CN104123284B - The method and server of a kind of recommendation - Google Patents

Abstract

The embodiment of the present invention discloses a recommendation method and server. The method combines the recommendation lists sent by each recommendation system through a preset combination strategy to realize multiple recommendation models or the system feeds back recommendation results to users. Feedback selection probability, update the combined recommendation list, realize real-time evaluation, and receive the user’s regular feedback of selection results, update the combined recommendation list, and then reflect the user’s current interests and historical hobbies at the same time.

Description

A recommended method and server

technical field

The invention belongs to the field of data processing, in particular to a recommended method and server.

Background technique

In today's information explosion, more and more commercial systems introduce recommendation technology, changing from the previous mode of people looking for content to content looking for people to meet the personalized needs of users. The effect of a single recommendation system recommendation result is limited, which is especially reflected in more and more recommendation competitions. The final winners of the competition often use multiple recommendation technologies/models or scoring results for fusion and integration. In the past, the recommendation was more analogous to the problem of predictive scoring, but some scholars believe that the form of recommendation list may be more suitable. The effect of the previous recommendation system is not easy to evaluate, and it is usually done offline, and the real-time effect evaluation has not been used. Historical preferences represent the user's hobbies and hobbies, which generally will not change for a long time and can be obtained by analyzing the user's historical behavior; while current preferences represent the user's current temporary interests and hobbies, which are generally not changed at any time. Changeable over time and the external environment.

In the usual prior art 1, data is generally imported from the business system to the recommendation system offline, and user feedback is the real-time data feedback from users to the recommendation system, which is used to update the recommendation model to improve the accuracy of future predictions. Commonly used user feedback methods include favorites, clicks, browsing (time), purchases, ratings, comments, etc. The disadvantage of prior art 1 is that the technology is relatively single, and it is difficult to give consideration to the accuracy of recommendation and the real-time performance of calculation.

In the usual prior art 2, the recommendation combination technology obtains a selection model through offline training, which is essentially a single recommendation system. The recommendation combination and the back-end recommendation technology are generally strongly related and deployed together. Recommended combination techniques: neural networks, case-based reasoning (CBR), decision trees, etc. The disadvantage of existing technology 2 is that the recommendation technology is not easy to expand, and the combination model needs to be re-trained offline every time a recommendation technology is added, and it is impossible to feed back the user's current interest in real time.

Contents of the invention

The purpose of the present invention is to provide a recommendation method and server to solve how to use real-time evaluation effects to update the recommendation list when there are multiple recommendation models or systems, while reflecting the user's current interests and historical hobbies.

In a first aspect, a recommended method, said method comprising:

Receive recommendation lists sent by each recommendation system;

Combining the recommendation list according to a preset combination strategy, presenting the combined recommendation list to the user, so that the user makes a selection according to the combined recommendation list;

A selection result fed back by the user is received, and the combined recommendation list is updated according to the selection result.

With reference to the first aspect, in a first possible implementation manner of the first aspect, combining the recommendation lists according to a preset combination strategy includes:

Predefine the proportion of the results sent by each recommendation system in the results sent by all recommendation systems;

According to the proportion, the recommendation lists sent by each recommendation system are combined.

With reference to the first aspect or the first possible implementation of the first aspect, in the second possible implementation of the first aspect, after receiving the selection result fed back by the user, updating the combination according to the selection result recommended list, including:

receiving a selection result fed back by the user, and calculating a selection probability according to the selection result;

Set the update coefficient in advance, and calculate the updated proportion of each recommendation system according to the selection probability, the update coefficient and the proportion of each recommendation system;

The combined recommendation list is updated according to the updated proportions.

With reference to the second possible implementation manner of the first aspect, in a third possible implementation manner of the first aspect, the receiving the selection result fed back by the user, and calculating the selection probability according to the selection result include:

calculating the ratio of the selection result to all the results of the combined recommendation list, where the ratio is the selection probability; or,

The weight of the user's choice is preset, and according to the weight of the user's choice, the ratio of the weight of the user's selection of the results recommended by each recommendation system to the weight of all the results of the combined recommendation list is obtained, and the ratio is the selection probability.

