CN115618121B

CN115618121B - Personalized information recommendation method, device, equipment and storage medium

Info

Publication number: CN115618121B
Application number: CN202211637885.7A
Authority: CN
Inventors: 冯恩宝; 陈游; 冉春雷; 陶泽胤; 易岚; 杨猛; 章芋文
Original assignee: Cloud Internet Wuhan Information Technology Co ltd
Current assignee: Cloud Internet Wuhan Information Technology Co ltd
Priority date: 2022-12-20
Filing date: 2022-12-20
Publication date: 2023-03-28
Anticipated expiration: 2042-12-20
Also published as: CN115618121A

Abstract

The embodiment of the application provides a personalized information recommendation method, device, equipment and storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: responding to a recommendation request of a target user, acquiring user data of the target user and information data of information to be processed in an information pool, and determining state data of the information to be processed; according to a preset personalized multi-channel recall strategy and user data, performing recall processing on information to be processed to obtain a first recall result containing candidate information, and inputting the first recall result into a preset sequencing model to obtain a first recommendation sequence; according to a preset popularization strategy, user data and state data, recalling information to be processed to obtain a second recall result containing candidate information; and inputting the first recommendation sequence and the second recall result into a preset mixing model, determining target candidate information, and pushing the target candidate information to a target user. The embodiment of the application can perform better personalized recommendation on the user.

Description

Personalized information recommendation method, device, equipment and storage medium

Technical Field

The present application relates to, but not limited to, the field of artificial intelligence technologies, and in particular, to a method, an apparatus, a device, and a storage medium for personalized information recommendation.

Background

In recent years, rapid development of internet technology and massive popularization of intelligent terminal equipment lead users to suffer from information overload due to explosive increase of data information; on one hand, the data information contains abundant value, but on the other hand, the data information has huge scale and uneven quality; the method is particularly important for screening out information which is interesting to the user from various complex data, and a recommendation system is generated accordingly.

At present, in an individualized information recommendation system, a recommendation mode of a two-stage architecture of multi-path recall and sorting is generally adopted, and in a sorting process, because recall results of the multi-path recall have the same weight, individualized requirements of a user are easily ignored, and better individualized recommendation cannot be performed on the user.

Disclosure of Invention

The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

The embodiment of the application provides a method, a device, equipment and a storage medium for recommending personalized information, which can better recommend the personalized information to a user.

In order to achieve the above object, a first aspect of an embodiment of the present application provides a method for recommending personalized information, including: responding to a recommendation request of a target user, and acquiring user data of the target user and information data of information to be processed in an information pool, wherein the information data comprises state data for representing a mark state; according to a preset personalized multi-channel recall strategy, the user data and the information data, recalling the information to be processed to obtain a first recall result containing candidate information, and inputting the first recall result into a preset sequencing model to obtain a first recommendation sequence; according to a preset popularization strategy, the user data and the state data, recalling the information to be processed to obtain a second recall result containing the candidate information; and inputting the first recommendation sequence and the second recall result into a preset mixing model, determining target candidate information, and pushing the target candidate information to the target user.

In some embodiments, the inputting the first recommendation sequence and the second recall result into a preset mixing model, determining target candidate information, and pushing the target candidate information to the target user includes: receiving a first weight and a second weight, wherein the sum of the first weight and the second weight is less than or equal to one; updating the model parameters of a preset mixed model according to the first weight and the second weight; inputting the first recommendation sequence and the second recall result into an updated hybrid model to obtain a target recommendation sequence, wherein the ratio of candidate information of the first recommendation sequence in the target recommendation sequence is the first weight, and the ratio of candidate information of the second recall result in the target recommendation sequence is the second weight; determining target candidate information according to the target recommendation sequence; and pushing the target candidate information to the target user.

In some embodiments, before the step of obtaining the user data of the target user and the information data of the to-be-processed information in the information pool in response to the recommendation request of the target user, the method further includes: receiving a mark updating instruction of the information to be processed; and responding to the mark updating instruction, and updating the mark state of the corresponding information to be processed.

In some embodiments, the user data includes user historical behavior data, the information data further includes information content data, and the user historical behavior data refers to implicit feedback of the user on the associated to-be-processed information; before the step of recalling the information to be processed according to the preset personalized multi-way recall strategy, the user data and the information data to obtain a first recall result containing candidate information, and inputting the first recall result into a preset ranking model to obtain a first recommendation sequence, the method comprises the following steps: determining an information content image according to a preset word frequency inverse document frequency algorithm and the information content data, wherein the information content image is used for determining the first recall result; determining historical scoring information of the historical user behavior data according to a preset scoring rule and the historical user behavior data; determining historical record time of the user historical behavior data and according to the calendarDetermining a time attenuation coefficient according to history recording time; determining a user behavior portrait according to the time attenuation coefficient and the historical scoring information, wherein the user behavior portrait is used for determining the first recall result; wherein, the calculation formula of the time attenuation coefficient is as follows:

in which>

For said time-attenuation coefficient, based on the sum of the absolute value of the time-dependent signal and the absolute value of the time-dependent signal>

And the difference value of the historical record time and the current time is obtained.

In some embodiments, the personalized multi-recall policy comprises: a collaborative filtering recall strategy, an alternating least squares recall strategy, and a topic recall strategy, the first recall result comprising: collaborative filtering recall results, alternate least squares recall results, and topic recall results; the recalling the information to be processed according to a preset personalized multichannel recalling strategy, the user data and the information data to obtain a first recalling result containing candidate information, and inputting the first recalling result into a preset sequencing model to obtain a first recommendation sequence, wherein the method comprises the following steps: determining the information similarity between any two pieces of information to be processed in the information pool according to the information content image and a preset similarity algorithm; according to the collaborative filtering recall strategy, the user historical behavior data and the information similarity, recalling the information to be processed to obtain a collaborative filtering recall result; recalling the information to be processed according to the alternative least square recall strategy and a preset information scoring matrix to obtain an alternative least square recall result; and recalling the information to be processed according to the theme recall strategy, the user data and the information data to obtain the theme recall result.

