submodule is specifically configured to: and dividing the plurality of pieces of first information to be recommended into at least one information interval by adopting an entropy-based discretization method according to the pre-ordering information of the plurality of pieces of first information to be recommended.

In another embodiment of the present invention, the information processing apparatus further includes: the system comprises a sample data acquisition module, a recommendation processing module and a recommendation processing module, wherein the sample data acquisition module is used for acquiring a plurality of sample data, at least one sample data in the sample data comprises recommended information, interval information corresponding to the recommended information, clicked information of the recommended information and user information, and the clicked information of the recommended information is used for representing whether the recommended information is clicked by the user; and the prediction model optimization module is used for taking the plurality of sample data as the input of the click rate prediction model and optimizing and training the click rate prediction model by adopting a preset optimization algorithm. Wherein the click rate prediction model comprises a logistic regression model, a decision tree model or a gradient lifting tree model.

In still another embodiment of the present invention, the recommendation information obtaining module includes a first information obtaining sub-module: and the information recommendation module is used for acquiring the plurality of pieces of first information to be recommended by adopting a recall model according to the user information. The first input information is obtained by splicing first information to be recommended, interval information corresponding to the first information to be recommended and source information of the first information to be recommended, and the source information is used for representing a recall model adopted for obtaining the first information to be recommended. Wherein the recall model comprises at least one of a matrix decomposition recall model, a collaborative filtering recall model, and a neural network recall model.

In a further embodiment of the present invention, the plurality of information to be recommended further includes second information to be recommended, and the recommendation information obtaining module further includes a second information obtaining sub-module, configured to obtain the second information to be recommended according to a predetermined recall rule. Wherein the predetermined recall rule comprises: at least one of a hotspot recall rule, a regional recall rule, and an incident recall rule.

In still another embodiment of the present invention, the information processing apparatus further includes: and the cross information determining module is used for determining cross information of the user information and the plurality of pieces of information to be recommended according to the user information and the plurality of pieces of information to be recommended before the click rate obtaining module obtains the plurality of predicted click rates. The click rate obtaining module is specifically configured to: and acquiring a plurality of predicted click rates corresponding to the plurality of information to be recommended one by adopting a click rate prediction model according to the pre-sequencing information of the first information to be recommended, the user information, the plurality of information to be recommended and the cross information.

In a further embodiment of the present invention, the information processing apparatus further includes an information recommending module, configured to recommend information to be recommended to the user according to the plurality of predicted click rates. Specifically, the information recommendation module includes: the information sorting submodule is used for sequentially sorting the information to be recommended according to the size of the predicted click rate corresponding to one; and the information recommending submodule is used for recommending the information to be recommended arranged at the preset position to the user.

In a third aspect of embodiments of the present invention, there is provided a computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform the information processing method provided according to the first aspect of embodiments of the present invention.

In a fourth aspect of embodiments of the present invention, a computing device is provided. The computing device includes one or more memories storing executable instructions, and one or more processors. The processor executes the executable instructions to implement the information processing method provided according to the first aspect of the embodiment of the present invention.

According to the information processing method, the device, the medium and the computing equipment, when the click rate of the information to be recommended is predicted by adopting the click rate prediction model, the pre-sequencing information of the first information to be recommended can be considered, so that the consideration factors in the click rate prediction can be enriched, and the accuracy of the obtained predicted click rate is improved. Therefore, the recommendation effect of information recommendation according to the predicted click rate can be improved, and the user experience is improved.

Drawings

The above and other objects, features and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description read in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which:

fig. 1 schematically illustrates an application scenario of an information processing method, apparatus, medium, and computing device according to embodiments of the present invention;

fig. 2 schematically shows a flowchart of an information processing method according to a first embodiment of the present invention;

fig. 3A schematically shows a flow chart of an information processing method according to a second embodiment of the present invention;

FIG. 3B schematically illustrates a flow chart for recommending information to a user based on a plurality of predicted click rates;

FIG. 4 is a flowchart schematically illustrating obtaining a predicted click rate corresponding to information to be recommended according to an embodiment of the present invention;

fig. 5 schematically shows a flowchart of an information processing method according to a third embodiment of the present invention;

fig. 6 schematically shows a flowchart of an information processing method according to a fourth embodiment of the present invention;

FIG. 7 is a flow architecture diagram schematically illustrating an information processing method according to an embodiment of the present invention;

fig. 8 schematically shows a block diagram of an information processing apparatus according to an embodiment of the present invention;

FIG. 9 schematically shows a program product adapted to perform an information processing method according to an embodiment of the invention; and

fig. 10 schematically shows a block diagram of a computing device adapted to perform an information processing method according to an embodiment of the present invention.

In the drawings, like or corresponding reference characters designate like or corresponding parts.

Detailed Description

The principles and spirit of the present invention will be described with reference to a number of exemplary embodiments. It is understood that these embodiments are given solely for the purpose of enabling those skilled in the art to better understand and to practice the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

As will be appreciated by one skilled in the art, embodiments of the present invention may be embodied as a system, apparatus, device, method, or computer program product. Accordingly, the present disclosure may be embodied in the form of: entirely hardware, entirely software (including firmware, resident software, micro-code, etc.), or a combination of hardware and software.

According to an embodiment of the invention, an information processing method, an information processing device, an information processing medium and a computing device are provided.

In this context, it is to be understood that the terms referred to are to be interpreted as follows:

logistic Regression (Logistic Regression), a machine learning method used to solve the classification problem, is used to estimate the probability of something. Such as the likelihood of an advertisement being clicked on by a user, etc. "probability" is used herein rather than mathematical "probability". The result of logistic regression is not a probability value in the mathematical definition and cannot be directly used as a probability value. The result is often used for weighted summation with other eigenvalues rather than direct multiplication. Both logistic Regression and Linear Regression (Linear Regression) are generalized Linear models. Logistic regression assumes that the dependent variable v obeys a bernoulli distribution, while linear regression assumes that the dependent variable y obeys a gaussian distribution. Thus, there are many similarities to linear regression, and the logistic regression algorithm is a linear regression except for the Sigmoid mapping function. It can be said that the logistic regression is theoretically supported by linear regression, but the logistic regression introduces nonlinear factors through Sigmoid function, so that the classification problem can be easily handled.

Click-through Rate (CTR), which refers to the ratio of the number of times that a given content on a website or an application is clicked to the number of times that the content is exposed to light, is an important index for measuring the recommendation effect in a recommendation system.

