CN111523032A

CN111523032A - Method, device, medium and electronic equipment for determining user preference

Info

Publication number: CN111523032A
Application number: CN202010322272.9A
Authority: CN
Inventors: 李嘉晨; 郭凯; 刘洋; 付东东; 刘雷; 胡磊
Original assignee: Beike Technology Co Ltd
Current assignee: Seashell Housing Beijing Technology Co Ltd
Priority date: 2020-04-22
Filing date: 2020-04-22
Publication date: 2020-08-11

Abstract

A method, apparatus, medium, and electronic device for determining user preferences are disclosed. The method comprises the following steps: acquiring behavior characteristic information of user behavior, wherein the behavior characteristic information comprises: a plurality of behavioral characteristic elements; predicting the behavior value of the user behavior relative to the target behavior according to each behavior feature element contained in the behavior feature information; and determining the user preference of the user according to the behavior object attribute information corresponding to the at least one user behavior of the user and the predicted behavior value of the at least one user behavior relative to the target behavior. The technical scheme provided by the disclosure is beneficial to improving the accuracy of determining the user preference, thereby being beneficial to improving the occurrence probability of the target behavior.

Description

Method, device, medium and electronic equipment for determining user preference

Technical Field

The present disclosure relates to computer technologies, and in particular, to a method for determining user preferences, an apparatus for determining user preferences, a storage medium, and an electronic device.

Background

Because the user preference can reflect the interest of the user, the user preference is often considered when providing recommendation information for the user, so that the recommendation information which is interested in the user can be provided for the user, and further better service is provided for the user. How to accurately determine the user preference is a technical problem of great concern.

Disclosure of Invention

The present disclosure is proposed to solve the above technical problems. Embodiments of the present disclosure provide a method for determining user preferences, an apparatus for determining user preferences, a storage medium, and an electronic device.

According to an aspect of an embodiment of the present disclosure, there is provided a method of determining user preferences, the method including: acquiring behavior characteristic information of user behavior, wherein the behavior characteristic information comprises: a plurality of behavioral characteristic elements; predicting the behavior value of the user behavior relative to the target behavior according to each behavior feature element contained in the behavior feature information; and determining the user preference of the user according to the behavior object attribute information corresponding to the at least one user behavior of the user and the predicted behavior value of the at least one user behavior relative to the target behavior.

In an embodiment of the present disclosure, the acquiring behavior feature information of a user behavior includes: obtaining incremental business data based on user behaviors; acquiring a plurality of behavior characteristic elements of the user behavior according to the field content of a preset field contained in the incremental business data; wherein the behavior feature information of the user behavior comprises: discrete behavior feature elements, and/or continuous behavior feature elements.

In another embodiment of the present disclosure, the predicting, according to each behavior feature element included in the behavior feature information, a behavior value of the user behavior with respect to a target behavior includes: determining a value prediction model corresponding to the user behavior according to the behavior type of the user behavior; the user behaviors of different behavior types correspond to different value prediction models; behavior characteristic elements contained in the behavior characteristic information of the user behaviors are used as model input and provided for a value prediction model corresponding to the user behaviors; performing prediction processing on the probability of converting the user behavior into the target behavior according to the model input of the value prediction model corresponding to the user behavior; and acquiring the behavior value of the user behavior relative to the target behavior according to the probability value output by the value prediction model.

In yet another embodiment of the present disclosure, the method further comprises: respectively acquiring behavior characteristic information samples of the user behaviors of a plurality of users aiming at each user behavior type according to stock business data based on the user behaviors; aiming at any behavior type, respectively taking a plurality of behavior characteristic elements contained in behavior characteristic information samples of a plurality of user behaviors with the behavior type as model input and providing the model input to a value prediction model to be trained corresponding to the user behaviors with the behavior type; respectively inputting the value prediction models to be trained according to the respective models, and performing prediction processing on the probability of converting the corresponding user behaviors into target behaviors; and adjusting the model parameters of the value prediction models to be trained according to the labels which are used for representing whether the user behaviors are converted into the target behaviors in the behavior characteristic information samples and the corresponding probability values output by the value prediction models to be trained.

In another embodiment of the present disclosure, the obtaining, according to the inventory business data based on the user behaviors, behavior feature information samples of the user behaviors of a plurality of users for each user behavior type includes: respectively acquiring a plurality of behavior characteristic elements of the user behaviors of a plurality of users aiming at each user behavior type according to the field content of a preset field contained in the stock business data; judging whether target behaviors exist in the user behaviors of the plurality of users within a preset time range after the user behaviors occur or not according to the stock business data; for each user behavior with a target behavior, setting a first label for representing and converting the corresponding user behavior characteristic information into the target behavior, and obtaining a positive behavior characteristic information sample; and for each user behavior without the target behavior, setting a second label for representing the non-converted target behavior for the corresponding user behavior characteristic information, and obtaining a negative behavior characteristic information sample.