In combination with the first aspect or the first possible implementation of the first aspect or the second possible implementation of the first aspect or the third possible implementation of the first aspect, the fourth possible implementation of the first aspect In the implementation mode, the method also includes:

The selection result fed back periodically by the user is received, and the recommendation list of each recommendation system is updated according to the selection result fed back periodically.

In a second aspect, a server includes:

a receiving unit, configured to receive the recommendation list sent by each recommendation system;

A combination unit, configured to combine the recommendation lists according to a preset combination strategy, and present the combined recommendation list to the user, so that the user can make a selection according to the combined recommendation list;

An updating unit, configured to receive a selection result fed back by the user, and update the combined recommendation list according to the selection result.

With reference to the second aspect, in a first possible implementation manner of the second aspect, the combination unit is specifically used for:

Predefine the proportion of the results sent by each recommendation system in the results sent by all recommendation systems;

According to the proportion, the recommendation lists sent by each recommendation system are combined.

With reference to the second aspect or the first possible implementation manner of the second aspect, in the second possible implementation manner of the second aspect, the update unit is specifically configured to:

receiving a selection result fed back by the user, and calculating a selection probability according to the selection result;

Set the update coefficient in advance, and calculate the updated proportion of each recommendation system according to the selection probability, the update coefficient and the proportion of each recommendation system;

The combined recommendation list is updated according to the updated proportions.

With reference to the second possible implementation of the second aspect, in a third possible implementation of the second aspect, the updating unit executes a step of receiving a selection result fed back by the user, and calculating the selection probability according to the selection result, include:

calculating the ratio of the selection result to all the results of the combined recommendation list, where the ratio is the selection probability; or,

The weight of the user's choice is preset, and according to the weight of the user's choice, the ratio of the weight of the user's selection of the results recommended by each recommendation system to the weight of all the results of the combined recommendation list is obtained, and the ratio is the selection probability.

In combination with the second aspect or the first possible implementation of the second aspect or the second possible implementation of the second aspect or the third possible implementation of the second aspect, the fourth possible implementation of the second aspect In an implementation manner, the server further includes a periodic feedback unit, configured to:

The selection result fed back periodically by the user is received, and the recommendation list of each recommendation system is updated according to the selection result fed back periodically.

Compared with the prior art, the present invention combines the recommendation lists sent by each recommendation system through a preset combination strategy, realizes multiple recommendation models or the system feeds back the recommendation results to the user, and updates the selection probability according to the user feedback. The combined recommendation list realizes real-time evaluation, and receives the user's regular feedback of selection results, and updates the combined recommendation list, because real-time evaluation can reflect the user's current interest, and regular feedback can reflect the user's historical interest. Therefore, the present invention It can reflect the user's current interests and historical interests at the same time.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings that need to be used in the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For Those of ordinary skill in the art can also obtain other drawings based on these drawings without any creative effort.

FIG. 1 is an application scenario diagram of a recommended method provided by an embodiment of the present invention;

Fig. 2 is a method flowchart of a recommended method provided by an embodiment of the present invention;

Fig. 3 is a method schematic diagram of a recommended method provided by an embodiment of the present invention;

Fig. 4 is a method schematic diagram of a recommended method provided by an embodiment of the present invention;

FIG. 5 is a device structural diagram of a server provided by an embodiment of the present invention;

Fig. 6 is a device structural diagram of a server provided by an embodiment of the present invention.

detailed description

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. within range.