In some embodiments, the user data further comprises user preference data, and the information to be processed further comprises information subject data; the recalling the information to be processed according to the theme recall strategy, the user data and the information data to obtain the theme recall result includes: determining a matching relation between the user data and the information to be processed according to the user preference data and the information subject data, wherein the information subject data is used for representing an information subject of the information to be processed, and the user preference data is used for representing a target information subject of the target user; according to the theme recall strategy and the matching relation, the information to be processed matched with the user data is used as theme matching information; and randomly screening the theme matching information according to the preset theme recommendation quantity, and taking a screening result as a theme recall result.

In some embodiments, the user data further includes user current behavior data, and the generalized policy includes: a new information recall policy, a hit information recall policy, and a tagged recall policy, the second recall result comprising: a new information recall result, a hit information recall result, and a tagged recall result; the recalling the information to be processed according to a preset popularization strategy, the user data and the state data to obtain a second recall result containing the candidate information comprises the following steps: according to the new information recall strategy and the user historical behavior data, using information to be processed associated with the user historical behavior data as historical information to be processed, and using the information to be processed except the historical information to be processed as a new information recall result; determining the current grading information of the current behavior data of the user according to the grading rule and the current behavior data of the user; sorting the information to be processed according to the trending information recall strategy and the current grading information, and determining a trending information recall result according to a sorting result; and according to the mark recall strategy and the state data, taking the information to be processed in the target mark state as the mark recall result.

In order to achieve the above object, a second aspect of the embodiments of the present application provides a personalized information recommendation apparatus, including: the system comprises an acquisition unit, a recommendation unit and a processing unit, wherein the acquisition unit is used for responding to a recommendation request of a target user and acquiring user data of the target user and information data of information to be processed in an information pool, and the information data comprises state data used for representing a mark state; the first recall unit is used for recalling the information to be processed according to a preset personalized multi-channel recall strategy, the user data and the information data to obtain a first recall result containing candidate information, and inputting the first recall result into a preset sequencing model to obtain a first recommendation sequence; the second recall unit is used for recalling the information to be processed according to a preset popularization strategy, the user data and the state data to obtain a second recall result containing the candidate information; and the recommending unit is used for inputting the first recommending sequence and the second recalling result into a preset mixing model, determining target candidate information and pushing the target candidate information to the target user.

In order to achieve the above object, a third aspect of the embodiments of the present application provides an electronic device, where the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the personalized information recommendation method according to the first aspect when executing the computer program.

In order to achieve the above object, a fourth aspect of the embodiments of the present application provides a storage medium, where the storage medium is a computer-readable storage medium, and the storage medium stores a computer program, where the computer program is executed by a processor to implement the personalized information recommendation method according to the first aspect.

The embodiment of the application provides a personalized information recommendation method, a personalized information recommendation device and a storage medium, and the method comprises the following steps: responding to a recommendation request of a target user, and acquiring user data of the target user and information data of information to be processed in an information pool, wherein the information data comprises state data for representing a mark state; recalling the information to be processed according to a preset personalized multipath recalling strategy, the user data and the information data to obtain a first recalling result containing candidate information, and inputting the first recalling result into a preset sequencing model to obtain a first recommendation sequence; recalling the information to be processed according to a preset popularization strategy, the user data and the state data to obtain a second recalling result containing the candidate information; and inputting the first recommendation sequence and the second recall result into a preset mixing model, determining target candidate information, and pushing the target candidate information to the target user. According to the scheme provided by the embodiment of the application, after a recommendation request of a target user is responded, a first recall result is obtained by recalling information to be processed by using user data and information data based on a personalized multi-way recall strategy, a second recall result is obtained by recalling the information to be processed by using the user data and the state data of the information to be processed based on a popularization strategy, then a first recommendation sequence is obtained by sequencing the first recall result by using a sequencing model, then the first recommendation sequence and the second recall result are mixed by using a mixing model to obtain target candidate information meeting personalized requirements of the user, and the target candidate information is pushed to the target user, wherein both a mark state and the mixing model can be adjusted by the user according to the personalized requirements, and better personalized recommendation can be carried out on the user.

Additional features and advantages of the application will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by the practice of the application. The objectives and other advantages of the application may be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

Drawings

The accompanying drawings are included to provide a further understanding of the claimed subject matter and are incorporated in and constitute a part of this specification, illustrate embodiments of the subject matter and together with the description serve to explain the principles of the subject matter and not to limit the subject matter.

FIG. 1 is a flowchart illustrating a method for recommending personalized information according to an embodiment of the present application;

FIG. 2 is a flow chart of a method for updating a mixture model according to another embodiment of the present application;

FIG. 3 is a flow chart of a method for updating a flag state according to another embodiment of the present application;

FIG. 4 is a flow chart of a method of data pre-processing according to another embodiment of the present application;

FIG. 5 is a flow chart of a method of obtaining a first recall result according to another embodiment of the present application;

FIG. 6 is a flow chart of a method of obtaining subject recall results according to another embodiment of the present application;

FIG. 7 is a flow chart of a method for obtaining a second recall result according to another embodiment of the present application;

FIG. 8 is a block diagram of a personalized information recommendation system according to another embodiment of the present application;

FIG. 9 is a schematic structural diagram of a personalized information recommendation device according to another embodiment of the present application;

fig. 10 is a schematic hardware structure diagram of an electronic device according to another embodiment of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of and not restrictive on the broad application.

In the description of the present application, the meaning of a plurality is one or more, the meaning of a plurality is two or more, and greater than, less than, more than, etc. are understood as excluding the present number, and greater than, less than, etc. are understood as including the present number.