In yet another embodiment of the present disclosure, the determining, according to behavior object attribute information corresponding to each of at least one user behavior of the user and a behavior value of each of the predicted at least one user behavior with respect to a target behavior, a user preference of the user includes: counting the predicted behavior values according to enumeration values under the behavior object attribute elements corresponding to at least one user behavior of the user; and determining the user preference of the user according to the statistical result.

In another embodiment of the present disclosure, the counting, according to enumerated values under attribute elements of each behavior object corresponding to at least one user behavior of the user, values of each predicted behavior, includes: cooling the predicted behavior values according to the occurrence time of each user behavior; and counting the behavior values after the cooling treatment according to the enumeration values under the behavior object attribute elements corresponding to the at least one user behavior of the user.

In yet another embodiment of the present disclosure, the determining, according to behavior object attribute information corresponding to each of at least one user behavior of the user and a behavior value of each of the predicted at least one user behavior with respect to a target behavior, a user preference of the user further includes: acquiring a behavior identifier of the at least one user behavior; and acquiring enumeration values under the behavior object attribute elements corresponding to the behavior identifiers of the user behaviors according to the behavior identifiers of the at least one user behavior.

According to another aspect of the embodiments of the present disclosure, there is provided an apparatus for determining user preferences, the apparatus including: an obtaining module, configured to obtain behavior feature information of a user behavior, where the behavior feature information includes: a plurality of behavioral characteristic elements; the prediction module is used for predicting the behavior value of the user behavior relative to the target behavior according to each behavior feature element contained in the behavior feature information acquired by the acquisition module; and the preference determining module is used for determining the user preference of the user according to the behavior object attribute information corresponding to the at least one user behavior of the user and the behavior value of the at least one user behavior predicted by the predicting module relative to the target behavior.

In an embodiment of the present disclosure, the obtaining module includes: the first submodule is used for acquiring incremental business data based on user behaviors; the second sub-module is used for acquiring a plurality of behavior characteristic elements of the user behavior according to the field content of the preset field contained in the incremental business data acquired by the first sub-module; wherein the behavior feature information of the user behavior comprises: discrete behavior feature elements, and/or continuous behavior feature elements.

In yet another embodiment of the present disclosure, the prediction module includes: the third sub-module is used for determining a value prediction model corresponding to the user behavior according to the behavior type of the user behavior; the user behaviors of different behavior types correspond to different value prediction models; the fourth sub-module is used for inputting behavior characteristic elements contained in the behavior characteristic information of the user behavior as models and providing the model with a value prediction model corresponding to the user behavior so as to predict the probability of converting the user behavior into a target behavior according to the model input by the value prediction model corresponding to the user behavior; and the fifth submodule is used for acquiring the behavior value of the user behavior relative to the target behavior according to the probability value output by the value prediction model.

In yet another embodiment of the present disclosure, the apparatus further includes: a training module, the training module comprising: the sixth submodule is used for respectively acquiring behavior characteristic information samples of the user behaviors of a plurality of users aiming at each user behavior type according to the stock business data based on the user behaviors; a seventh sub-module, configured to, for any behavior type, respectively use multiple behavior feature elements included in behavior feature information samples of multiple user behaviors having the behavior type as model inputs, provide the model with a value prediction model to be trained corresponding to the user behavior having the behavior type, and perform prediction processing on probabilities that the corresponding user behaviors are respectively converted into target behaviors through the value prediction models to be trained according to the respective model inputs; and the eighth submodule is used for adjusting the model parameters of the value prediction models to be trained according to the labels which are used for representing whether the user behaviors are converted into the target behaviors in the behavior characteristic information samples and the corresponding probability values output by the value prediction models to be trained.

In yet another embodiment of the present disclosure, the sixth submodule is further configured to: respectively acquiring a plurality of behavior characteristic elements of the user behaviors of a plurality of users aiming at each user behavior type according to the field content of a preset field contained in the stock business data; judging whether target behaviors exist in the user behaviors of the plurality of users within a preset time range after the user behaviors occur or not according to the stock business data; for each user behavior with a target behavior, setting a first label for representing and converting the corresponding user behavior characteristic information into the target behavior, and obtaining a positive behavior characteristic information sample; and for each user behavior without the target behavior, setting a second label for representing the non-converted target behavior for the corresponding user behavior characteristic information, and obtaining a negative behavior characteristic information sample.

In yet another embodiment of the present disclosure, the determining the preference module includes: a ninth sub-module, configured to count, according to an enumeration value under each behavior object attribute element corresponding to each at least one user behavior of the user, values of each predicted behavior; a tenth submodule, configured to determine the user preference of the user according to the statistical result.

