Information recommendation method, apparatus, and non-transitory computer readable medium

1 . A computer-implemented method of allocating resources in a graphical user interface (GUI) of an e-commerce application on a display device to reduce user interface switching events, the method comprising:

receiving, from the display device, a first preset operation of the user, wherein the first preset operation comprises selecting, via a click operation, an interactive interface element;

acquiring, in response to determining that a user enters a preset interface from a homepage of a shopping application based on receiving the first preset operation of the user, operation information within a preset length of time before the user performs the first preset operation,

wherein the preset interface comprises a goods details interface configured to display a main object and further comprises a preset display area having a plurality of preset display regions, each of the preset display regions displaying an object other than the main object, and

wherein the homepage displays a plurality of interactive interface elements corresponding to objects of different categories from the main object;

determining behavior data of the user based on the operation information, comprising:

receiving, via a click operation, a second preset operation comprising selection of an object category or a selectable object displayed in the preset interface;

inputting, in response to determining that the operation information comprises the second preset operation, the operation information into a pre-trained behavior data recognition model comprising a neural network, wherein the pre-trained behavior data recognition model is configured to: receive the operation information, classify a user behavior in the operation information according to learning from training, and output a first array of only two elements;

outputting the first array of only two elements, wherein one of the two elements is a first proportion which represents a proportion of a first behavior and the other of the two elements is a second proportion which represents a proportion of a second behavior,

wherein the first proportion and the second proportion sum to unity, and wherein in the first array, the first proportion and the second proportion are not preset; and

determining, in response to determining that the operation information does not comprise the second preset operation, a portal to enter the preset interface and generating a second array of two elements in response to determining that the portal is a portal in a search results interface obtained based on a third preset operation of the user, wherein the third preset operation of the user comprises inputting a search operation into the homepage;

wherein the two elements comprise a third proportion which represents the proportion of the first behavior and a fourth proportion which represents the proportion of the second behavior, wherein the third proportion and the fourth proportion sum to unity, wherein, in the second array, the third proportion and the fourth proportion are preset;

wherein the first behavior represents that the user has a definite browsing object and the second behavior represents that the user has no definite browsing object; and

determining recommendation information for the user based on the behavior data, and displaying the recommendation information in the preset interface, wherein objects represented by the recommendation information comprise objects of a same category as the main object displayed in the preset interface, and one or more objects of different categories from the main object, wherein the one or more objects of the different categories from the main object are associated with one or more of the plurality of interactive interface elements,

wherein the determining the recommendation information for the user based on the behavior data comprises:

allocating, based on a total number of the preset display regions in the preset display area in the preset interface and in proportion to one of the first proportion or the third proportion of the first behavior, a corresponding first number of the preset display regions in the preset display area that display a number of the objects of the same category as the main object, and

allocating, based on the total number of preset display regions in the preset display area in the preset interface and in proportion to one of the second proportion or the fourth proportion of the second behavior, a corresponding second number of the preset display regions in the preset display area that display a number of the objects of the different categories from the main object; and

outputting instructions to the display device to cause the display device to render the number of the objects of the same category as the main object in the allocated first number of the preset display regions, and the number of the objects of the different categories from the main object in the allocated second number of the preset display regions, such that the objects of the same category as the main object and the objects of the different categories from the main object are displayed at the same time in the allocated first number of the preset display regions and the allocated second number of the preset display regions.

2 . The computer-implemented method according to claim 1 , wherein the determining behavior data of the user browsing the preset interface based on the operation information, further comprises:

determining, in response to determining that the portal is a portal in a preset recommendation interface, a third preset array representing a fifth proportion of the first behavior and a sixth proportion of the second behavior.

3 . The computer-implemented method according to claim 1 , wherein the pre-trained behavior data recognition model is obtained by training through a following method:

acquiring a training sample set, wherein training samples in the training sample set comprise: sample operation information with a unit of the preset length of time and sample labels; and

training, using a machine learning algorithm, using the sample operation information in the training samples as input of an initial behavior data recognition model, and using the sample labels corresponding to the input sample operation information as expected output of the initial behavior data recognition model, to obtain the pre-trained behavior data recognition model.