Referring to FIG. 1 , FIG. 1 is an application scenario diagram of a recommended method provided by an embodiment of the present invention. As shown in Figure 1, the user 101 selects an item of interest from the recommendation list according to the recommendation list provided by the server 102. To update the system, the recommendation list pushed by the server 102 to the user 101 next time can reflect the preferences of the user 101 last time, and the server 102 will regularly receive the selection results fed back by the user 101, and the server 102 will update the system according to the selection results fed back regularly. The recommendation list pushed by the server 102 to the user 101 can reflect the user's historical interest and current interest at the same time.

Referring to FIG. 2 , FIG. 2 is a method flowchart of a recommended method provided by an embodiment of the present invention. As shown in Figure 2, the method includes the following steps:

Step 201, receiving recommendation lists sent by each recommendation system;

Specifically, as shown in FIG. 3 , FIG. 3 is a method schematic diagram of a recommended method provided by an embodiment of the present invention. As shown in Figure 3, the recommendation front-end system receives the recommendation lists sent by the recommendation system 1, the recommendation system 2, and the recommendation system 3, and the recommendation front-end system combines the recommendation lists sent by the recommendation system 1, the recommendation system 2, and the recommendation system 3 according to the combination strategy. The recommendation list is combined, and the recommendation front-end system sends the combined recommendation list to the service system, so that the service system presents the combined recommendation list to the user. The recommendation front-end system receives the selection probability fed back by the user in real time, and is used to update the combined recommendation list. At the same time, the recommendation system 1, the recommendation system 2, and the recommendation system 3 receive the selection results regularly fed back by the user, and are used to update the recommendation system 1. The recommendation list of recommendation system 2 and recommendation system 3. Each recommendation system has different recommendation results calculated according to the database. For example, recommendation system 1 recommends more children's products, and recommendation system 2 recommends more electronic products or books or clothes and other fields.

Step 202, combining the recommendation lists according to a preset combination strategy, and presenting the combined recommendation list to the user, so that the user can make a selection according to the combined recommendation list;

Optionally, combining the recommendation lists according to a preset combination strategy includes:

Predefine the proportion of the results sent by each recommendation system in the results sent by all recommendation systems;

According to the proportion, the recommendation lists sent by each recommendation system are combined.

Specifically, suppose that the recommendation list A sent by the recommendation system 1 to the recommendation front-end system is {a1, a2, a3}, the recommendation list B sent by the recommendation system 2 to the recommendation front-end system is {b1, b2, b3}, and the recommendation system 3 sends The recommendation list C for the recommendation front-end system is {c1, c2, c3}. According to the combination strategy of the preset recommendation results, assuming that the recommendation results of recommendation system 1, recommendation system 2, and recommendation system 3 account for 1/3 of the recommendation results of all recommendation systems, the combined recommendation list can be {a1 , b1, c2}, assuming that the recommendation results of recommendation system 1, recommendation system 2, and recommendation system 3 account for 3/5, 1/5, and 1/5 of the recommendation results of all recommendation systems, respectively, the combined recommendation list Can be {a1, a2, a3, b2, c3}.

At the same time, the proportion of the combination strategy can be freely defined. Assuming that it is Children's Day, the proportion of the recommendation system 1 can be increased, because the recommendation system 1 mostly recommends children's products.

Step 203, receiving a selection result fed back by the user, and updating the combined recommendation list according to the selection result.

Optionally, the receiving a selection result fed back by the user, and updating the combined recommendation list according to the selection result include:

receiving a selection result fed back by the user, and calculating a selection probability according to the selection result;

Set the update coefficient in advance, and calculate the updated proportion of each recommendation system according to the selection probability, the update coefficient and the proportion of each recommendation system;

The combined recommendation list is updated according to the updated proportions.

Optionally, the receiving the selection result fed back by the user and calculating the selection probability according to the selection result include:

calculating the ratio of the selection result to all the results of the combined recommendation list, where the ratio is the selection probability; or,

The weight of the user's choice is preset, and according to the weight of the user's choice, the ratio of the weight of the user's selection of the results recommended by each recommendation system to the weight of all the results of the combined recommendation list is obtained, and the ratio is the selection probability.