In yet another embodiment of the present disclosure, the ninth sub-module is further configured to: cooling the predicted behavior values according to the occurrence time of each user behavior; and counting the behavior values after the cooling treatment according to the enumeration values under the behavior object attribute elements corresponding to the at least one user behavior of the user.

In another embodiment of the present disclosure, the determining a preference module further includes: an eleventh submodule, configured to obtain a behavior identifier of the at least one user behavior; and the twelfth submodule is used for acquiring enumeration values under the behavior object attribute elements corresponding to the behavior identifiers of the user behaviors according to the behavior identifiers of the at least one user behavior.

According to still another aspect of the embodiments of the present disclosure, there is provided a computer-readable storage medium storing a computer program for executing the above-mentioned method of determining user preferences.

According to still another aspect of the embodiments of the present disclosure, there is provided an electronic apparatus including: a processor; a memory for storing the processor-executable instructions; the processor is used for reading the executable instructions from the memory and executing the instructions to realize the method for determining the user preference.

Based on the method and the device for determining the user preference provided by the above embodiment of the present disclosure, the behavior value of the user behavior is predicted by using each behavior feature element included in the behavior feature information, and since the behavior feature element included in the behavior feature information can describe a user behavior from multiple aspects, the present disclosure can enable two user actions that are identical in terms of behavior amount to have different behavior values due to different behavior feature elements included in the respective behavior feature information; that is, since the present disclosure converts discrete values for counting, such as 0 and 1, into continuous values, such as behavioral value, the present disclosure can sufficiently distinguish two user actions; the behavior value of the user behavior in the disclosure can represent the possibility that the user behavior is converted into the target behavior, and the target behavior of the user often reflects the user preference, so that the user preference is determined by utilizing the behavior value of each user behavior, and the user behaviors with different behavior values can play different roles in the process of determining the user preference. Therefore, the technical scheme provided by the disclosure is beneficial to improving the accuracy of determining the user preference, and is further beneficial to improving the occurrence probability of the target behavior.

The technical solution of the present disclosure is further described in detail by the accompanying drawings and examples.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and together with the description, serve to explain the principles of the disclosure.

The present disclosure may be more clearly understood from the following detailed description, taken with reference to the accompanying drawings, in which:

FIG. 1 is a schematic diagram of one embodiment of a suitable scenario for use with the present disclosure;

FIG. 2 is a flow chart of one embodiment of a method of determining user preferences of the present disclosure;

FIG. 3 is a flow diagram of one embodiment of the present disclosure to predict behavioral value of a user behavior relative to a target behavior;

FIG. 4 is a flow diagram of one embodiment of training a predictive model according to the present disclosure;

FIG. 5 is a flowchart of one embodiment of the present disclosure for determining user preferences of the user based on behavior object attribute information and behavior value;

FIG. 6 is a schematic diagram illustrating an embodiment of an apparatus for determining user preferences according to the present disclosure;

fig. 7 is a block diagram of an electronic device provided in an exemplary embodiment of the present disclosure.

Detailed Description

Example embodiments according to the present disclosure will be described in detail below with reference to the accompanying drawings. It is to be understood that the described embodiments are merely a subset of the embodiments of the present disclosure and not all embodiments of the present disclosure, with the understanding that the present disclosure is not limited to the example embodiments described herein.

It should be noted that: the relative arrangement of the components and steps, the numerical expressions, and numerical values set forth in these embodiments do not limit the scope of the present disclosure unless specifically stated otherwise.

It will be understood by those of skill in the art that the terms "first," "second," and the like in the embodiments of the present disclosure are used merely to distinguish one element from another, and are not intended to imply any particular technical meaning, nor is the necessary logical order between them.

It is also understood that in embodiments of the present disclosure, "a plurality" may refer to two or more than two and "at least one" may refer to one, two or more than two.

It is also to be understood that any reference to any component, data, or structure in the embodiments of the disclosure, may be generally understood as one or more, unless explicitly defined otherwise or stated otherwise.

In addition, the term "and/or" in the present disclosure is only one kind of association relationship describing the associated object, and means that there may be three kinds of relationships, such as a and/or B, and may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the character "/" in the present disclosure generally indicates that the former and latter associated objects are in an "or" relationship.

It should also be understood that the description of the various embodiments of the present disclosure emphasizes the differences between the various embodiments, and the same or similar parts may be referred to each other, so that the descriptions thereof are omitted for brevity.

Meanwhile, it should be understood that the sizes of the respective portions shown in the drawings are not drawn in an actual proportional relationship for the convenience of description.

The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the disclosure, its application, or uses.

Techniques, methods, and apparatus known to those of ordinary skill in the relevant art may not be discussed in detail but are intended to be part of the specification where appropriate.

It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, further discussion thereof is not required in subsequent figures.