4 . The computer-implemented method according to claim 3 , wherein the acquiring a training sample set, comprises:

acquiring log information;

dividing the log information according to a unit of the preset length of time to obtain a plurality of pieces of sample operation information; and

determining, for each piece of sample operation information in the plurality of pieces of sample operation information, based on a number of object categories comprised in the sample operation information, and a number of page views for each category in the sample operation information, a sample label for the sample operation information.

5 . An apparatus configured for allocating resources in a graphical user interface (GUI) of an e-commerce application on a display device to reduce user interface switching events, the apparatus comprising:

at least one processor; and

a memory storing:

a pre-trained behavior data recognition model based on a neural network which was trained using data which represents a relationship between operation information and categories of user behavior, and is configured to output a data structure comprising a first array, wherein each element of the first array represents a corresponding output value of the pre-trained behavior data recognition model for a respective one of a plurality of predefined behavior categories for which the model was trained, in response to being input with operation information which represents user behavior, and

instructions that, when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:

receiving, from the display device, a first preset operation of the user, wherein the first preset operation comprises selecting, via a click operation, an interactive interface element;

acquiring, in response to determining that a user enters a preset interface from a homepage of a shopping application based on receiving the first preset operation of the user, operation information within a preset length of time before the user performs the first preset operation,

wherein the preset interface comprises a goods details interface configured to display a main object and further comprises a preset display area having a plurality of preset display regions, each of the preset display regions displaying an object other than the main object, and

wherein the homepage displays a plurality of interactive interface elements corresponding to objects of different categories from the main object;

determining behavior data of the user based on the operation information comprising:

parsing data in the operation information to determine whether the operation information comprises a second preset operation or does not comprise the second preset operation, and outputting a corresponding determination result,

in response to receiving, via a click operation, a second preset operation comprising selection of an object category or a selectable object displayed in the preset interface and determining that the operation information does comprise the second preset operation, inputting the operation information into the pre-trained behavior data recognition model and classifying the operation information into a first behavior and a second behavior and outputting the data structure comprising the first array of only two elements, wherein one of the two elements is a first proportion which represents a proportion of a first behavior and the other of the two elements is a second proportion which represents a proportion of a second behavior wherein the first proportion and second proportion sum to unity and wherein, in the first array, the first proportion and the second proportion are not preset, and

in response to determining that the operation information does not comprise the second preset operation, further parsing the data in the operation information to verify that a portal used to enter the preset interface is a portal in search results interface obtained based on a third preset operation of the user, wherein the third preset operation of the user comprises inputting a search operation into the homepage, and generating a second array of only two elements,

wherein the two elements comprise a third proportion which represents the proportion of the first behavior and a fourth proportion which represents the proportion of the second behavior, wherein the third proportion and the fourth proportion sum to unity,

wherein in the second array, the third proportion and the fourth proportion are preset values,

wherein the first behavior represents that the user has a definite browsing object and the second behavior represents that the user has no definite browsing object;

determining recommendation information for the user based on the behavior data, and displaying the recommendation information in the preset interface, wherein objects represented by the recommendation information comprise objects of a same category as the main object displayed in the preset interface, and one or more objects of different categories from the main object, wherein the one or more objects of the different categories from the main object are associated with one or more of the plurality of interactive interface elements,

wherein the determining recommendation information for the user based on the behavior data comprises:

allocating, based on a total number of the preset display regions in the preset display area in the preset interface and in proportion to one of the first proportion or the third proportion of the first behavior, a corresponding first number of the preset display regions in the preset display area that display a number of the objects of the same category as the main object, and

allocating, based on the total number of preset display regions in the preset display area in the preset interface and in proportion to one of the second proportion or the fourth proportion of the second behavior, a corresponding second number of the preset display regions in the preset display area that display a number of the objects of the different categories from the main object; and

outputting instructions to the display device to cause the display device to render the number of the objects of the same category as the main object in the allocated first number of the preset display regions, and the number of the objects of the different categories from the main object in the allocated second number of the preset display regions, such that the objects of the same category as the main object and the objects of the different categories from the main object are displayed at the same time in the allocated first number of the preset display regions and the allocated second number of the preset display regions,