Matrix Decomposition (MF), which decomposes a Matrix into a product of several matrices, and the Matrix Decomposition may include Decomposition methods such as triangle Decomposition, full rank Decomposition, QR (orthogonal triangle) Decomposition, Jordan Decomposition, and Singular Value Decomposition (SVD), wherein a common Matrix Decomposition method includes: triangle decomposition, QR decomposition, and singular value decomposition.

In Collaborative Filtering (CF), simply speaking, interested information is recommended to a user by using the preferences of groups with mutual interests and common experiences, and individuals give responses (such as scores) to the information to a considerable extent through a Collaborative mechanism and record the responses so as to achieve the purpose of Filtering, thereby helping others to filter the information. Wherein, the response is not necessarily limited to the particular interest, and the record of the response of the particular uninteresting information is also important. Collaborative filtering can be further divided into rating (rating) and population filtering (social filtering).

A neural network is a machine learning technology which simulates the neural network of the human brain and can realize artificial intelligence. The neural network comprises: the input layer, the hidden layer and the output layer, when the network is designed, the number of nodes of the input layer and the output layer is fixed, and the hidden layer can be freely appointed. Each layer is composed of neurons, which are a model containing input, output, and computational functions.

MDLP (minimum description length principle) feature discretization: the method is applied to a continuous characteristic supervision discretization method, and a data segmentation point is searched in an information gain mode.

Moreover, it is to be understood that the number of any elements in the figures are intended to be illustrative rather than restrictive, and that any nomenclature is used solely for differentiation and not intended to be limiting.

The principles and spirit of the present invention are explained in detail below with reference to several representative embodiments of the invention.

Summary of The Invention

In the information to be recommended acquired by adopting the recall model, the pre-ranking information can represent the interest degree of the user in the information to be recommended to a certain extent. In the prior art, pre-ranking information is often not considered when the predicted click rate of the information to be recommended is determined in the ranking stage. Therefore, when information is recommended according to the sorting result in the sorting stage, pre-sorting information is undoubtedly lost during sorting, so that the recommendation efficiency is not ideal, and the performance of the recommendation system is low. The inventor finds that if the recall model is fused with the click rate prediction model through a fusion method, the pre-ranking information of the information to be recommended can be effectively considered when the click rate is predicted, so that the click rate prediction accuracy is improved, and the information recommendation effect is improved.

Having described the general principles of the invention, various non-limiting embodiments of the invention are described in detail below.

Application scene overview

Reference is first made to fig. 1.

Fig. 1 schematically illustrates an application scenario of an information processing method, apparatus, medium, and computing device according to embodiments of the present invention. It should be noted that fig. 1 is only an example of an application scenario in which the embodiment of the present invention may be applied to help those skilled in the art understand the technical content of the present invention, and does not mean that the embodiment of the present invention may not be applied to other devices, systems, environments or scenarios.

As shown in fig. 1, the application scenario 100 includes terminal devices 111, 112, 113, a network 120, and a database 130. Network 120 serves as a medium for providing communication links between terminal devices 111, 112, 113 and database 130. Network 120 may include various connection types, such as wireless communication links or fiber optic cables, among others.

The terminal devices 111, 112, 113 may interact with the database 130 through the network 120 in response to the user's operation to acquire information to be recommended that is recommended to the user. Various client applications may be installed on the terminal devices 111, 112, 113, such as a web browser application, a news-browsing type application, a search type application, social platform software, etc. (by way of example only).

The terminal devices 111, 112, 113 may be various electronic devices having display screens and supporting web browsing to present recommendation information to the user. The terminal devices 111, 112, 113 include, but are not limited to, smart phones, tablet computers, laptop portable computers, desktop computers, and the like.

The terminal devices 111, 112, and 113 may further have a processing function, for example, to process the information to be recommended acquired from the database 130, so as to obtain a predicted click rate of the information to be recommended; and sorting the information to be recommended according to the predicted click rate so as to recommend the information which is interested by the user to the user according to a sorting result.

According to an embodiment of the present invention, as shown in fig. 1, the application scenario 100 may further include a server 140, for example. The server 140 may be a server providing various services, such as a background management server (for example only) providing support for information recommended to the user by the terminal devices 111, 112, 113. Accordingly, the database 130 may be, for example, a database integrated in the server 140.

According to the embodiment of the present invention, the server 140 may further process the information to be recommended in response to the request of the terminal devices 111, 112, and 113, for example, to obtain the predicted click rate of the information to be recommended. And sorting the information to be recommended according to the predicted click rate so as to determine the information to be recommended according to a sorting result. And finally, feeding back the information needing to be recommended to the terminal equipment 111, 112 and 113 for the terminal equipment to display to the user.

It should be noted that the information processing method provided by the embodiment of the present disclosure may be generally executed by the terminal devices 111, 112, 113 or the server 140. Accordingly, the information processing apparatus provided by the embodiment of the present invention may be generally provided in the terminal device 111, 112, 113 or the server 140. The information processing method provided by the embodiment of the present invention may also be executed by a server or a server cluster that is different from the server 140 and is capable of communicating with the terminal devices 111, 112, 113 and/or the server 140. Accordingly, the information processing apparatus provided in the embodiment of the present invention may also be provided in a server or a server cluster different from the server 140 and capable of communicating with the terminal devices 111, 112, 113 and/or the server 140.

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

Exemplary method

An information processing method according to an exemplary embodiment of the present invention is described below with reference to fig. 2 to 7 in conjunction with the application scenario of fig. 1. It should be noted that the above application scenarios are merely illustrated for the convenience of understanding the spirit and principles of the present invention, and the embodiments of the present invention are not limited in this respect. Rather, embodiments of the present invention may be applied to any scenario where applicable.

Fig. 2 schematically shows a flowchart of an information processing method according to a first embodiment of the present invention.

As shown in fig. 2, the information processing method according to the first embodiment of the present invention includes operations S210 to S230. The information processing method may be executed by, for example, the terminal devices 111, 112, 113 or the server 140 in fig. 1. The operations S210 to S230 may be specifically used to implement the information recall in the information recall phase and the click rate prediction in the ranking phase of the information recommendation system.

In operation S210, user information of a user is acquired.

According to an embodiment of the present invention, the operation S210 may specifically be, for example: and acquiring user information corresponding to the account information from a server or a database according to the account information input when the user logs in the application program installed in the terminal equipment. The user information may include, for example, user basic information (age, gender) and/or user preference information (which may be a pre-selected type of information: sports, finance, and/or entertainment, etc.), etc.