Embodiments of the present disclosure may be implemented in electronic devices such as terminal devices, computer systems, servers, etc., which are operational with numerous other general purpose or special purpose computing system environments or configurations. Examples of well known terminal devices, computing systems, environments, and/or configurations that may be suitable for use with an electronic device, such as a terminal device, computer system, or server, include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, hand-held or laptop devices, microprocessor-based systems, set top boxes, programmable consumer electronics, network pcs, minicomputer systems, mainframe computer systems, distributed cloud computing environments that include any of the above, and the like.

Electronic devices such as terminal devices, computer systems, servers, etc. may be described in the general context of computer system-executable instructions, such as program modules, being executed by a computer system. Generally, program modules may include routines, programs, objects, components, logic, data structures, etc. that perform particular tasks or implement particular abstract data types. The computer system/server may be implemented in a distributed cloud computing environment. In a distributed cloud computing environment, tasks may be performed by remote processing devices that are linked through a communications network. In a distributed cloud computing environment, program modules may be located in both local and remote computer system storage media including memory storage devices.

Summary of the disclosure

In implementing the present disclosure, the inventors found that, at present, the implementation manner of determining the user preference is generally: for any user, in a time range before the current time, the behavior quantities of the user on the enumerated values under the behavior object attribute are accumulated, then, a linear weighting method is used for calculating all the behavior quantities of the user, the ratios of the behavior quantities on the enumerated values to all the behavior quantities are calculated, each ratio is used as the preference degree of the user on the corresponding enumerated value, and the user preference can be obtained by comparing all the ratios. However, since the behavior amount is a measure for the number of times of the behavior, and the like, and cannot reflect the difference between different behaviors, the user preference cannot be accurately determined by using the ratio calculated by the behavior amount.

For example, assuming that a user performs two actions on the same enumerated value under the same behavior object attribute or a user performs one action on each of the two enumerated values under the same behavior object attribute, if in one of the actions, the user performs multiple rounds (for example, 20 rounds) of conversations with a conversation party (for example, a house broker, etc.), and in the other action, the user does not perform a conversation with the conversation party, then it is obviously inappropriate to respectively mark the two actions as one action amount.

As another example, assume a behavior object attribute is a second-hand room, the behavior object attribute includes: a general house type and a cottage type, these two enumerated values. Assume that a user has N1 web browsing actions at an enumerated value of a general house type, and the user has N2 web browsing actions at an enumerated value of a cottage type. If N1 is greater than N2, existing implementations for determining user preferences determine that the user's preference level for the enumerated value for the common home type is somewhat higher than the user's preference level for the enumerated value for the cottage type. However, since there is often a large gap in the supply quantities of ordinary houses and villas, for example, 95% of the premises in a city are ordinary houses and 5% are villas; therefore, although the user browses the web page related to the cottage less frequently than the web page related to the general home, it does not necessarily mean that the user's preference degree in the enumerated value of the general home type is higher than the user's preference degree in the enumerated value of the cottage type.

Brief description of the drawings

One example of an application scenario for determining user preferences provided by the present disclosure is illustrated in fig. 1.

In fig. 1, a user 100 uses a terminal device 101 (such as a computer or a smart mobile phone) of the user to access a server 102 on a network side, the server 102 pushes corresponding information to the terminal device 101 according to an access requirement of the user 100, and the terminal device 101 forms a corresponding page (such as a page in the form of a web page or a page in the form of an APP) according to the information pushed by the server 102 and displays the page to the user 100.

The present disclosure may form a piece of incremental service information based on the page browsing behavior of the user 100, and obtain a piece of behavior feature information of the user 100 by using the incremental service information, for example, extract field contents of n (n is an integer greater than 1) fields from the piece of incremental service information, where the field contents of the n fields are n behavior feature elements included in the behavior feature information of the user 100.

The present disclosure may obtain the behavior value of the page browsing behavior of the user 100 relative to the target behavior by using the behavior feature information. For example, the present disclosure may use a plurality of (e.g., all) behavior feature elements included in the piece of behavior feature information as model inputs, provide the model inputs to the value prediction model, and obtain the behavior value of the current page browsing behavior of the user 100 relative to the target behavior according to the output of the value prediction model.

The behavior object attribute information corresponding to the page browsing behavior and the behavior value of the page browsing behavior relative to the target behavior can be used as a basic data storage for determining the user preference. The present disclosure may obtain and store a piece of basic data of the user 100 by performing the above operation each time an incremental service information is formed.

When the user preference of the user 100 needs to be determined, the user preference of the user 100 can be obtained according to the value statistical result by performing behavior value statistics (such as behavior value weighted statistics based on time weight) for the attribute information of the behavior object on all the currently stored basic data of the user 100. For example, the user preferences of the user 100 obtained by the present disclosure are, from high to low according to the preference degrees, respectively: 140-180 square meters of second-hand villas, 80-100 square meters of second-hand two-room ordinary houses, 40-60 square meters of second-hand one-room ordinary houses, 80-100 square meters of new-room two-room ordinary houses and the like.