wherein the operations further comprise training the pre-trained behavior data recognition model on historical operation data comprising a training sample set, and using labels as a target metric to thereby obtain the pre-trained data behavior recognition model, such that, via the training, the pre-trained data behavior recognition model is configured to internally perform classification into the first behavior and the second behavior in response to input of the operation information, and then output corresponding values as a first element and a second element which express the proportions of the first behavior and the second behavior, and

wherein the training the pre-trained behavior data recognition model further comprises acquiring the training sample set and acquiring the labels comprises acquiring log information, dividing the acquired log information according to a unit of the preset length of time to obtain a plurality of pieces of sample operation information, and obtaining, for each piece of sample operation information in the plurality of pieces of sample operation information, based on a number of object categories comprised in the sample operation information, and a number of page views for each category in the sample operation information, a sample label for the sample operation information.

6 . The apparatus according to claim 5 , wherein determining the behavior data of the user based on the operation information further comprises determining, in response to determining that the portal is a portal in a preset recommendation interface, a third preset array representing a fifth proportion of the first behavior and a sixth proportion of the second behavior.

7 . The apparatus according to claim 5 , wherein the pre-trained behavior data recognition model is obtained by training through a following method:

acquiring the training sample set, wherein training samples in the training sample set comprise: sample operation information with the unit of the preset length of time and sample labels; and

training, using a machine learning algorithm, using the sample operation information in the training samples as input of an initial behavior data recognition model, and using the sample labels corresponding to the input sample operation information as expected output of the initial behavior data recognition model, to obtain the pre-trained behavior data recognition model.

8 . A non-transitory computer readable medium, storing a computer program thereon, wherein, the computer program, when executed by a processor, implements operations of allocating resources in a graphical user interface (GUI) of an e-commerce application on a display device to reduce user interface switching events, the operations comprising:

receiving, from the display device, a first preset operation of the user, wherein the first preset operation comprises selecting, via a click operation, an interactive interface element;

acquiring, in response to determining that a user enters a preset interface from a homepage of a shopping application based on receiving the first preset operation of the user, operation information within a preset length of time before the user performs the first preset operation,

wherein the preset interface comprises a goods details interface configured to display a main object and further comprises a preset display area having a plurality of preset display regions, each of the preset display regions displaying an object other than the main object, and

wherein the homepage displays a plurality of interactive interface elements corresponding to objects of different categories from the main object;

determining behavior data of the user based on the operation information, comprising:

inputting, in response to receiving, via a click operation, a second preset operation comprising selection of an object category or a selectable object displayed in the preset interface and determining that the operation information comprises a second preset operation, the operation information into a pre-trained behavior data recognition model comprising a neural network, wherein the pre-trained behavior data recognition model is configured to: receive the operation information, classify a user behavior in the operation information according to learning from training, and output a first array of only two elements;

outputting the first array of two elements, wherein one of the two elements is a first proportion which represents a proportion of a first behavior and the other of the two elements is a second proportion which represents a proportion of a second behavior,

wherein the first proportion and second proportion sum to unity,

wherein in the first array, the first proportion and the second proportion are not preset, wherein the pre-trained behavior data recognition model was trained on historical user operation data and used labels as a target metric to train the behavior data recognition model and thereby obtain the pre-trained data behavior recognition model, such that the pre-trained data behavior recognition model is configured to internally perform classification into the first behavior and the second behavior in response to being input with the operation information,

determining, in response to determining that the operation information does not comprise the second preset operation, a portal to enter the preset interface and generating a second array of two elements, in response to determining that the portal is a portal in a search results interface obtained based on a third preset operation of the user, wherein the third preset operation of the user comprises inputting a search operation into the homepage, and wherein the two elements comprise a third proportion which represents the proportion of the first behavior and a fourth proportion which represents the proportion of the second behavior, wherein the third proportion and the fourth proportion sum to unity, wherein, in the second array, the third proportion and the fourth proportion are preset,

wherein the first behavior represents that the user has a definite browsing object and the second behavior representing represents that the user has no definite browsing object;