In operation S220, according to the user information, a plurality of pieces of information to be recommended are obtained, where the plurality of pieces of information to be recommended include first information to be recommended having pre-ranking information.

According to the embodiment of the present invention, the operation of acquiring the first information to be recommended in operation S220 may be, for example: and acquiring the first information to be recommended by adopting a recall model according to the user information. The recall model may include, for example, a matrix decomposition recall model, a collaborative filtering recall model, and/or a neural network recall model. The obtaining of the first to-be-recommended information by using the recall model may specifically include: and taking the user information as the input of the recall model, acquiring recommendation information corresponding to the user information in the massive recommendation information from the database 130 or the server 140 by adopting a matrix decomposition method, a collaborative filtering method and/or a neural network, and taking the acquired recommendation information as the first information to be recommended.

The recall model takes user information into consideration when acquiring the information to be recommended, so that the plurality of pieces of first information to be recommended acquired by each model have accurate ranking scores corresponding to each model. The ranking score can be used for representing the matching degree of the first information to be recommended and the user information obtained by adopting each model, and the higher the matching degree is, the higher the ranking score is. The pre-ranking information may be, for example, the ranking score and/or the ranking position ranked from high to low according to the precise ranking scores of the plurality of pieces of first information to be recommended. For example, if the ranking position corresponding to the information with the highest ranking score in the plurality of pieces of first information to be recommended is position 1, the ranking position in the pre-ranking information may be represented as Order 1, for example.

According to the embodiment of the present invention, the obtained plurality of pieces of information to be recommended may further include, for example, second information to be recommended obtained by a rule recall method. The method for acquiring the second information to be recommended in operation S220 may further include: and acquiring recommendation information matched with the recall rule from the massive recommendation information of the database 130 or the server 140 according to the preset recall rule, and taking the acquired recommendation information matched with the recall rule as the second information to be recommended. The predetermined recall rule may include, for example: hotspot recall rules, geographic recall rules, and/or incident recall rules, among others.

The recommendation information acquired according to the hotspot recall rule may be, for example, information that the click rate is higher than a predetermined click rate in a predetermined time. The recommendation information acquired according to the region recall rule may be, for example, information that a described event occurs in a predetermined region. The recommendation information acquired according to the emergency recall rule may be, for example, information including keywords that can characterize an emergency, such as "earthquake", "explosion", or "debris flow". It is to be understood that the types of the recommendation information obtained according to the rule recall method are only examples to facilitate understanding of the present invention, and the present invention is not limited thereto.

In operation S230, according to the pre-ranking information of the first information to be recommended, the user information, and the plurality of information to be recommended, a click rate prediction model is used to obtain a plurality of predicted click rates corresponding to the plurality of information to be recommended one by one.

According to an embodiment of the present invention, the operation S230 may specifically be, for example: and simultaneously inputting the pre-sequencing information, the user information and the plurality of pieces of information to be recommended of the first information to be recommended into a click rate prediction model, and calculating by the click rate prediction model to obtain the predicted click rate of each piece of information to be recommended of the plurality of pieces of information to be recommended. The click rate prediction model may include, for example, a logistic regression model, a decision tree model, or a gradient boosting tree model.

According to the embodiment of the invention, in order to facilitate obtaining the predicted click rate by the click rate prediction model, the first information to be recommended can be in one-to-one correspondence with the pre-ranking information and the recall model for obtaining the first information to be recommended. Operation S230 may specifically include: splicing first information to be recommended, pre-sequencing information of the first information to be recommended and identification information of a recall model for acquiring the first information to be recommended to form input information; and then inputting the input information formed by splicing, other information to be recommended except the first information to be recommended in the information to be recommended and the user information into a click rate prediction model, and calculating to obtain a plurality of predicted click rates which are in one-to-one correspondence with the information to be recommended.

According to an embodiment of the present invention, the operation S230 may specifically determine, for example, through operations S431 to S433 described in fig. 4, a plurality of predicted click rates corresponding to a plurality of pieces of information to be recommended, which is not described in detail herein.

In summary, in the information processing method according to the embodiment of the present invention, when predicting the click rate of the information to be recommended, the pre-ranking information of the first information to be recommended, which is obtained by recalling the model, may be considered at the same time. Therefore, the accuracy of the predicted click rate of the first to-be-recommended information obtained by the click rate prediction model can be improved. The recommendation effect when recommending information to the user according to the predicted click rate is improved, and the user experience is improved.

FIG. 3A schematically shows a flowchart of an information processing method according to a second embodiment of the present invention, and FIG. 3B schematically shows a flowchart of recommending information to a user according to a plurality of predicted click rates.

According to the embodiment of the invention, after the plurality of predicted click rates of the plurality of information to be recommended are obtained through the operation S210 to the operation S230, the plurality of information to be recommended can be selected according to the plurality of predicted click rates, and the information recommended to the user is obtained through selection. Therefore, as shown in fig. 3A, the information processing method according to the second embodiment of the present invention may further include operation S340 in addition to operation S210 to operation S230. The operation S340 is performed after the operation S230.

In operation S340, information to be recommended is recommended to the user according to the plurality of predicted click rates.

According to an embodiment of the present invention, the operation S340 may specifically include, for example: determining information to be recommended to a user according to a plurality of predicted click rates; and then the information to be recommended to the user is displayed to the user through the terminal equipment 111, 112 and 113. In operation S340, the information to be recommended corresponding to the predicted click rate greater than the predetermined click rate may be determined as the information to be recommended to the user.

According to an embodiment of the present invention, as shown in fig. 3B, the operation S340 may further include operations S341 to S342. In operation S341, sequentially sorting the plurality of pieces of information to be recommended according to the one-to-one correspondence predicted click rate; in operation S342, information to be recommended arranged at a predetermined position is recommended to the user.

According to the embodiment of the invention, when the click rate prediction model obtains the predicted click rates of a plurality of pieces of information to be recommended, the user information is considered, so that the obtained information to be recommended with a large predicted click rate is usually the information with a high matching degree with the user information, namely the information to be recommended with a large predicted click rate is the information which is interested by the user. Therefore, in operation S341, information to be recommended with a high predicted click rate (i.e., with interest to the user) is ranked at a front position, and information to be recommended with a low predicted click rate (i.e., with no interest to the user) is ranked at a rear position. The operation S342 may specifically be that the information to be recommended ranked in the top n positions is presented to the user through the terminal device 111, 112, 113 as the information to be recommended to the user. The predetermined positions are the first n positions, n may be any positive integer value such as 5, 10, 12, and the like, and the value of n may be specifically set according to a user requirement, for example.