Exemplary method

FIG. 2 is a flow chart of one embodiment of a method of determining user preferences of the present disclosure. The method of the embodiment shown in fig. 2 comprises the steps of: s200, S201, and S202. The following describes each step.

And S200, acquiring behavior characteristic information of user behaviors.

The behavior feature information in the present disclosure may refer to information for describing a feature that a user has in a process of performing a user behavior. Generally, a user behavior has a piece of behavior feature information, and the piece of behavior feature information may include: a plurality of behavioral characteristic elements. The behavior feature element in the present disclosure may refer to a basic information unit for describing a feature that a user has in performing a user behavior. It can also be considered that a user behavior has a plurality of feature points, each feature point is a behavior feature element, and all the feature points are combined together to describe a complete user behavior. A piece of behavior feature information may include not only a plurality of behavior feature elements, but also behavior identifiers for characterizing a unique user behavior.

Each behavior characteristic element contained in the behavior characteristic information in the present disclosure can be set according to the specific requirements of the actual application field. For example, the behavioral characteristic information may include: the method comprises the following steps of obtaining behavior characteristic elements such as the geographic position of a user, the number of conversation rounds of the user and a conversation party, the entrance of the user to execute user behaviors, the occurrence time of the user behaviors, the entrance of business opportunities, the occurrence time of the business opportunities and the like. A business opportunity may refer to an opportunity to achieve a business goal or goal. For example, the business may have successfully left a contact for the user or the user has opened a page to communicate with the conversing party, etc. The method and the device can obtain the specific value of each behavior characteristic element in the behavior characteristic information of the user behavior by using the service data.

S201, predicting the behavior value of the user behavior relative to the target behavior according to the behavior feature elements contained in the behavior feature information of the user behavior.

The behavior value of the user behavior relative to the target behavior in the present disclosure may refer to a possibility that the user performs the target behavior by continuing to perform the subsequent corresponding user behavior after performing the user behavior, and finally performing the preset target behavior. The target behavior can be set according to the specific requirements of the actual application field. For example, the target behavior may be a targeted transaction behavior or a successful commitment behavior or a user's house-in-watch behavior, and so on. The present disclosure is not limited thereto.

Predicting a behavior value of a user behavior relative to a target behavior in the present disclosure may refer to predicting a likelihood that a user will execute a target behavior within a predetermined time period (e.g., N days) in the future on the premise that the user behavior is executed. The method and the device can process the behavior characteristic information of the user behavior by utilizing a preset model, so that the behavior value of the user behavior relative to the target behavior is obtained.

S202, determining the user preference of the user according to behavior object attribute information corresponding to at least one user behavior of the user and the predicted behavior value of the at least one user behavior relative to the target behavior.

The behavior object in the present disclosure may refer to a target to which a user behavior is directed. Behavior objects may appear as different content in different application domains. For example, in the field of real estate, an behavioral object in the present disclosure may be a house. As another example, in the field of retail goods, the behavioral object in the present disclosure may be a retail good or the like.

The behavior object attribute information in the present disclosure may refer to information for describing characteristics that the behavior object itself has. For example, when the behavior object is a house, the behavior object attribute information includes: a plurality of attribute elements of house property, house location, house area, hall structure, house type, and house structure.

User preferences in this disclosure may refer to a user's tendencies in behavior object attribute information. For example, for the real estate domain, the finalized user preferences may include: new houses, city district, 60-80 square meters, two rooms and one hall, common houses, brick-concrete structures and the like.

The behavior value of the user behavior is predicted by utilizing the behavior characteristic elements contained in the behavior characteristic information, and because the behavior characteristic elements contained in the behavior characteristic information can describe the user behavior from multiple aspects, two user actions which are completely the same in the aspect of the behavior quantity can have different behavior values due to different behavior characteristic elements contained in the respective behavior characteristic information; that is, since the present disclosure converts discrete values for counting, such as 0 and 1, into continuous values, such as behavioral value, the present disclosure can sufficiently distinguish two user actions; the behavior value of the user behavior in the disclosure can represent the possibility that the user behavior is converted into the target behavior, and the user preference can be reflected more often when the target behavior occurs to the user, so that the user preference is determined by utilizing the behavior value of each user behavior, and the user behaviors with different behavior values can play different roles in the process of determining the user preference. Therefore, the technical scheme provided by the disclosure is beneficial to improving the accuracy of determining the user preference, and is further beneficial to improving the occurrence probability of the target behavior.

In one optional example, the present disclosure may utilize business data based on user behavior to obtain behavior feature information of user behavior. The user behavior-based business data in the present disclosure may refer to business data that is generated by operating a device such as a computer or a smart mobile phone to describe a user behavior due to a user's own person and/or a network-side maintenance person or the like. The service data in the present disclosure may include information such as an operation log or an access log formed on the server side.