It should be noted that although functional blocks are partitioned in a schematic diagram of an apparatus and a logical order is shown in a flowchart, in some cases, the steps shown or described may be performed in a different order than the partitioning of blocks in the apparatus or the order in the flowchart. The terms "first," "second," and the like in the description, in the claims, or in the drawings described above, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.

First, several terms referred to in the present application are resolved:

artificial Intelligence (AI): is a new technical science for researching and developing theories, methods, technologies and application systems for simulating, extending and expanding human intelligence; artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produces a new intelligent machine that can react in a manner similar to human intelligence, and research in this field includes robotics, language recognition, image recognition, natural language processing, and expert systems, among others. The artificial intelligence can simulate the information process of human consciousness and thinking. Artificial intelligence is also a theory, method, technique and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results.

At present, in an information recommendation system, a recommendation mode of a two-stage architecture of multi-way recall and sorting is generally adopted, and in the sorting process, because each recall result of the multi-way recall has the same weight, the personalized demand of a user is easily ignored, and better personalized recommendation cannot be performed on the user.

Aiming at the problems that the individual requirements of users are easy to ignore and better individual recommendation cannot be carried out on the users, the application provides an individual information recommendation method, an apparatus, a device and a storage medium, wherein the method comprises the following steps: responding to a recommendation request of a target user, and acquiring user data of the target user and information data of information to be processed in an information pool, wherein the information data comprises state data for representing a mark state; according to a preset personalized multi-channel recall strategy, user data and information data, recalling information to be processed to obtain a first recall result containing candidate information, and inputting the first recall result into a preset sequencing model to obtain a first recommendation sequence; according to a preset popularization strategy, user data and state data, recalling information to be processed to obtain a second recall result containing candidate information; and inputting the first recommendation sequence and the second recall result into a preset mixing model, determining target candidate information, and pushing the target candidate information to a target user. According to the scheme provided by the embodiment of the application, after a recommendation request of a target user is responded, a first recall result is obtained by recalling information to be processed by using user data and information data based on a personalized multi-way recall strategy, a second recall result is obtained by recalling the information to be processed by using the user data and the state data of the information to be processed based on a popularization strategy, then a first recommendation sequence is obtained by sequencing the first recall result by using a sequencing model, then the first recommendation sequence and the second recall result are mixed by using a mixing model to obtain target candidate information meeting personalized requirements of the user, and the target candidate information is pushed to the target user, wherein both a mark state and the mixing model can be adjusted by the user according to the personalized requirements, and better personalized recommendation can be carried out on the user.

The method, the apparatus, the device and the storage medium for recommending personalized information provided in the embodiments of the present application are described in detail with reference to the following embodiments, which first describe a method for recommending personalized information in the embodiments of the present application.

The embodiment of the application provides a personalized information recommendation method, and relates to the technical field of artificial intelligence. The personalized information recommendation method provided by the embodiment of the application can be applied to a terminal, a server side and software running in the terminal or the server side. In some embodiments, the terminal may be a smartphone, tablet, laptop, desktop computer, or the like; the server side can be configured as an independent physical server, can also be configured as a server cluster or a distributed system formed by a plurality of physical servers, and can also be configured as a cloud server for providing basic cloud computing services such as cloud service, a cloud database, cloud computing, cloud functions, cloud storage, network service, cloud communication, middleware service, domain name service, security service, CDN (content distribution network) and big data and artificial intelligence platforms; the software may be an application for implementing a personalized information recommendation method, but is not limited to the above form.

The application is operational with numerous general purpose or special purpose computing system environments or configurations. For example: personal computers, server computers, hand-held or portable devices, tablet-type devices, multiprocessor systems, microprocessor-based systems, set top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and the like. The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

In each embodiment of the present application, when data related to the user identity or characteristic, such as user information, user behavior data, user history data, and user location information, is processed, permission or consent of the user is obtained, and the data collection, use, and processing comply with relevant laws and regulations and standards of relevant countries and regions. In addition, when the embodiment of the present application needs to acquire sensitive personal information of a user, individual permission or individual consent of the user is obtained through a pop-up window or a jump to a confirmation page, and after the individual permission or individual consent of the user is definitely obtained, necessary user-related data for enabling the embodiment of the present application to operate normally is acquired.

The embodiments of the present application will be further explained with reference to the drawings.

Referring to fig. 1, fig. 1 is a flowchart illustrating a method for recommending personalized information according to an embodiment of the present application. The personalized information recommendation method includes but is not limited to the following steps:

step S110, responding to a recommendation request of a target user, and acquiring user data of the target user and information data of information to be processed in an information pool, wherein the information data comprises state data for representing a mark state;

step S120, recalling the information to be processed according to a preset personalized multipath recalling strategy, user data and information data to obtain a first recalling result containing candidate information, and inputting the first recalling result into a preset sequencing model to obtain a first recommendation sequence;

step S130, according to a preset popularization strategy, user data and state data, recalling information to be processed to obtain a second recall result containing candidate information;

step S140, inputting the first recommended sequence and the second recall result into a preset mixing model, determining target candidate information, and pushing the target candidate information to the target user.

It can be understood that in the internet, information can refer to information such as articles, videos and commodities, and accurate recommendation of information to users is realized by using a personalized information recommendation method, for example, the method can be applied to a news recommendation scene, specifically, in a data preparation stage, user data and information data are obtained, in a recall stage, a first recommendation sequence is determined by using a two-stage architecture of a personalized multi-way recall strategy and a sequencing model, a second recall result is obtained by using a popularization strategy, then, a target candidate information is determined by mixing the first recommendation sequence and the second recall result through a mixing model, and in the recommendation stage, the target candidate information is pushed to a target user, so that personalized requirements of the user are effectively met; based on the method, after a recommendation request of a target user is responded, a first recall result is obtained by recalling information to be processed by using user data and information data based on a personalized multi-way recall strategy, a second recall result is obtained by recalling the information to be processed by using the user data and the state data of the information to be processed based on a popularization strategy, then a first recommendation sequence is obtained by sequencing the first recall result by using a sequencing model, then the first recommendation sequence and the second recall result are mixed by using a mixing model to obtain target candidate information meeting personalized requirements of the user, and the target candidate information is pushed to the target user.