According to an embodiment of the present invention, in order to reduce resource consumption as much as possible, the operations S210 to S230 may be performed periodically, for example, with the first time period as a cycle. And operations S341 to S342 may be performed in response to a user' S acquisition request. The first time period may be, for example, one day, 12 hours, or 6 hours. The user's acquisition request may be generated in response to a user's operation of "sliding down" a page or clicking a "refresh" control while the user browses recommendation information using an application on the terminal device 111, 112, 113, for example.

According to the embodiment of the present invention, when the execution frequency of operations S341 to S342 is higher than the execution frequency of operations S210 to S230, when operation S341 is executed for the first time after execution of operations S210 to S230, for example, all the information to be recommended acquired through operation S220 may be sorted. When the operation S341 is subsequently performed, the information to be recommended, except the information to be recommended already recommended to the user, in all the information to be recommended acquired through the operation S220 is sorted.

FIG. 4 is a flowchart schematically illustrating obtaining a predicted click rate corresponding to information to be recommended according to an embodiment of the present invention.

According to the embodiment of the invention, the ranking scores of the first information to be recommended acquired by adopting the recall model aiming at different users are considered to be possibly in different intervals. Therefore, if the click rate prediction model is directly input by using the ranking score, the accuracy of the click rate prediction cannot be effectively improved, and the recommendation performance is not obviously improved. Therefore, the embodiment of the present invention may preferably use the ranking position obtained according to the ranking score as the pre-ranking information.

Moreover, in order to reduce the influence of the abnormal ranking score on the click rate prediction model, the fitting effect is improved, and the robustness of the recommendation system based on the click rate prediction model is improved. Under the condition that the first information to be recommended is multiple, the multiple pieces of first information to be recommended can be discretized into a plurality of information intervals according to the ranking positions and/or ranking scores of the multiple pieces of first information to be recommended. Accordingly, as shown in fig. 4, operation S230 in fig. 2 may specifically include operations S431 to S433.

In operation S431, according to the pre-ranking information of the first information to be recommended, the first information to be recommended is divided into at least one information interval, so as to obtain a plurality of interval information corresponding to the first information to be recommended one by one. The information of the plurality of intervals is used for representing the information intervals to which the plurality of pieces of first information to be recommended belong respectively.

According to an embodiment of the present invention, operation S431 may specifically be, for example: according to the sorting positions represented by the pre-sorting information of the plurality of pieces of first information to be recommended, the pieces of first information to be recommended with the sorting positions close to each other are divided into the same information interval. For example, according to the principle that the ranking positions are divided equally, the first information to be recommended with the ranking positions of 1-5 is divided into the same information interval, the first information to be recommended with the ranking positions of 6-10 is divided into the same information interval, and so on, at least one information interval is obtained. Or, in operation S431, the score intervals of the ranking scores represented by the pre-ranking information of the first pieces of information to be recommended may be divided first, and then the first pieces of information to be recommended, of which the ranking scores belong to the same score interval, are divided into the same information interval. Alternatively, the operation S431 may also divide the information interval of the first information to be recommended by comprehensively considering the ranking position and the ranking score.

According to an embodiment of the present invention, in order to better reflect the authenticity of the data of the pre-ordering information, operation S431 may specifically be, for example: according to the pre-ordering information of the first information to be recommended, the first information to be recommended is divided into at least one information interval by adopting a discretization method based on entropy (or based on information gain). The entropy-based discretization method can specifically adopt a similar idea as a decision tree model, and a synthesis method or a splitting method is used for determining synthesis or classification according to entropy calculation and preset judgment. According to the embodiment of the present invention, the entropy-based discretization method may specifically be, for example, an MDLP discretization method, so as to find the dividing point of each information interval in a manner of information gain.

According to the embodiment of the present invention, the section information corresponding to the first information to be recommended obtained in operation S431 may specifically be, for example, a section number representing an information section to which the first information to be recommended belongs. Accordingly, while the plurality of first information to be recommended is divided into at least one information section in operation S431, a section number may be further allocated to the at least one information section.

In operation S432, a plurality of first input information corresponding to a plurality of first information to be recommended one by one is obtained according to the plurality of first information to be recommended and the plurality of interval information.

According to the embodiment of the invention, in order to facilitate that when the click rate prediction model obtains the predicted click rate, the first information to be recommended and the interval information of the first information to be recommended can be in one-to-one correspondence, and the first input information corresponding to each piece of first information to be recommended can be obtained by splicing the first information to be recommended and the interval information corresponding to the first information to be recommended.

According to the embodiment of the invention, it is considered that the pre-ranking information of the first information to be recommended, which is recorded with the same content and obtained by using different recall models, may be different, and therefore the first information to be recommended, which is recorded with the same content and obtained by using different recall models, may have different section information. Therefore, in order to completely express each piece of first information to be recommended, the first input information of each piece of first information to be recommended may be specifically obtained by splicing each piece of first information to be recommended, the interval information corresponding to each piece of first information to be recommended, and the source information of each piece of first information to be recommended. The source information is used for characterizing a recall model adopted when each piece of first information to be recommended is obtained, and specifically, the source information may be, for example, identification information of the recall model. Therefore, when two pieces of first information to be recommended, in which the same content is recorded, are obtained through two different recall models, because the two pieces of source information of the first information to be recommended, in which the same content is recorded, are different, the two pieces of first information to be recommended, in which the same content is recorded, are two different pieces of first information to be recommended, and thus two different pieces of first input information can be obtained.

According to the embodiment of the present invention, after the first input information is obtained, operation S433 may be executed to input the user information, the information to be recommended, except for the first information to be recommended, of the information to be recommended, and the first input information into the click rate prediction model, and obtain a plurality of predicted click rates corresponding to the information to be recommended one by one.

In summary, when the predicted click rate of the first information to be recommended is determined, the first information to be recommended is divided into a plurality of information intervals according to the pre-ranking information, and the interval information of the information interval to which the first information to be recommended belongs is used as a feature to be input into the click rate prediction model, so that the influence of abnormal pre-ranking information on a prediction result can be avoided, the accuracy of the determined predicted click rate of the first information to be recommended is further improved, and the information recommendation effect and the user experience based on the predicted click rate are further improved.

Fig. 5 schematically shows a flowchart of an information processing method according to a third embodiment of the present invention.