Optionally, the present disclosure may obtain the behavior feature information of the user behavior by using the incremental service data based on the user behavior. For example, when a piece of incremental business data based on the user behavior is formed, the field content of the corresponding field may be extracted from the piece of incremental business data by using the corresponding field in the incremental business data corresponding to each behavior feature element, so as to obtain a plurality of behavior feature elements (e.g., all behavior feature elements) of the user behavior. For another example, the present disclosure may perform the extraction operation of the field content on each incremental service data formed in a current period of time at a predetermined time, to obtain behavior feature information of a plurality of user behaviors. As can be seen from the above description, the behavior feature elements in the present disclosure generally correspond to corresponding fields in the incremental business data. It should be particularly noted that, the present disclosure may also obtain corresponding behavior feature elements by performing corresponding calculation, conversion, mapping, or other processing on field contents of corresponding fields in the incremental service data.

Optionally, one behavior feature element in the present disclosure may be a discrete behavior feature element, and may also be a continuous behavior feature element. The discrete behavior feature element may be a non-ordinal-based discrete behavior feature element, or an ordinal-based discrete behavior feature element. The non-ordinal-based discrete behavior feature elements may refer to non-numeric behavior feature elements. For example, the non-ordinal based discrete behavior feature elements may be gender, home address, native place, entry to generate business opportunities, or user text ratings, etc. The ordinal-based discrete behavior feature elements may refer to behavior feature elements that have discontinuous numerical values and are usually obtained by numerical methods. For example, an ordinal-based discrete behavior feature element may be: the number of conversation turns between the user and the conversation party, the number of times of business opportunity generation, the date of business opportunity generation, the week to which the date belongs, the number of participants (such as the number of participants with house behavior of the user, etc.), the PV (PageView) amount of the user, or the rating level given by the user, etc. The continuous behavior feature element may be in a digital form, and the value of the number may be any behavior feature element of a numerical value in a numerical value interval. For example, the continuous behavior feature elements may be: the time of business opportunity generation, the time of user starting to browse the page, or the time of user finishing browsing the page, etc.

The method and the device can obtain the behavior characteristic information of the user behavior in time by utilizing the incremental business data, so that the behavior values of the user behavior of the user relative to the target behavior can be accumulated at any time, and therefore the user preference can be conveniently and rapidly determined according to all the accumulated behavior values at any time. Because the behavior feature information of the present disclosure may include a discrete behavior feature element or a continuous behavior feature element, the present disclosure may describe the user behavior from multiple dimensions, so that a phenomenon that the user behavior is described only from a behavior amount and a difference between different user behaviors (as different user behaviors under a behavior type) cannot be effectively distinguished can be avoided, that is, the behavior feature information in the present disclosure may sufficiently embody a difference between different user behaviors (as different user behaviors under a behavior type), and the present disclosure may accurately obtain a behavior value of the user behavior relative to a target behavior by using the behavior feature information.

In one optional example, each user behavior in the present disclosure has a corresponding behavior type. The behavior type of the user behavior can be set according to the requirements of the actual application field. For example, in the real estate domain, the behavior types of user behavior may include: PV behavior of the house details page, keyword search behavior, user house attention behavior, user business generation behavior, and user house-with-watch behavior, among others. The number of behavior types and the specific content of the behavior types are not limited by this disclosure.

Optionally, the present disclosure may utilize a value prediction model successfully trained in advance to predict the behavior value of the user behavior relative to the target behavior. The method can preset a value prediction model for each behavior type, and the behavior types and the value prediction models are generally in one-to-one relationship, namely different behavior types correspond to different value prediction models. One example of predicting the behavior value of a user behavior relative to a target behavior using a value prediction model successfully trained in advance is shown in fig. 3.

In fig. 3, S300, a value prediction model corresponding to a user behavior is determined according to a behavior type of the user behavior.

Optionally, the present disclosure may determine the behavior type of the user behavior according to information (such as a behavior type identifier) included in the behavior feature information of the user behavior. For example, if the behavior feature information of a user behavior includes a behavior type identifier 1, it indicates that the behavior type of the user behavior is a user-generated business opportunity behavior type, and the behavior type identifier 1 corresponds to the value prediction model 1; and if the behavior type identifier contained in the behavior feature information of a user behavior is 2, the behavior type of the user behavior is the house-watching behavior type of the user, and the behavior type identifier 2 corresponds to the value prediction module 2.

Optionally, under the condition that the behavior feature information of the user behaviors of different behavior types is distinguished in advance and stored in different information sets, the value prediction model corresponding to the user behavior can be determined according to the set identifier of the information set where the behavior feature information is located. For example, the user behaviors of which the behavior types are the types of business opportunity behavior generated by the user are all stored in the information set 1, and the user behaviors of which the behavior types are the types of house behavior with watching are all stored in the information set 2. Information set 1 corresponds to value prediction model 1 and information set 2 corresponds to value prediction model 2.