Specifically, initialize the three recommendation systems in Figure 3, assuming that the proportions of each recommendation system are P ₁ (t), P ₂ (t), and P ₃ (t),

and

p _i (t) represents the proportion of the i-th recommendation system at time t

p _i (t+1) represents the proportion of the i-th recommendation system at the next moment after time t

η is the update factor

λ _i (t) represents the probability that the result recommended by the i-th recommendation system is selected by the user at time t

Suppose the recommendation list A sent by recommendation system 1 is {i1, i2, i3}, the recommendation list B sent by recommendation system 2 is {i2, i3, i4, i5}, and recommendation system 3 does not send the recommendation list, and finally the combination is presented to the user The recommendation list list is {i1, i2, i3, i4, i5}. In order to avoid the impact of the position of the combination of recommendation results on the user's choice, we randomly arrange the order after combining the results. Suppose the user chooses {i2,i4}, in fact, the user chooses {i2(A),i2(B),i4(B)}, so

Specifically, when the user’s operation behavior is different, for example, the user’s behavior for i2 is to purchase, while the behavior for i4 is just browsing, then the selection probability of each recommendation subsystem’s recommendation result being selected by the user needs to consider the operation behavior of each user Weight, the weight of purchasing behavior is obviously greater than that of browsing behavior. Assuming that the weight of purchasing behavior is 0.3 and the weight of browsing behavior is 0.2, the probability of user selection is {i2(A)*0.3, i2(B)*0.3, i4(B)*0.2 },so

For λ _A (t), 3/8>1/3, the recommendation effect of recommendation system 1 in the second case is better than that in the first case, because the i2 recommended by recommendation system 1 is purchased by the user.

As another optional method, the method also includes:

The selection result periodically fed back by the user is received, and the recommendation list of each recommendation system is updated according to the selection result fed back periodically.

Specifically, the recommendation system 1, the recommendation system 2, and the recommendation system 3 regularly receive the selection results regularly fed back by users, and update the recommendation list sent by themselves according to the results.

The present invention combines the recommendation lists sent by each recommendation system through a preset combination strategy to realize multiple recommendation models or systems to feed back recommendation results to users, and update the combined recommendation lists according to the selection probability fed back by users to realize Evaluate in real time, and receive regular feedback from users on the selection results, update the combined recommendation list, and then reflect the user's current interests and historical interests at the same time.

Referring to FIG. 4 , FIG. 4 is a method schematic diagram of a recommended method provided by an embodiment of the present invention. The method comprises the steps of:

Step 401, the user logs in for the first time, and the system does not have any user information, then the recommendation front-end system selects an equal amount of recommendation results from the recommendation list of each recommendation subsystem to present to the user, or the system assigns initial values to each recommendation subsystem according to the combination strategy ;

Step 402, the recommendation front-end system obtains the recommendation list of each recommendation system for combined display;

Step 403, the user performs operations such as selecting to browse or purchase or bookmark the recommended results;

Step 404, the recommendation front-end system captures the user's behavior in real time, assigns different weights to different user behaviors to represent the user's current preference, and calculates the selection probability of the recommendation result;

Step 405, the recommendation front-end system adjusts the proportion of the output of each recommendation system in the recommended final result list according to the user's current preferences according to certain rules;

Step 406, the user can now see a recommendation result different from the previous one;

Step 407, the user logs out to end the session, and each recommendation subsystem obtains the overall behavior of the user during the session;

Step 408, each recommendation subsystem adjusts the recommendation list according to the user's historical preferences;

In step 409, the user logs in to the system again, and jumps to step 401.