According to the embodiment of the invention, before the click rate of the information to be recommended is predicted, an initial click rate prediction model needs to be trained. Further, the click rate prediction model can be optimized after click rate prediction is performed. Therefore, as shown in fig. 5, the information processing method according to the third embodiment of the present invention may further include operations S550 to S560 in addition to the operations S210 to S230. The operations S550 to S560 may be performed before the operations S210 to S230 or after the operations S210 to S230.

In operation S550, a plurality of sample data is acquired.

The plurality of sample data are input data of a click rate prediction model. The plurality of sample data should include information to be recommended, i.e., recommended information, which has been recommended to the user through operation S230, or recommended information, which has been recommended to the user through an existing information processing method. It is understood that each sample data in the plurality of sample data should further include clicked information, and the clicked information is used for characterizing whether recommended information included in the sample data is clicked by the user. Specifically, the plurality of sample data may be tagged with corresponding clicked information. For example, when the recommended information included in the clicked information representation sample data is clicked by the user, the label of the sample data may be 1; and when the recommended information included in the clicked information representation sample data is not clicked by the user, the label of the sample data can be-1.

According to an embodiment of the present invention, in order to enable the training of the optimized click rate prediction model to consider the pre-ranking information of the first information to be recommended, at least one sample data of the plurality of sample data should include: recommended information, section information corresponding to the recommended information, clicked information of the recommended information, and user information. The recommended information included in the at least one sample datum should be the recommended information obtained through the recall model, that is, the first information to be recommended described in operation S220. The section information corresponding to the recommended information may be determined through operation S431 described in fig. 4, and is not described herein again.

In operation S560, a predetermined optimization algorithm is used to optimize and train the click rate prediction model by using a plurality of sample data as input of the click rate prediction model.

According to an embodiment of the present invention, the operation S560 may specifically be: inputting a plurality of sample data into the click rate prediction model, and respectively obtaining the predicted click rates of the recommended information included in the plurality of sample data through the click rate prediction model. Comparing the predicted click rate with clicked information contained in a plurality of sample data, and calculating through a loss function to obtain a loss value of the click rate prediction model; and then, adjusting and optimizing each parameter in the click rate prediction model according to the loss value. The predetermined optimization algorithm is the loss function, and the loss function may be specifically a cross entropy loss function or the like.

According to The embodiment of The present invention, The operation S560 may further specifically optimize The click rate prediction model according to The predicted click rate of The recommended information included in The plurality of sample data by using a Forward-Backward segmentation (debos) algorithm or an ftrl (follow The regulated leader) algorithm. Preferably, an FTRL algorithm is adopted, so that on the premise of ensuring that the click rate prediction model obtained by optimization has higher precision, the sparsity of the click rate prediction model can be improved by losing certain precision.

Fig. 6 schematically shows a flowchart of an information processing method according to a fourth embodiment of the present invention.

According to the embodiment of the invention, the obtained information to be recommended and the user information are considered to have a correlation. In order to further embody the association relationship and obtain better data characteristics, the cross information of the user information and the information to be recommended can be simultaneously input when the information to be recommended, the pre-sorting information and the user information are input into the click rate prediction model. Therefore, the accuracy of the predicted click rate determined by the click rate prediction model is further improved. Therefore, as shown in fig. 6, the information processing method according to the fourth embodiment of the present invention may further include operation S670 in addition to operation S210 to operation S230. The operation S670 should be performed between the operations S220 and S230.

In operation S670, cross information between the user information and the plurality of pieces of information to be recommended is determined according to the user information and the plurality of pieces of information to be recommended.

According to an embodiment of the present invention, the cross information may specifically be determined, for example, by: and performing feature intersection on the user information and the plurality of information to be recommended by using a One-Hot vector mode. Specifically, the plurality of pieces of information to be recommended may have pieces of information to be recommended, in which a sports event is recorded, for example, and when the user information has pieces of information that the user likes sports, the pieces of information that characterize the user likes sports and the pieces of information to be recommended, in which the sports event is recorded, may be cross-merged into one cross feature through operation S670.

Accordingly, operation S230 may be implemented by operation S680 shown in fig. 6. In operation S680, according to the pre-ranking information, the user information, the multiple pieces of information to be recommended, and the cross information of the first information to be recommended, multiple predicted click rates corresponding to the multiple pieces of information to be recommended one by one are obtained by using a click rate prediction model. Specifically, the method comprises the following steps: and calculating the predicted click rate of the plurality of information to be recommended by taking the pre-sequencing information, the user information, the plurality of information to be recommended and the cross information of the first information to be recommended as the input of a click rate prediction model.

In summary, in the information processing method according to the embodiment of the present invention, when the click rate prediction model is used to determine the predicted click rate of the information to be recommended, the cross information (i.e., cross characteristics obtained by cross) between the user information and the information to be recommended can be considered at the same time. Therefore, the accuracy of the click rate prediction model can be further improved, and the information recommendation effect and the user experience are further improved.

Fig. 7 schematically shows a flowchart architecture diagram of an information processing method according to an embodiment of the present invention.

As shown in fig. 7, in an embodiment of the present invention, the overall flow of the information processing method may include:

the method comprises the steps of firstly recalling information to be recommended from million pieces of information in a database by adopting a recall model and a preset recall rule to obtain a recommended information candidate set 1 and a recommended information candidate set 2. Wherein, the information to be recommended in the recommendation information candidate set 1 is recalled through a recall model. The information to be recommended recalled by different recall models belongs to different recommendation information candidate subsets, and the information to be recommended in the recommendation information candidate set 1 has pre-ranking information. The information to be recommended in the recommendation information candidate set 2 is recalled by a predetermined recall rule. The recall model comprises a matrix decomposition recall model, a collaborative filtering recall model and a neural network recall model. The predetermined recall rules include a hotspot recall rule, a regional recall rule, and an incident recall rule. In consideration of the fact that the recall model needs to use user information when recalling the information to be recommended, the user information can be acquired before recalling the information to be recommended. The user information may be obtained through operation S210 described in fig. 2, for example, and is not described herein again;

and then respectively fusing recommendation information and pre-sequencing information for information to be recommended in different recommendation information candidate subsets in the recommendation information candidate set 1 to obtain input information of the click rate prediction model. The fusion of the recommendation information and the pre-ranking information to obtain the input information can be specifically realized through operations S431 to S432 described in fig. 4, and details are not described here;

and finally, inputting the obtained input information, the user information and the information to be recommended included in the recommendation information candidate set 2 into a click rate prediction model, and obtaining the predicted click rate of each information to be recommended after the information to be recommended is processed by the click rate prediction model. For example, the obtained predicted click rate of the information to be recommended 1 is 0.15, the obtained predicted click rate of the information to be recommended 2 is 0.12, … …, and the obtained predicted click rate of the information to be recommended N is 0.04.