Optionally, the information set 1 may specifically be a business behavior feature table generated by the user, and the information set 2 may specifically be a house behavior feature table with a watch of the user. The user generating the business behavior feature table may include: the business opportunity identification, the number of conversation turns of the business opportunity user and the conversation party, the entrance for generating the business opportunity, the business opportunity generation date, the week to which the business opportunity generation date belongs, the time for generating the business opportunity and the like. The watch-with-house behavior feature table of the user may include: the list items of the house with watch identification, the evaluation of the house with watch broker by the user, the start time of the house with watch, the end time of the house with watch, the number of people with watch, the Nth time of the current house with watch, and the like.

S301, behavior characteristic elements contained in behavior characteristic information of user behaviors are used as model input and provided for a value prediction model corresponding to the user behaviors.

Optionally, the present disclosure may form the behavior feature information of the user behavior into a model input according to a preset format of the model input. For example, the present disclosure may perform normalization or type conversion on corresponding behavior feature elements in the behavior feature information, so that all behavior feature elements in the behavior feature information satisfy the input requirement of the value prediction model. In a general case, all behavior feature elements included in a piece of behavior feature information are input as a piece of model.

Optionally, each value prediction model in the present disclosure may be: a value prediction model formed from a decision tree (e.g., Xgboost). The training process for each value prediction model can be seen in the description below with respect to fig. 4.

And S302, predicting the probability of converting the user behavior into the target behavior according to the model input through the value prediction model corresponding to the user behavior.

Optionally, the value prediction model in the present disclosure forms a probability value within a predetermined value range for each model input. For example, the predetermined range of values may be 0-1. The probability value may indicate a probability that the user continues to execute the subsequent corresponding user behavior after executing the user behavior, and finally executes the preset target behavior, that is, a possibility that the user behavior is finally converted into the target behavior.

And S303, acquiring the behavior value of the user behavior relative to the target behavior according to the probability value output by the value prediction model.

Optionally, the present disclosure may directly use the probability value output by the value prediction model as the behavior value of the user behavior relative to the target behavior, or may further process (e.g., map) the probability value output by the value prediction model, and use the processing result as the behavior value of the user behavior relative to the target behavior.

According to the method and the device, a value prediction model (such as a value prediction model formed by Xgboost) is respectively set for the user behavior of each behavior type, so that each value prediction model can carry out value prediction processing in a targeted manner, and the behavior value of the user behavior relative to the target behavior can be conveniently and accurately obtained.

In an alternative example, one process of the present disclosure for training predictive models is illustrated in FIG. 4.

In fig. 4, S400, behavior feature information samples of user behaviors of a plurality of users are respectively obtained for each user behavior type according to inventory business data based on the user behaviors.

Optionally, any behavior feature element in each behavior feature information sample in the present disclosure generally corresponds to a corresponding field in the inventory incremental business data. The method and the device can utilize the corresponding fields in the stock business data corresponding to each behavior feature element to respectively extract the field contents of the corresponding fields from each stock business data, thereby obtaining a plurality of behavior feature elements in each behavior feature information sample of a user behavior. The present disclosure may also obtain corresponding behavior feature elements by performing processing such as calculation, conversion, or mapping on field contents of corresponding fields in the stock business data.

Optionally, the inventory business data in the present disclosure may include a plurality of sets, each set corresponding to a type of behavior. When a behavior feature information sample of one behavior type needs to be obtained, a plurality of behavior feature elements contained in the behavior feature information sample can be obtained from inventory business data contained in a corresponding set.

Optionally, behavior feature information samples in the present disclosure include behavior feature elements that are the same as feature elements included in the behavior feature information in the above embodiments, and the behavior feature information samples in the present disclosure include label information. The marking information carried by the behavior characteristic information sample can show that: for a user behavior of a user, whether the target behavior exists within a predetermined time range after the user behavior occurs. If the target behavior exists within a predetermined time range after the user behavior occurs, the label information of the behavior feature information sample of the user behavior may be set to a first label (e.g., 1), otherwise, the label information of the behavior feature information sample of the user behavior may be set to a second label (e.g., 0). The first label can characterize that the user behavior can be converted into the target behavior within a preset time range. Wherein the second label may characterize that the user behavior is unable to be converted into the target behavior within a predetermined time frame.

In a specific example, for a user behavior of a user, the time when the user behavior occurs is taken as the starting time, and within N days after the starting time, whether the user has the target behavior occurs or not is determined. If the target behavior occurs to the user within N days after the start time, the label information of the behavior feature information sample of the user behavior may be set to 1, otherwise, the label information of the behavior feature information sample of the user behavior may be set to 0.