The step of recalling the information to be processed to obtain a first recall result containing the candidate information means that a personalized multi-way recall strategy is utilized to perform recall processing on the information to be processed, a first batch of candidate information is determined in the information to be processed, and then the first recall result is obtained through the first batch of candidate information; recalling the information to be processed to obtain a second recalling result containing the candidate information, wherein the recalling processing is carried out on the information to be processed by utilizing the popularization type strategy, the candidate information of a second batch is determined in the information to be processed, and the second recalling result is obtained through the candidate information of the second batch.

It should be noted that the information pool is a storage space for storing a large amount of information to be processed, and the information to be processed in the information pool can be changed in real time, and the change can be adding information to be processed, deleting information to be processed, and updating information to be processed.

Notably, the user data includes, but is not limited to: a user ID, a user attention topic and user registration time; information data includes, but is not limited to: information ID, information content data, information subject, information creation time.

Referring to fig. 2, in an embodiment, step S140 in the embodiment shown in fig. 1 includes, but is not limited to, the following steps:

step S210, receiving a first weight and a second weight, wherein the sum of the first weight and the second weight is less than or equal to one;

step S220, updating the model parameters of the preset mixed model according to the first weight and the second weight;

step S230, inputting the first recommendation sequence and the second recall result into the updated hybrid model to obtain a target recommendation sequence, wherein the ratio of the candidate information of the first recommendation sequence in the target recommendation sequence is a first weight, and the ratio of the candidate information of the second recall result in the target recommendation sequence is a second weight;

step S240, determining target candidate information according to the target recommendation sequence;

step S250, pushing the target candidate information to the target user.

It can be understood that the un-updated mixed model corresponds to a preset weight, for example, before updating, the weight of the mixed model is preset to 60% of the candidate information of the first recommendation sequence and 40% of the candidate information of the second recall result, so that the first recommendation sequence and the second recall result can be effectively mixed, and the reliability of the target recommendation sequence is further ensured; in addition, the user can adjust the first weight and the second weight based on personalized requirements, after the user inputs the first weight and the second weight, the model parameters of the hybrid model are updated according to the first weight and the second weight, then in the hybrid process, the proportion of the candidate information of the first recommendation sequence can be adjusted according to the first weight, the proportion of the candidate information of the second recall result can be adjusted according to the second weight, the target recommendation sequence is determined according to the updated proportion, the target candidate information can be guaranteed to meet the personalized requirements of the user, and therefore better personalized recommendation can be conducted on the user.

It should be noted that, a user may individually adjust the first weight or the second weight according to an individual requirement, and the weight that is not adjusted may continue to use a preset weight, and only needs to satisfy that the sum of the weights is less than or equal to one, so that the adjustment efficiency can be improved.

In addition, referring to fig. 3, in an embodiment, before step S110 in the embodiment shown in fig. 1, the following steps are further included, but not limited to:

step S310, receiving a mark updating command of the information to be processed;

step S320, in response to the mark update command, updating the mark state of the corresponding to-be-processed information.

It can be understood that, the user can adjust the marking state of the information to be processed according to the personalized requirement, so as to realize the manual marking of the information to be processed, and in the process of manual marking, the user inputs the corresponding marking update instruction according to the information to be processed which needs to be adjusted, and then updates the marking state of the corresponding information to be processed according to the marking update instruction, for example, the marking state is updated to a target marking state or a non-target marking state, which is equivalent to marking the information to be processed or canceling the marking of the information to be processed; the manual marking processing is to perform a recall strategy of subsequent manual recommendation, and can ensure that target candidate information meets the personalized requirements of the user, so as to perform better personalized recommendation on the user.

In addition, referring to fig. 4, in an embodiment, the user data includes user historical behavior data, and the information data further includes information content data, where the user historical behavior data refers to implicit feedback of the user on the associated to-be-processed information; before step S120 in the embodiment shown in fig. 1, the following steps are included, but not limited to:

step S410, determining an information content image according to a preset word frequency inverse document frequency algorithm and information content data, wherein the information content image is used for determining a first recall result;

step S420, determining historical scoring information of the historical behavior data of the user according to a preset scoring rule and the historical behavior data of the user;

step S430, determining the historical recording time of the user historical behavior data, and determining a time attenuation coefficient according to the historical recording time;

step S440, determining a user behavior portrait according to the time attenuation coefficient and the historical scoring information, wherein the user behavior portrait is used for determining a first recall result;

wherein, the calculation formula of the time attenuation coefficient is as follows:

，

wherein the content of the first and second substances,

is a time decay factor, is->

The difference between the historical time and the current time.

It can be understood that, after the data preparation stage, a data processing stage is required, in the data processing stage, the information content data is processed by a Term Frequency-Inverse Document Frequency (TF-IDF) algorithm to determine the information content image; historical scoring information of the historical behavior data of the user is also determined through scoring rules, and specifically, implicit feedback referred by the historical behavior data of the user includes but is not limited to: the user behavior portrait can be accurately determined by combining historical scoring information and a time attenuation coefficient.