In summary, the information processing method according to the embodiment of the present invention can consider the pre-ranking information of the information to be recommended recalled by the recall model, that is, the available information generated by the recall model can be utilized, so that the accuracy of the determined predicted click rate can be improved, and thus the recommendation performance of the information recommendation system constructed based on the information processing method can be improved. Compared with a recommendation system without considering the pre-sequencing information in the prior art, the recommendation performance can be improved by 2.6%, and the online click rate of the recommended information is improved by 3%.

Exemplary devices

Having described the method of the exemplary embodiment of the present invention, next, an information processing apparatus of the exemplary embodiment of the present invention will be described with reference to fig. 8.

Fig. 8 schematically shows a block diagram of an information processing apparatus according to an embodiment of the present invention.

As shown in fig. 8, the information processing apparatus 800 may include a user information acquisition module 810, a recommendation information acquisition module 820, and a click rate acquisition module 830 according to an embodiment of the present invention. The information processing apparatus 800 can be used to implement the information processing method according to the embodiment of the present invention.

The user information acquiring module 810 is used to acquire user information of a user (operation S210).

The recommendation information obtaining module 820 is configured to obtain a plurality of pieces of information to be recommended according to the user information, where the plurality of pieces of information to be recommended include first information to be recommended having pre-ranking information (operation S220).

The click rate obtaining module 830 is configured to obtain multiple predicted click rates corresponding to multiple pieces of information to be recommended one by using a click rate prediction model according to the pre-ranking information of the first information to be recommended, the user information, and the multiple pieces of information to be recommended (operation S230).

According to an embodiment of the present invention, the information to be recommended includes a plurality of pieces of first information to be recommended. As shown in fig. 8, the click rate obtaining module 830 includes an information interval dividing sub-module 831, a first input information obtaining sub-module 832, and a predicted click rate obtaining sub-module 833. The information interval dividing sub-module 831 is configured to divide the first information to be recommended into at least one information interval according to the pre-ordering information of the first information to be recommended, so as to obtain a plurality of interval information corresponding to the first information to be recommended (operation S431). The information of the plurality of intervals is used for representing information intervals to which the plurality of pieces of first information to be recommended belong. The first input information obtaining sub-module 832 is configured to obtain a plurality of first input information corresponding to a plurality of first information to be recommended one by one according to the plurality of first information to be recommended and the plurality of interval information (operation S432). The first input information is obtained by splicing a first piece of information to be recommended and interval information corresponding to the first piece of information to be recommended. The predicted click rate obtaining sub-module 833 is configured to input the user information, the information to be recommended, except the first information to be recommended, in the information to be recommended, and the first input information into the click rate prediction model, and obtain a plurality of predicted click rates corresponding to the information to be recommended one by one (operation S433).

According to an embodiment of the present invention, the information interval division submodule 831 is specifically configured to: according to the pre-ordering information of the first information to be recommended, the first information to be recommended is divided into at least one information interval by an entropy-based discretization method.

According to an embodiment of the present invention, as shown in fig. 8, the information processing apparatus 800 further includes a sample data obtaining module 840 and a prediction model optimizing module 850. The sample data acquiring module 840 is configured to acquire a plurality of sample data (operation S550). At least one sample data in the plurality of sample data comprises recommended information, section information corresponding to the recommended information, clicked information of the recommended information and user information. And the clicked information of the recommended information is used for representing whether the recommended information is clicked by the user or not. The prediction model optimization module 850 is configured to optimize the training click rate prediction model by using a predetermined optimization algorithm, with a plurality of sample data as input of the click rate prediction model (operation S560). The click rate prediction model comprises a logistic regression model, a decision tree model or a gradient lifting tree model.

According to an embodiment of the present invention, as shown in fig. 8, the recommendation information obtaining module 820 includes a first information obtaining sub-module 821. The first information obtaining sub-module 821 is configured to obtain a plurality of pieces of first information to be recommended by using a recall model according to the user information. The first input information is obtained by splicing first information to be recommended, interval information corresponding to the first information to be recommended and source information of the first information to be recommended, and the source information is used for representing a recall model adopted for obtaining the first information to be recommended. Wherein the recall model comprises at least one of a matrix decomposition recall model, a collaborative filtering recall model, and a neural network recall model.

According to an embodiment of the present invention, the plurality of pieces of information to be recommended further includes second information to be recommended. As shown in fig. 8, the recommended information obtaining module 820 further includes a second information obtaining sub-module 822. The second information obtaining sub-module 822 is configured to obtain second information to be recommended according to a predetermined recall rule. Wherein the predetermined recall rule comprises: at least one of a hotspot recall rule, a regional recall rule, and an incident recall rule.

According to an embodiment of the present invention, as shown in fig. 8, the information processing apparatus 800 further includes a cross information determination module 860. The cross information determining module 860 is configured to determine cross information between the user information and the plurality of pieces of information to be recommended according to the user information and the plurality of pieces of information to be recommended before the click rate obtaining module 830 obtains the plurality of predicted click rates (operation S670). The click rate obtaining module 830 is specifically configured to: according to the pre-ranking information, the user information, the information to be recommended and the cross information of the first information to be recommended, a click rate prediction model is adopted to obtain a plurality of predicted click rates corresponding to the information to be recommended one by one (operation S680).

According to an embodiment of the present invention, as shown in fig. 8, the information processing apparatus 800 further includes an information recommending module 870. The information recommending module 870 is configured to recommend information to be recommended to the user according to the plurality of predicted click rates (operation S340). Specifically, the information recommendation module 870 may include an information ranking sub-module 871 and an information recommendation sub-module 872. The information sorting sub-module 871 is configured to sequentially sort the plurality of information to be recommended according to the size of the predicted click rate corresponding to one (operation S341). The information recommending sub-module 872 is configured to recommend information to be recommended, which is arranged at a predetermined position, to the user (operation S342).

Exemplary Medium

Having described the method of the exemplary embodiment of the present invention, a computer-readable storage medium suitable for executing an information processing method of the exemplary embodiment of the present invention will be described next with reference to fig. 9.

There is also provided, according to an embodiment of the present invention, a computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform an information processing method according to an embodiment of the present invention.