Referring to FIG. 5 , FIG. 5 is a device structural diagram of a server provided by an embodiment of the present invention. As shown in Figure 4, the server includes the following units:

a receiving unit 501, configured to receive the recommendation list sent by each recommendation system;

Specifically, as shown in FIG. 3 , FIG. 3 is a method schematic diagram of a recommended method provided by an embodiment of the present invention. As shown in Figure 3, the recommendation front-end system receives the recommendation lists sent by the recommendation system 1, the recommendation system 2, and the recommendation system 3, and the recommendation front-end system combines the recommendation lists sent by the recommendation system 1, the recommendation system 2, and the recommendation system 3 according to the combination strategy. The recommendation list is combined, and the recommendation front-end system sends the combined recommendation list to the service system, so that the service system presents the combined recommendation list to the user. The recommendation front-end system receives the selection probability fed back by the user in real time, and is used to update the combined recommendation list. At the same time, the recommendation system 1, the recommendation system 2, and the recommendation system 3 receive the selection results regularly fed back by the user, and are used to update the recommendation system 1. Recommender system 2. Recommendation list of recommender system 3.

The combination unit 502 is configured to combine the recommendation list according to a preset combination strategy, and present the combined recommendation list to the user, so that the user can make a selection according to the combined recommendation list;

Optionally, the combining unit 502 is specifically configured to:

Predefine the proportion of the results sent by each recommendation system in the results sent by all recommendation systems;

According to the proportion, the recommendation lists sent by each recommendation system are combined.

Specifically, suppose that the recommendation list A sent by the recommendation system 1 to the recommendation front-end system is {a1, a2, a3}, the recommendation list B sent by the recommendation system 2 to the recommendation front-end system is {b1, b2, b3}, and the recommendation system 3 sends The recommendation list C for the recommendation front-end system is {c1, c2, c3}. According to the pre-combined recommendation strategy, assuming that the recommendation results of recommendation system 1, recommendation system 2, and recommendation system 3 account for 1/3 of the recommendation results of all recommendation systems, the combined recommendation list can be {a1, b1, c2}, assuming that the recommended results of recommendation system 1, recommendation system 2, and recommendation system 3 account for 3/5, 1/5, and 1/5 of the recommended results of all recommendation systems, respectively, the combined recommendation list can be { a1, a2, a3, b2, c3}.

At the same time, the proportion of the combination strategy can be freely defined. Assuming that it is Children's Day, the proportion of the recommendation system 1 can be increased, because the recommendation system 1 mostly recommends children's products.

The updating unit 503 is configured to receive a selection result fed back by the user, and update the combined recommendation list according to the selection result.

Optionally, the update unit 503 is specifically configured to:

receiving a selection result fed back by the user, and calculating a selection probability according to the selection result;

Set the update coefficient in advance, and calculate the updated proportion of each recommendation system according to the selection probability, the update coefficient and the proportion of each recommendation system;

The combined recommendation list is updated according to the updated proportions.

The execution step in the update unit receives the selection result fed back by the user, and calculates the selection probability according to the selection result, including:

calculating the ratio of the selection result to all the results of the combined recommendation list, where the ratio is the selection probability; or,

The weight of the user's choice is preset, and according to the weight of the user's choice, the ratio of the weight of the user's selection of the results recommended by each recommendation system to the weight of all the results of the combined recommendation list is obtained, and the ratio is the selection probability.

Specifically, initialize the three recommendation systems in Figure 3, assuming that the proportions of each recommendation system are P ₁ (t), P ₂ (t), and P ₃ (t),

and

p _i (t) represents the proportion of the i-th recommendation system at time t

p _i (t+1) represents the proportion of the i-th recommendation system at the next moment after time t

η is the update factor

λ _i (t) represents the probability that the result recommended by the i-th recommendation system is selected by the user at time t

Suppose the recommendation list A sent by recommendation system 1 is {i1, i2, i3}, the recommendation list B sent by recommendation system 2 is {i2, i3, i4, i5}, and recommendation system 3 does not send the recommendation list, and finally the combination is presented to the user The recommendation list list is {i1, i2, i3, i4, i5}. In order to avoid the impact of the position of the combination of recommendation results on the user's choice, we randomly arrange the order after combining the results. Suppose the user chooses {i2,i4}, in fact, the user chooses {i2(A),i2(B),i4(B)}, so