In specific practice, the implementation steps of the TF-IDF algorithm are as follows:

step 1, calculating word frequency through a first calculation formula, wherein the first calculation formula is as follows:

，

wherein the content of the first and second substances,

means for the entry pick>

In a document->

Is present, is present at a time>

Or that the entry +>

In a document->

The frequency of occurrence of;

step 2, calculating the inverse document frequency through a second calculation formula, wherein the second calculation formula is as follows:

，

wherein

Represents the number of all documents, and>

indicates that the word contains an entry +>

The number of documents in the document set (c),

is to prevent the inclusion of an entry->

The number of (2) is 0 and cannot be calculated;

step 3, calculating the word frequency inverse document frequency through a third calculation formula, wherein the third calculation formula is as follows:

。

as shown in FIG. 5, in one embodiment, the personalized multi-recall policy comprises: a collaborative filtering recall strategy, an alternating least squares recall strategy, and a subject recall strategy, the first recall result comprising: collaborative filtering recall results, alternate least squares recall results, and topic recall results; step S120 in the embodiment shown in fig. 1 includes, but is not limited to, the following steps:

step S510, determining the information similarity between any two pieces of information to be processed in the information pool according to the information content image and a preset similarity algorithm;

step S520, according to the collaborative filtering recall strategy, the user historical behavior data and the information similarity, recalling the information to be processed to obtain a collaborative filtering recall result;

step S530, recalling the information to be processed according to the alternative least square recall strategy and a preset information scoring matrix to obtain an alternative least square recall result;

and S540, recalling the information to be processed according to the theme recall strategy, the user data and the information data to obtain a theme recall result.

It can be understood that before the collaborative filtering recall strategy is adopted, the information similarity between different information needs to be calculated through a similarity algorithm, for example, the similarity algorithm can be an euclidean distance calculation algorithm, firstly, feature extraction is carried out on the information content images, feature vectors of the information content images are determined, then, euclidean distances between different feature vectors are calculated, the euclidean distances are used as the information similarity, the information similarity between any two pieces of information to be processed can be accurately obtained, the collaborative filtering recall strategy is adopted in combination with historical behavior data of a user, the information to be processed is sorted according to the similarity from high to low, and N pieces of information to be processed which are sorted are used as collaborative filtering recall results, wherein N is a positive integer; in addition, an Alternating least square recall (ALS) strategy is adopted, on the basis that the user scores the information matrix of the information, denser hidden vectors are used for representing the user and the information, the hidden interests and hidden characteristics of the user and the information are mined, and the problem that the capacity of a collaborative filtering model for processing the sparse matrix is insufficient is solved to a certain extent; in addition, a theme recall strategy is adopted, each piece of information has a corresponding theme, and a user also has an interested theme, so that the theme recall strategy can be used for recommending the potentially interested information for the target user; and finally, performing mixed sorting on the collaborative filtering recall result, the alternative least square recall result and the theme recall result to obtain a first recommendation sequence, so that target candidate information can be ensured to meet the personalized requirements of the user, and better personalized recommendation is performed on the user.

It should be noted that the implementation principle of ALS is to iteratively solve a series of least squares regression problems; after determining an information scoring matrix of information by a user, fixing one of a user factor matrix or an information factor matrix in each iteration, and then updating the other matrix by using the fixed matrix and scoring data; in the next iteration, fixing the matrix updated in the last iteration, and then updating another matrix; repeating the iteration process until convergence or the iteration times reach a preset value, wherein the specific implementation steps are as follows:

by using

The information scoring matrix representing the user's information can be decomposed approximately into the user's preference matrix for implicit characteristics>

（/>

Dimension) and the attribution matrix of information to implicit characteristics>

（/>

Dimension) transposed product, i.e.

The a goal of the ALS algorithm is to solve for @inthe formula above>

And &>

；

First, the degree of fit needs to be measured by constructing a cost function, ALS is quantified by the sum of the squares of the reconstruction error for each element, where the cost function is constructed as:

，

wherein the content of the first and second substances,

indicates that the user is pickand place>

For information on/off>

Is scored in->

And/or>

Respectively indicate the user->

And information->

Is hidden vector of->

Is a regularization parameter, is>

For calculating a two-norm; by setting the regularization parameters, overfitting can be prevented; in the execution of the algorithm, the matrix may be initialized at random first>

And is fixed and then paired/>

In the loss function->

The upper partial derivative is calculated to be equal to 0, so that the obtained product

，

Simple solution and can obtain

，

；

Wherein the content of the first and second substances,

is a unit matrix, is selected>

Is->

Is transposed matrix of->

Is->

Is transposed matrix of->

Is->

The transposed matrix of (2);

repeating the iteration process until convergence or the iteration times reach a preset value to obtain the final product

And/or>

Select the matrix->

Is compared with a certain row and the matrix->

Any row of the data processing method is subjected to dot product, the prediction score of a user on certain to-be-processed information is calculated, after the prediction scores of all the to-be-processed information are determined, the to-be-processed information is sorted according to the prediction scores from high to low, and J pieces of the to-be-processed information which are sorted are used as alternating least square recall results, wherein J is a positive integer.

As shown in fig. 6, in an embodiment, the user data further includes user preference data, and the information to be processed further includes information subject data; step S540 in the embodiment shown in fig. 5 includes, but is not limited to, the following steps:

step S610, determining a matching relation between the user data and the information to be processed according to the user preference data and the information theme data, wherein the information theme data is used for representing an information theme of the information to be processed, and the user preference data is used for representing a target information theme of a target user;

step S620, according to the theme recall strategy and the matching relation, the information to be processed matched with the user data is used as theme matching information;

step S630, according to the preset recommended number of topics, the random screening processing is performed on the topic matching information, and the screening result is used as the topic recall result.

It can be understood that, before the topic recall policy is adopted, a topic which is interested by a user and an information topic of information to be processed need to be determined, where a target information topic refers to a topic which is interested by the user, for any information to be processed, if the information topic is the same as the target information topic, it is equivalent to that the information to be processed is matched with user data, if the information topic is different from the target information topic, it is equivalent to that the information to be processed is not matched with the user data, after a matching relationship between all the information to be processed and the user data is determined, all topic matching information matched with the target information topic is determined through the topic recall policy, then through a random screening method, candidate information randomly screened out from the topic matching information is used as a topic recall result, the number of the candidate information is a topic recommendation number, for example, the topic recommendation number is 100, if the total number of the topic matching information exceeds 100, the number of the randomly screened candidate information is 100, and if the total number of the topic matching information does not exceed 100, all the topic matching information is used as candidate information.