In some possible embodiments, aspects of the present invention may also be implemented in the form of a program product including program code for causing a computing device to perform steps of the method for performing information processing according to various exemplary embodiments of the present invention described in the above section "exemplary method" of this specification when the program product is run on the computing device, for example, the computing device may perform step S210 as shown in fig. 2: acquiring user information of a user; step S220: according to the user information, obtaining a plurality of pieces of information to be recommended, wherein the plurality of pieces of information to be recommended comprise first information to be recommended with pre-sequencing information; step S230: according to the pre-ranking information, the user information and the information to be recommended of the first information to be recommended, a click rate prediction model is adopted to obtain a plurality of predicted click rates which are in one-to-one correspondence with the information to be recommended.

The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable 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 disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

As shown in fig. 9, a program product 900 adapted to perform an information processing method according to an embodiment of the present invention is depicted, which may employ a portable compact disc read only memory (CD-ROM) and include program code, and may be executed on a computing device, such as a personal computer. However, the program product of the present invention is not limited in this regard and, in the present document, a readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

A readable signal medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable signal medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

Program code embodied on a readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user computing device, partly on a remote computing device, or entirely on the remote computing device or server. In situations involving remote computing devices, the remote computing devices may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to external computing devices (e.g., through the internet using an internet service provider).

Exemplary computing device

Having described the method, medium, and apparatus of exemplary embodiments of the present invention, a computing device adapted to perform the information processing method of exemplary embodiments of the present invention is described next with reference to fig. 10.

The embodiment of the invention also provides the computing equipment. As will be appreciated by one skilled in the art, aspects of the present invention may be embodied as a system, method or program product. Thus, various aspects of the invention may be embodied in the form of: an entirely hardware embodiment, an entirely software embodiment (including firmware, microcode, etc.) or an embodiment combining hardware and software aspects that may all generally be referred to herein as a "circuit," module "or" system.

In some possible embodiments, a computing device according to the present invention may include at least one processor, and at least one memory. Wherein the memory stores program code that, when executed by the processor, causes the processor to perform the steps in the information processing method according to various exemplary embodiments of the present invention described in the above section "exemplary method" of the present specification. For example, the processor may perform step S210 as shown in fig. 2: acquiring user information of a user; step S220: according to the user information, obtaining a plurality of pieces of information to be recommended, wherein the plurality of pieces of information to be recommended comprise first information to be recommended with pre-sequencing information; step S230: according to the pre-ranking information, the user information and the information to be recommended of the first information to be recommended, a click rate prediction model is adopted to obtain a plurality of predicted click rates which are in one-to-one correspondence with the information to be recommended.

A computing device 1000 adapted to execute the information processing method according to this embodiment of the present invention is described below with reference to fig. 10. The computing device 1000 as shown in FIG. 10 is only one example and should not be taken to limit the scope of use and functionality of embodiments of the present invention.

As shown in fig. 10, computing device 1000 is embodied in the form of a general purpose computing device. Components of computing device 1000 may include, but are not limited to: the at least one processor 1001, the at least one memory 1002, and the bus 1003 connecting the various system components (including the memory 1002 and the processor 1001).

The bus 1003 may include a data bus, an address bus, and a control bus.

The memory 1002 can include volatile memory, such as Random Access Memory (RAM)10021 and/or cache memory 10022, and can further include Read Only Memory (ROM) 1023.

Memory 1002 may also include a program/utility 10025 having a set (at least one) of program modules 10024, such program modules 10024 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.

Computing device 1000 may also communicate with one or more external devices 1004 (e.g., keyboard, pointing device, bluetooth device, etc.), which may through input/output (I/O) interface 1005. Moreover, computing device 1000 may also communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the internet) through network adapter 1006. As shown, the network adapter 1006 communicates with the other modules of the computing device 1000 over a bus 1003. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with computing device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.

It should be noted that although in the above detailed description several units/modules or sub-units/sub-modules of the apparatus are mentioned, such a division is merely exemplary and not mandatory. Indeed, the features and functionality of two or more of the units/modules described above may be embodied in one unit/module according to embodiments of the invention. Conversely, the features and functions of one unit/module described above may be further divided into embodiments by a plurality of units/modules.

Moreover, while the operations of the method of the invention are depicted in the drawings in a particular order, this does not require or imply that the operations must be performed in this particular order, or that all of the illustrated operations must be performed, to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step execution, and/or one step broken down into multiple step executions.

While the spirit and principles of the invention have been described with reference to several particular embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, nor is the division of aspects, which is for convenience only as the features in such aspects may not be combined to benefit. The invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims (16)

1. An information processing method comprising:

acquiring user information of a user;

obtaining a plurality of pieces of information to be recommended according to the user information, wherein the plurality of pieces of information to be recommended comprise first information to be recommended with pre-sequencing information; and

according to the pre-ordering information of the first information to be recommended, the user information and the plurality of information to be recommended, a click rate prediction model is adopted to obtain a plurality of predicted click rates which are in one-to-one correspondence with the plurality of information to be recommended;

the pre-ordering information comprises an ordering score and/or an ordering position obtained according to the ordering score;

the method for obtaining the plurality of predicted click rates corresponding to the plurality of information to be recommended in a one-to-one mode by adopting the click rate prediction model comprises the following steps:

dividing the plurality of pieces of first information to be recommended into at least one information interval according to the pre-ordering information of the plurality of pieces of first information to be recommended to obtain a plurality of pieces of interval information which are in one-to-one correspondence with the plurality of pieces of first information to be recommended, wherein the plurality of pieces of interval information are used for representing the information intervals to which the plurality of pieces of first information to be recommended belong;

according to the plurality of pieces of first information to be recommended and the plurality of pieces of interval information, obtaining a plurality of pieces of first input information which correspond to the plurality of pieces of first information to be recommended one by one, wherein the first input information is obtained by splicing one piece of first information to be recommended and the interval information corresponding to the one piece of first information to be recommended; and

inputting the user information, the other information to be recommended in the information to be recommended except the first information to be recommended and the first input information into the click rate prediction model, and obtaining a plurality of predicted click rates corresponding to the information to be recommended one by one.

2. The method of claim 1, wherein dividing the plurality of first to-be-recommended information into at least one information interval comprises:

and dividing the plurality of pieces of first information to be recommended into at least one information interval by adopting an entropy-based discretization method according to the pre-ordering information of the plurality of pieces of first information to be recommended.

3. The method of claim 1, further comprising:

obtaining a plurality of sample data, wherein at least one sample data in the plurality of sample data comprises recommended information, interval information corresponding to the recommended information, clicked information of the recommended information and the user information, and the clicked information of the recommended information is used for representing whether the recommended information is clicked by the user; and

taking the plurality of sample data as the input of the click rate prediction model, optimally training the click rate prediction model by adopting a preset optimization algorithm,

wherein the click rate prediction model comprises a logistic regression model, a decision tree model or a gradient lifting tree model.