determining recommendation information for the user based on the behavior data, and displaying the recommendation information in the preset interface, wherein objects represented by the recommendation information comprise objects of a same category as the main object displayed in the preset interface, and one or more objects of different categories from the main object, wherein the one or more objects of the different categories from the main object are associated with one or more of the plurality of interactive interface elements,

wherein the determining the recommendation information for the user based on the behavior data comprises:

allocating, based on a total number of the preset display regions in a preset display area in the preset interface and in proportion to one of the first proportion or the third proportion of the first behavior, a corresponding second number of the preset display regions in the preset display area that display a number of the objects of the same category as the main object; and

allocating, based on the total number of preset display regions in the preset display area in the preset interface and in proportion to one of the second proportion or the fourth proportion of the second behavior, a corresponding second number of the preset display regions in the preset display area that display a number of the objects of the different categories from the main object; and

outputting instructions to the display device to cause the display device to render the number of the objects of the same category as the main object in the allocated first number of the preset display regions, and the number of the objects of the different categories from the main object in the allocated second number of the preset display regions, such that the objects of the same category as the main object and the objects of the different categories from the main object are displayed at the same time in the allocated first number of the preset display regions and the allocated second number of the preset display regions.

9 . The non-transitory computer readable medium according to claim 8 , wherein determining the behavior data of the user browsing the preset interface based on the operation information further comprises:

determining, in response to determining that the portal is a portal in a preset recommendation interface, a third preset array representing a fifth proportion of the first behavior and a sixth proportion of the second behavior.

10 . The non-transitory computer readable medium according to claim 8 , wherein the pre-trained behavior data recognition model is obtained by training through a following method:

acquiring a training sample set, wherein training samples in the training sample set comprise: sample operation information with a unit of the preset length of time and sample labels; and

training, using a machine learning algorithm, using the sample operation information in the training samples as input of an initial behavior data recognition model, and using the sample labels corresponding to the input sample operation information as expected output of an initial behavior data recognition model, to obtain the pre-trained behavior data recognition model.

11 . The computer-implemented method according to claim 1 , wherein:

the total number of preset display regions in the preset display area comprises a total number of preset areas configured for displaying objects of the same category as the main object and objects of a different category from the main object which are available for use, and

in the second array, the third proportion of the first behavior is equal to the the fourth proportion of the second behavior.

12 . The method according to claim 1 , wherein the allocating, based on the total number of the preset display regions in the preset display area in the preset interface and in proportion to one of the first proportion or the third proportion of the first behavior, the corresponding first number of the preset display regions in the preset display area for displaying the number of the objects of the same category as the main object, comprises:

generating the first number by multiplying the total number of the preset display regions by the first proportion or the third proportion, wherein the first number is a number of the preset display regions to be allocated for displaying one or more objects of the same category as the main object, and

the allocating, based on the total number of the preset display regions in the preset display area in the preset interface and in proportion to one of the second proportion or the fourth proportion of the second behavior, the corresponding second number of the preset display regions in the preset display area for displaying the number of the objects of the different categories from the main object, comprises: generating the second number by multiplying the total number of the preset display regions by the second proportion, wherein the corresponding second number is a number of the preset display regions allocated for displaying the one or more objects of the different categories from the main object.

13 . The non-transitory computer readable medium according to claim 8 , wherein the allocating, based on the total number of the preset display regions in the preset display area in the preset interface and in proportion to one of the first proportion or the third proportion of the first behavior, the corresponding first number of the preset display regions in the preset display area for displaying the number of the objects of the same category as the main object, comprises:

generating the first number by multiplying the total number of the preset display regions by the first proportion, wherein the first number is a number of the preset display regions to be allocated for displaying one or more objects of the same category as the main object, and

the allocating, based on the total number of the preset display regions in the preset display area in the preset interface and in proportion to one of the second proportion or the fourth proportion of the second behavior, the corresponding second number of the preset display regions in the preset display area for displaying the number of the objects of the different categories from the main object, comprises: generating the second number by multiplying the total number of the preset display regions by the second proportion, wherein the corresponding second number is a number of the preset display regions allocated for displaying the one or more objects of the different categories from the main object.