As shown in fig. 7, in an embodiment, the user data further includes current behavior data of the user, and the generalized policy includes: a new information recall policy, a trending information recall policy, and a tagged recall policy, the second recall result comprising: new information recall results, hit information recall results and tagged recall results; step S130 in the embodiment shown in fig. 1 includes, but is not limited to, the following steps:

step S710, according to the new information recall strategy and the user historical behavior data, using the information to be processed related to the user historical behavior data as historical information to be processed, and using the information to be processed except the historical information to be processed as a new information recall result;

step S720, determining the current rating information of the current behavior data of the user according to the rating rule and the current behavior data of the user;

step S730, sorting the information to be processed according to the popular information recall strategy and the current grading information, and determining a popular information recall result according to a sorting result;

step S740, according to the mark recall strategy and the status data, using the information to be processed in the target mark status as the mark recall result.

It can be understood that, by using a new information recall policy, through the historical behavior data of the user, the to-be-processed information associated with the historical behavior data of the user is determined to be used as the historical to-be-processed information, then the historical to-be-processed information is filtered out from the to-be-processed information, and the remaining to-be-processed information is used as a new information recall result, which is equivalent to using the to-be-processed information which does not interact with the user in the to-be-processed information as new information; in addition, before the hot information recall strategy is adopted, current scoring information of current behavior data of the user needs to be determined through a scoring rule, then the hot information recall strategy is adopted, the information to be processed is sorted through the current scoring information, the first K pieces of information to be processed are used as hot information recall results, namely the first K pieces of information to be processed are used as hot information, N is a positive integer, the hot information is determined through the current behavior data of the user, the real-time performance of recommendation can be improved, the recommendation effect is improved, and the hot information is recommended more accurately; in addition, a mark recall strategy is adopted, the mark state of the information to be processed is determined through state data, the information to be processed in the target mark state is used as a mark recall result, which is equivalent to the information to be processed in the target mark state being used as high-quality information of an artificial mark, the mark state of the information to be processed can be set according to the individual requirements of the user, which is equivalent to executing a recall strategy of artificial recommendation, the target candidate information can be ensured to meet the individual requirements of the user, and therefore better individual recommendation is carried out on the user.

It is worth noting that in the cold-start recall process of the newly registered user, a combination mode of a new information recall strategy and a hot information recall strategy is adopted for recall, and the obtained recall result can more easily arouse the interest of the newly registered user, attract the user to browse, improve the retention rate of the user and has good recommendation effect; the new registered user means that the user behavior representation of the user is different from the existing user behavior representation.

In addition, referring to fig. 8, fig. 8 is a schematic structural diagram of a personalized information recommendation system according to another embodiment of the present application.

It can be understood that the stages that the personalized information recommendation method needs to be sequentially performed include, but are not limited to: a data preparation stage, a data processing stage, a recall stage, a sorting stage and a recommendation stage; in the data preparation stage, offline data and online data need to be prepared, wherein the offline data can refer to historical behavior data of a user, and the online data can refer to current behavior data of the user; in the data processing stage, the online data and the offline data need to be processed respectively; in the recall stage, adopting an individualized multi-path recall strategy to recall the information to be processed to obtain a first recall result, and adopting a popularization type recall strategy to recall the information to be processed to obtain a second recall result; in the sorting stage, a first recall result of the personalized multi-path recall strategy needs to be mixed and sorted to obtain a first recommendation sequence; in the recommendation stage, the first recommendation sequence and the second recall result are mixed to obtain a target recommendation sequence, and then target candidate information in the target recommendation sequence is pushed to a target user, so that the target candidate information can meet the personalized requirements of the user, and better personalized recommendation is performed on the user.

It is worth noting that the recall result of different recall strategies can be effectively and individually recommended to the user by adjusting the two-stage type general architecture of recall and sequencing; the ALS matrix decomposition method is adopted to replace the existing recall strategy based on users, and the recall efficiency is improved under the condition of limited resources; real-time recommendation is made by using the user online behavior data of the user, so that the recommendation effect is improved; the cold start recall of the newly registered user is realized, and the recommendation effect is improved; information recommendation of corresponding theme categories is made for users who focus on the theme, and the recommendation effect is improved.

In addition, referring to fig. 9, the present application further provides a personalized information recommendation apparatus 900, including:

an obtaining unit 910, configured to, in response to a recommendation request of a target user, obtain user data of the target user and information data of information to be processed in an information pool, where the information data includes state data used for representing a state of a tag;

the first recall unit 920 is configured to perform recall processing on information to be processed according to a preset personalized multi-channel recall policy, user data and information data to obtain a first recall result including candidate information, and input the first recall result into a preset ranking model to obtain a first recommendation sequence;

a second recall unit 930, configured to perform recall processing on the information to be processed according to a preset popularization policy, user data, and status data, so as to obtain a second recall result including candidate information;

the recommending unit 940 is configured to input the first recommending sequence and the second recalling result into a preset hybrid model, determine target candidate information, and push the target candidate information to the target user.

It can be understood that the specific implementation of the personalized information recommendation apparatus 900 is substantially the same as the specific implementation of the personalized information recommendation method, and is not described herein again; based on the method, after a recommendation request of a target user is responded, a first recall result is obtained by recalling information to be processed by using user data and information data based on a personalized multi-way recall strategy, a second recall result is obtained by recalling the information to be processed by using the user data and the state data of the information to be processed based on a popularization strategy, then a first recommendation sequence is obtained by sequencing the first recall result by using a sequencing model, then the first recommendation sequence and the second recall result are mixed by using a mixing model to obtain target candidate information meeting personalized requirements of the user, and the target candidate information is pushed to the target user.