Specifically, when the user's operation behavior is different, for example: the user's behavior for i2 is purchase, while the behavior for i4 is only browsing. , then the selection probability of each recommendation subsystem’s recommendation results being selected by the user needs to consider the weight of each user’s operation behavior. Obviously, the weight of purchase behavior is greater than that of browsing behavior. Assuming that the weight of purchase behavior is 0.3 and the weight of browsing behavior is 0.2, the user’s selection probability is {i2(A)*0.3,i2(B)*0.3,i4(B)*0.2}, so

For λ _A (t), 3/8>1/3, the recommendation effect of recommendation system 1 in the second case is better than that in the first case, because the i2 recommended by recommendation system 1 is purchased by the user.

As an optional embodiment, the server further includes a periodic feedback unit 504, configured to:

The selection result periodically fed back by the user is received, and the recommendation list of each recommendation system is updated according to the selection result fed back periodically.

Specifically, the recommendation system 1, the recommendation system 2, and the recommendation system 3 regularly receive the selection results regularly fed back by users, and update the recommendation list sent by themselves according to the results.

The present invention combines the recommendation lists sent by each recommendation system through a preset combination strategy to realize multiple recommendation models or systems to feed back recommendation results to users, and update the combined recommendation lists according to the selection probability fed back by users to realize Evaluate in real time, and receive regular feedback from users on the selection results, update the combined recommendation list, and then reflect the user's current interests and historical interests at the same time.

Referring to FIG. 6 , FIG. 6 is a device structural diagram of a server provided by an embodiment of the present invention. Referring to FIG. 6 , FIG. 6 is a server 600 provided by an embodiment of the present invention, and the specific embodiment of the present invention does not limit the specific implementation of the server. The server 600 includes:

A processor (processor) 601 , a communication interface (Communications Interface) 602 , a memory (memory) 603 , and a bus 604 .

The processor 601 , the communication interface 602 , and the memory 603 communicate with each other through the bus 604 .

A communication interface 602, configured to communicate with other devices;

Processor 601, configured to execute programs.

Specifically, the program may include program code, and the program code includes computer operation instructions.

The processor 601 may be a central processing unit CPU, or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention.

The memory 603 is used to store the program A. The memory 603 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory. Specific procedures may include:

Receive recommendation lists sent by each recommendation system;

Combining the recommendation list according to a preset combination strategy, presenting the combined recommendation list to the user, so that the user makes a selection according to the combined recommendation list;

A selection result fed back by the user is received, and the combined recommendation list is updated according to the selection result.

The combination of the recommendation list according to a preset combination strategy includes:

Predefine the proportion of the results sent by each recommendation system in the results sent by all recommendation systems;

According to the proportion, the recommendation lists sent by each recommendation system are combined.

The receiving the selection result fed back by the user, and updating the combined recommendation list according to the selection result include:

receiving a selection result fed back by the user, and calculating a selection probability according to the selection result;

Set the update coefficient in advance, and calculate the updated proportion of each recommendation system according to the selection probability, the update coefficient and the proportion of each recommendation system;

The combined recommendation list is updated according to the updated proportions.

The receiving the selection result fed back by the user and calculating the selection probability according to the selection result include:

calculating the ratio of the selection result to all the results of the combined recommendation list, where the ratio is the selection probability; or,

The weight of the user's choice is preset, and according to the weight of the user's choice, the ratio of the weight of the user's selection of the results recommended by each recommendation system to the weight of all the results of the combined recommendation list is obtained, and the ratio is the selection probability.

The method also includes:

The selection result periodically fed back by the user is received, and the recommendation list of each recommendation system is updated according to the selection result fed back periodically.

The above descriptions are only preferred embodiments of the present invention, and do not constitute a limitation to the protection scope of the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included within the scope of the claims of the present invention.