4. The method of claim 1, wherein:

according to the user information, acquiring a plurality of pieces of information to be pushed comprises: acquiring the plurality of pieces of first information to be recommended by adopting a recall model according to the user information; and

the first input information is obtained by splicing first information to be recommended, interval information corresponding to the first information to be recommended and source information of the first information to be recommended, the source information is used for representing a recall model adopted for obtaining the first information to be recommended,

wherein the recall model comprises at least one of a matrix decomposition recall model, a collaborative filtering recall model, and a neural network recall model.

5. The method of claim 4, wherein the plurality of pieces of information to be recommended further includes second information to be recommended, and the obtaining the plurality of pieces of information to be pushed according to the user information further includes:

acquiring the second information to be recommended according to a preset recall rule,

the predetermined recall rule comprises: at least one of a hotspot recall rule, a regional recall rule, and an incident recall rule.

6. The method of claim 1, wherein:

before obtaining the plurality of predicted click rates, the method further comprises: determining cross information of the user information and the plurality of pieces of information to be recommended according to the user information and the plurality of pieces of information to be recommended; and

obtaining the plurality of predicted click rates comprises: and acquiring a plurality of predicted click rates corresponding to the plurality of information to be recommended one by adopting a click rate prediction model according to the pre-sequencing information of the first information to be recommended, the user information, the plurality of information to be recommended and the cross information.

7. The method of claim 1, further comprising: recommending information to be recommended to the user according to the plurality of predicted click rates, wherein the recommending information comprises:

sequencing the plurality of information to be recommended in sequence according to the size of the predicted click rate corresponding to one; and

and recommending the information to be recommended arranged at a preset position to the user.

8. An information processing apparatus comprising:

the user information acquisition module is used for acquiring user information of a user;

the recommendation information acquisition module is used for acquiring a plurality of pieces of information to be recommended according to the user information, wherein the plurality of pieces of information to be recommended comprise first information to be recommended with pre-sequencing information; and

the click rate obtaining module is used for obtaining a plurality of predicted click rates which are in one-to-one correspondence with the plurality of information to be recommended by adopting a click rate prediction model according to the pre-ranking information of the first information to be recommended, the user information and the plurality of information to be recommended;

the pre-ordering information comprises an ordering score and/or an ordering position obtained according to the ordering score;

the plurality of pieces of information to be recommended include a plurality of pieces of first information to be recommended, and the click rate obtaining module includes:

the information interval dividing submodule is used for dividing the first information to be recommended into at least one information interval according to the pre-ordering information of the first information to be recommended to obtain a plurality of interval information which is in one-to-one correspondence with the first information to be recommended, and the interval information is used for representing the information interval to which the first information to be recommended belongs;

the first input information acquisition submodule is used for acquiring a plurality of pieces of first input information which correspond to the plurality of pieces of first information to be recommended one by one according to the plurality of pieces of first information to be recommended and the plurality of pieces of interval information, and the first input information is obtained by splicing one piece of first information to be recommended and the interval information corresponding to the one piece of first information to be recommended; and

and the predicted click rate obtaining sub-module is used for inputting the user information, the other information to be recommended except the first information to be recommended in the plurality of information to be recommended and the plurality of first input information into the click rate prediction model, and obtaining a plurality of predicted click rates corresponding to the plurality of information to be recommended one by one.

9. The apparatus of claim 8, wherein the information interval partitioning submodule is to: and dividing the plurality of pieces of first information to be recommended into at least one information interval by adopting an entropy-based discretization method according to the pre-ordering information of the plurality of pieces of first information to be recommended.

10. The apparatus of claim 8, further comprising:

the system comprises a sample data acquisition module, a recommendation processing module and a recommendation processing module, wherein the sample data acquisition module is used for acquiring a plurality of sample data, at least one sample data in the sample data comprises recommended information, interval information corresponding to the recommended information, clicked information of the recommended information and user information, and the clicked information of the recommended information is used for representing whether the recommended information is clicked by the user; and

the prediction model optimization module is used for taking the plurality of sample data as the input of the click rate prediction model, optimizing and training the click rate prediction model by adopting a preset optimization algorithm,

wherein the click rate prediction model comprises a logistic regression model, a decision tree model or a gradient lifting tree model.

11. The apparatus of claim 8, wherein:

the recommendation information acquisition module comprises a first information acquisition submodule: the system comprises a plurality of pieces of first information to be recommended, a plurality of pieces of information to be recommended and a plurality of pieces of information to be recommended, wherein the information to be recommended is acquired by a recall model according to the user information;

the first input information is obtained by splicing first information to be recommended, interval information corresponding to the first information to be recommended and source information of the first information to be recommended, the source information is used for representing a recall model adopted for obtaining the first information to be recommended,

wherein the recall model comprises at least one of a matrix decomposition recall model, a collaborative filtering recall model, and a neural network recall model.

12. The apparatus of claim 11, wherein the plurality of information to be recommended further includes second information to be recommended, and the recommendation information obtaining module further includes:

a second information obtaining submodule, configured to obtain the second information to be recommended according to a predetermined recall rule,

wherein the predetermined recall rule comprises: at least one of a hotspot recall rule, a regional recall rule, and an incident recall rule.

13. The apparatus of claim 8, further comprising:

the cross information determining module is used for determining cross information of the user information and the plurality of pieces of information to be recommended according to the user information and the plurality of pieces of information to be recommended before the click rate obtaining module obtains the plurality of predicted click rates;

the click rate obtaining module is used for obtaining a plurality of predicted click rates which are in one-to-one correspondence with the plurality of information to be recommended by adopting a click rate prediction model according to the pre-ranking information of the first information to be recommended, the user information, the plurality of information to be recommended and the cross information.

14. The apparatus of claim 8, further comprising:

an information recommending module, configured to recommend information to be recommended to the user according to the plurality of predicted click rates, where the information recommending module includes:

the information sorting submodule is used for sequentially sorting the information to be recommended according to the size of the predicted click rate corresponding to one; and

and the information recommending submodule is used for recommending the information to be recommended arranged at a preset position to the user.

15. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, implement a method according to any one of claims 1 to 7.

16. A computing device, comprising:

one or more memories storing executable instructions; and

one or more processors executing the executable instructions to implement the method of any one of claims 1-7.

Images