In addition, referring to fig. 10, fig. 10 illustrates a hardware structure of an electronic apparatus of another embodiment, the electronic apparatus including:

the processor 1001 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits, and is configured to execute a relevant program to implement the technical solution provided in the embodiment of the present Application;

the Memory 1002 may be implemented in the form of a Read Only Memory (ROM), a static storage device, a dynamic storage device, or a Random Access Memory (RAM). The memory 1002 may store an operating system and other application programs, and when the technical solution provided by the embodiment of the present disclosure is implemented by software or firmware, the relevant program codes are stored in the memory 1002, and the processor 1001 calls the personalized information recommendation method for executing the embodiment of the present disclosure, for example, the method steps S110 to S140 in fig. 1, the method steps S210 to S250 in fig. 2, the method steps S310 to S320 in fig. 3, the method steps S410 to S430 in fig. 4, the method steps S510 to S540 in fig. 5, the method steps S610 to S630 in fig. 6, and the method steps S710 to S740 in fig. 7 described above are executed;

an input/output interface 1003 for implementing information input and output;

the communication interface 1004 is used for realizing communication interaction between the device and other devices, and can realize communication in a wired manner (for example, USB, network cable, etc.) or in a wireless manner (for example, mobile network, WIFI, bluetooth, etc.);

a bus 1005 that transfers information between the various components of the device (e.g., the processor 1001, the memory 1002, the input/output interface 1003, and the communication interface 1004);

wherein the processor 1001, the memory 1002, the input/output interface 1003, and the communication interface 1004 realize communication connection with each other inside the apparatus through the bus 1005.

An embodiment of the present application further provides a storage medium, which is a computer-readable storage medium for computer-readable storage, and the storage medium stores one or more programs, and the one or more programs are executable by one or more processors to implement the personalized information recommendation method, for example, the method steps S110 to S140 in fig. 1, the method steps S210 to S250 in fig. 2, the method steps S310 to S320 in fig. 3, the method steps S410 to S430 in fig. 4, the method steps S510 to S540 in fig. 5, the method steps S610 to S630 in fig. 6, and the method steps S710 to S740 in fig. 7 described above are performed.

The memory, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs as well as non-transitory computer executable programs. Further, the memory may include high speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory located remotely from the processor, and these remote memories may be connected to the processor through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The method, the device, the equipment and the storage medium for recommending the personalized information are characterized in that user data of a target user and information data of information to be processed in an information pool are obtained by responding to a recommendation request of the target user, wherein the information data comprise state data for representing a mark state; according to a preset personalized multi-channel recall strategy, user data and information data, recalling information to be processed to obtain a first recall result containing candidate information, and inputting the first recall result into a preset sequencing model to obtain a first recommendation sequence; recalling the information to be processed according to a preset popularization strategy, user data and state data to obtain a second recall result containing candidate information; and inputting the first recommendation sequence and the second recall result into a preset mixed model, determining target candidate information, and pushing the target candidate information to a target user. Based on the method, after a recommendation request of a target user is responded, a first recall result is obtained by recalling information to be processed by using user data and information data based on a personalized multi-way recall strategy, a second recall result is obtained by recalling the information to be processed by using the user data and the state data of the information to be processed based on a popularization strategy, then a first recommendation sequence is obtained by sequencing the first recall result by using a sequencing model, then the first recommendation sequence and the second recall result are mixed by using a mixing model to obtain target candidate information meeting personalized requirements of the user, and the target candidate information is pushed to the target user.

The embodiments described in the embodiments of the present application are for more clearly illustrating the technical solutions of the embodiments of the present application, and do not constitute a limitation to the technical solutions provided in the embodiments of the present application, and it is obvious to those skilled in the art that the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems with the evolution of technology and the emergence of new application scenarios.

Those skilled in the art will appreciate that the solutions shown in fig. 1-7 are not meant to limit embodiments of the present application and may include more or fewer steps than those shown, or may combine certain steps, or different steps.

The above-described embodiments of the apparatus are merely illustrative, wherein the units illustrated as separate components may or may not be physically separate, i.e. may be located in one place, or may also be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.

One of ordinary skill in the art will appreciate that all or some of the steps of the methods, systems, functional modules/units in the devices disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof.

The terms "first," "second," "third," "fourth," and the like in the description of the application and the above-described figures, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

It should be understood that in the present application, "at least one" means one or more, "a plurality" means two or more. "and/or" for describing an association relationship of associated objects, indicating that there may be three relationships, e.g., "a and/or B" may indicate: only A, only B and both A and B are present, wherein A and B may be singular or plural. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. "at least one of the following" or similar expressions refer to any combination of these items, including any combination of single item(s) or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", wherein a, b, c may be single or plural.

In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the above-described division of units is only one type of division of logical functions, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes multiple instructions for causing a computer device (which may be a personal computer, a server, or a network device) to perform all or part of the steps of the method of the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing programs, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

The preferred embodiments of the present application have been described above with reference to the accompanying drawings, and the scope of the claims of the embodiments of the present application is not limited thereby. Any modifications, equivalents and improvements that may occur to those skilled in the art without departing from the scope and spirit of the embodiments of the present application are intended to be within the scope of the claims of the embodiments of the present application.