Claims (8)

1. A recommended method, characterized in that the method comprises: The server receives the recommendation list sent by each recommendation system; Combine the recommendation lists according to a preset combination strategy, and present the combined recommendation list to the user, so that the user can select an item of interest according to the combined recommendation list, and the item includes children's products or books or clothes; receiving a selection result fed back by the user, and updating the combined recommendation list according to the selection result; The receiving the selection result fed back by the user, and updating the combined recommendation list according to the selection result include: receiving a selection result fed back by the user, and calculating a selection probability according to the selection result; Set the update coefficient in advance, and calculate the updated proportion of each recommendation system according to the selection probability, the update coefficient and the proportion of each recommendation system; updating the combined recommendation list according to the updated proportion; The formula for calculating the proportion of each recommendation system after updating is: and p _i (t) represents the proportion of the i-th recommendation system at time t; p _i (t+1) represents the proportion of the i-th recommendation system at the next moment after time t; η is the update coefficient; λ _i (t) represents the probability that the result recommended by the i-th recommendation system is selected by the user at time t. 2. The method according to claim 1, wherein said combining the recommendation list according to a preset combination strategy comprises: Predefine the proportion of the results sent by each recommendation system in the results sent by all recommendation systems; According to the proportion, the recommendation lists sent by each recommendation system are combined. 3. The method according to claim 1, wherein the receiving the selection result fed back by the user and calculating the selection probability according to the selection result include: calculating the ratio of the selection result to all the results of the combined recommendation list, where the ratio is the selection probability; or, The weight of the user's choice is preset, and according to the weight of the user's choice, the ratio of the weight of the user's selection of the results recommended by each recommendation system to the weight of all the results of the combined recommendation list is obtained, and the ratio is the selection probability. 4. The method according to any one of claims 1-3, wherein the method further comprises: The selection result periodically fed back by the user is received, and the recommendation list of each recommendation system is updated according to the selection result fed back periodically. 5. A server, characterized in that the server comprises: a receiving unit, configured to receive the recommendation list sent by each recommendation system; The combination unit is configured to combine the recommendation lists according to a preset combination strategy, and present the combined recommendation list to the user, so that the user can select an item of interest according to the combined recommendation list, and the item including children's products or books or clothing; An update unit, configured to receive a selection result fed back by the user, and update the combined recommendation list according to the selection result; The update unit is specifically used for: receiving a selection result fed back by the user, and calculating a selection probability according to the selection result; Set the update coefficient in advance, and calculate the updated proportion of each recommendation system according to the selection probability, the update coefficient and the proportion of each recommendation system; updating the combined recommendation list according to the updated proportion; The formula for calculating the proportion of each recommendation system after updating is: and p _i (t) represents the proportion of the i-th recommendation system at time t; p _i (t+1) represents the proportion of the i-th recommendation system at the next moment after time t; η is the update coefficient; λ _i (t) represents the probability that the result recommended by the i-th recommendation system is selected by the user at time t. 6. The server according to claim 5, wherein the combining unit is specifically used for: Predefine the proportion of the results sent by each recommendation system in the results sent by all recommendation systems; According to the proportion, the recommendation lists sent by each recommendation system are combined. 7. The server according to claim 5, characterized in that, the execution step in the update unit receives the selection result fed back by the user, and calculates the selection probability according to the selection result, comprising: calculating the ratio of the selection result to all the results of the combined recommendation list, where the ratio is the selection probability; or, The weight of the user's choice is preset, and according to the weight of the user's choice, the ratio of the weight of the user's selection of the results recommended by each recommendation system to the weight of all the results of the combined recommendation list is obtained, and the ratio is the selection probability. 8. The server according to any one of claims 5-7, wherein the server further comprises a periodic feedback unit, configured to: The selection result periodically fed back by the user is received, and the recommendation list of each recommendation system is updated according to the selection result fed back periodically.