Claims

1. A personalized information recommendation method is characterized by comprising the following steps:

responding to a recommendation request of a target user, and acquiring user data of the target user and information data of information to be processed in an information pool, wherein the information data comprises information content data and state data used for representing a mark state, and the user data comprises user historical behavior data and user current behavior data;

according to a preset personalized multi-way recall strategy, the user data and the information data, recalling the information to be processed to obtain a first recall result containing candidate information, and inputting the first recall result into a preset sequencing model to obtain a first recommendation sequence, wherein the personalized multi-way recall strategy comprises the following steps: a collaborative filtering recall strategy, an alternating least squares recall strategy, and a topic recall strategy, the first recall result comprising: collaborative filtering recall results, alternate least squares recall results, and topic recall results;

recalling the information to be processed according to a preset popularization strategy, the user data and the state data to obtain a second recall result containing the candidate information, wherein the popularization strategy comprises: a new information recall policy, a trending information recall policy, and a tagged recall policy, the second recall result comprising: a new information recall result, a hit information recall result, and a tagged recall result;

inputting the first recommendation sequence and the second recall result into a preset mixing model, determining target candidate information, and pushing the target candidate information to the target user;

the recalling the information to be processed according to a preset personalized multipath recalling strategy, the user data and the information data to obtain a first recalling result containing candidate information includes:

determining an information content portrait according to a preset word frequency inverse document frequency algorithm and the information content data;

determining the information similarity between any two pieces of information to be processed in the information pool according to the information content image and a preset similarity algorithm;

recalling the information to be processed according to the collaborative filtering recall strategy, the user historical behavior data and the information similarity to obtain a collaborative filtering recall result;

recalling the information to be processed according to the alternative least square recall strategy and a preset information scoring matrix to obtain an alternative least square recall result;

according to the theme recall strategy, the user data and the information data, the information to be processed is recalled to obtain a theme recall result;

wherein, the recalling the information to be processed according to a preset popularization strategy, the user data and the state data to obtain a second recall result containing the candidate information comprises:

according to the new information recall strategy and the user historical behavior data, using information to be processed associated with the user historical behavior data as historical information to be processed, and using the information to be processed except the historical information to be processed as a new information recall result;

determining the current grading information of the current user behavior data according to grading rules and the current user behavior data;

sorting the information to be processed according to the trending information recall strategy and the current grading information, and determining a trending information recall result according to a sorting result;

and according to the mark recall strategy and the state data, taking the information to be processed in the target mark state as the mark recall result.

2. The method of claim 1, wherein the inputting the first recommended sequence and the second recall result into a preset blending model, determining target candidate information, and pushing the target candidate information to the target user comprises:

receiving a first weight and a second weight, wherein the sum of the first weight and the second weight is less than or equal to one;

updating the model parameters of a preset mixed model according to the first weight and the second weight;

inputting the first recommendation sequence and the second recall result into an updated hybrid model to obtain a target recommendation sequence, wherein the ratio of candidate information of the first recommendation sequence in the target recommendation sequence is the first weight, and the ratio of candidate information of the second recall result in the target recommendation sequence is the second weight;

determining target candidate information according to the target recommendation sequence;

and pushing the target candidate information to the target user.

3. The method of claim 1, wherein the step of obtaining the user data of the target user and the information data of the information to be processed in the information pool in response to the recommendation request of the target user is preceded by the steps of:

receiving a mark updating instruction of the information to be processed;

and responding to the mark updating instruction, and updating the mark state of the corresponding information to be processed.

4. The method of claim 1, wherein the historical behavior data of the user is implicit feedback of the user on the associated information to be processed; before the step of recalling the information to be processed according to the preset personalized multi-way recall strategy, the user data and the information data to obtain a first recall result containing candidate information, and inputting the first recall result into a preset ranking model to obtain a first recommendation sequence, the method comprises the following steps:

determining historical recording time of the user historical behavior data, and determining a time attenuation coefficient according to the historical recording time;

determining a user behavior portrait according to the time attenuation coefficient and historical scoring information, wherein the user behavior portrait is used for determining the first recall result;

，

wherein the content of the first and second substances,

is the time decay factor, is>

5. The method of claim 1, wherein the user data further comprises user preference data, the information to be processed further comprises information subject data; the recalling the to-be-processed information according to the theme recalling strategy, the user data and the information data to obtain the theme recalling result comprises the following steps:

determining a matching relation between the user data and the information to be processed according to the user preference data and the information subject data, wherein the information subject data is used for representing an information subject of the information to be processed, and the user preference data is used for representing a target information subject of the target user;

according to the theme recall strategy and the matching relation, the information to be processed matched with the user data is used as theme matching information;

and randomly screening the theme matching information according to the preset theme recommendation quantity, and taking a screening result as a theme recall result.

6. A personalized information recommendation apparatus, comprising:

the system comprises an acquisition unit, a recommendation unit and a recommendation unit, wherein the acquisition unit is used for responding to a recommendation request of a target user and acquiring user data of the target user and information data of information to be processed in an information pool, the information data comprises information content data and state data used for representing a mark state, and the user data comprises user historical behavior data and user current behavior data;

the first recall unit is configured to perform recall processing on the information to be processed according to a preset personalized multi-recall policy, the user data, and the information data to obtain a first recall result including candidate information, and input the first recall result into a preset ranking model to obtain a first recommended sequence, where the personalized multi-recall policy includes: a collaborative filtering recall strategy, an alternating least squares recall strategy, and a topic recall strategy, the first recall result comprising: collaborative filtering recall results, alternate least squares recall results, and topic recall results;

a second recall unit, configured to perform recall processing on the information to be processed according to a preset popularization policy, the user data, and the state data, so as to obtain a second recall result including the candidate information, where the popularization policy includes: a new information recall policy, a trending information recall policy, and a tagged recall policy, the second recall result comprising: a new information recall result, a hit information recall result, and a tagged recall result;

the recommending unit is used for inputting the first recommending sequence and the second recalling result into a preset mixing model, determining target candidate information and pushing the target candidate information to the target user;

the recalling the information to be processed according to a preset personalized multipath recalling strategy, the user data and the information data to obtain a first recalling result containing candidate information comprises the following steps:

7. An electronic device, comprising a memory storing a computer program and a processor, wherein the processor implements the steps of the personalized information recommendation method according to any one of claims 1 to 5 when executing the computer program.

8. A storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the personalized information recommendation method according to any one of claims 1 to 5.