Alternatively, all behavioral characteristic information samples with the first label may be referred to as positive samples. All behavioral characteristic information samples with the second label may be referred to as negative samples. The number of positive samples and the number of negative samples in training samples used in the present disclosure for a value prediction model to be trained are typically approximately the same. For example, the difference between the number of positive samples and the number of negative samples is typically less than a predetermined difference. Because the number of positive samples in the stock business data in the same time range is usually smaller than the number of negative samples, the time window corresponding to the stock business data of the positive samples is often larger than the time window corresponding to the stock business data of the negative samples. For example, the present disclosure may obtain negative samples from the business inventory data of the last half year and positive samples from the business inventory data of the last two years.

Optionally, the present disclosure may divide all the behavior feature information samples obtained into a training set, a testing set, and the like. The behavior feature information samples in the training set are used to adjust the model parameters. And the behavior characteristic information samples of the test set can be used for detecting whether the current value prediction model is trained successfully. In addition, the training set typically contains substantially the same number of positive samples and negative samples, and the test set typically contains substantially the same number of positive samples and negative samples.

S401, aiming at any behavior type, respectively taking the behavior characteristic elements in the behavior characteristic information sample of the user behavior with the behavior type as model input, and providing the model for the value prediction model to be trained corresponding to the user behavior with the behavior type.

Optionally, the present disclosure may provide each behavior feature element in the plurality of behavior feature information samples in the training set corresponding to each behavior type to the to-be-trained value prediction model corresponding to the corresponding behavior type, respectively. For example, the positive samples and the negative samples with the same number are respectively selected randomly from the training sets corresponding to each behavior type according to the preset batch processing number, all behavior characteristic elements in each positive sample and each negative sample are respectively used as model inputs of the to-be-trained value prediction models corresponding to the corresponding behavior types and are provided for the corresponding to-be-trained value prediction models.

S402, the corresponding user behaviors are subjected to prediction processing according to the probability of converting into the target behaviors respectively through the value prediction models to be trained according to the input of the models respectively.

Optionally, for any value prediction model to be trained, a probability value is formed for each model input of the value prediction model to be trained, and the probability value can represent the possibility that a user behavior corresponding to the model input can be converted into a target behavior in the future.

Optionally, the present disclosure may enable the behavior characteristic information sample of the user behavior to form the model input according to a preset format of the model input. For example, the method and the device can perform normalization or type conversion or mapping on corresponding behavior feature elements in the behavior feature information sample so that the behavior feature elements meet the input requirements of the value prediction model to be trained. In general, all behavior feature elements contained in a behavior feature information sample are input as a model of a value prediction model.

And S403, adjusting model parameters of the value prediction models to be trained according to labels used for representing whether the user behaviors are converted into the target behaviors in the behavior feature information samples and corresponding probability values output by the value prediction models to be trained.

Optionally, for any value prediction model to be trained, the method may perform loss calculation on each probability value output by the value prediction model to be trained and label information (label information is, for example, 0 or 1) of a corresponding user behavior feature information sample according to a preset loss function, and propagate a result of the loss calculation in the value prediction model to be trained so as to adjust model parameters of the value prediction model to be trained. The model parameters of the value prediction model to be trained may include structural parameters of a binary tree, and the like.

Optionally, for any value prediction model to be trained, when the training for the value prediction model to be trained reaches a predetermined iteration condition, the training process for the value prediction model to be trained is ended.

Optionally, the predetermined iteration condition in the present disclosure may include: and the accuracy of a prediction result obtained by aiming at the output of the user behavior characteristic information sample in the test set according to the value prediction model to be trained meets the preset requirement. And under the condition that the accuracy of a prediction result obtained by the value prediction model to be trained aiming at the output of the user behavior characteristic information sample in the test set reaches a preset requirement, successfully training the value prediction model to be trained this time. The predetermined iteration condition in the present disclosure may further include: and training a value prediction model to be trained, wherein the number of the user behavior characteristic information samples in the used training set reaches the requirement of the preset number, and the like. When the number of the used user behavior characteristic information samples meets the requirement of a preset number, and the accuracy of a prediction result obtained by the value prediction model aiming at the output of the user behavior characteristic information samples in the test set does not meet the preset requirement, the value prediction model to be trained is not trained successfully. The successfully trained value prediction model may be used to predict the behavioral value of the corresponding user behavior in the above embodiments.

In one optional example, each attribute element included in the behavior object attribute information in the present disclosure typically has multiple enumerated values. An enumerated value in each attribute element may be considered a slot value under the corresponding attribute element. For example, assume that the behavioral object in the present disclosure is a house, and the behavioral object attribute information includes: house nature, house location, house area, hall structure, house type, and house structure, these six attribute elements, then:

the properties of the house may include: new house and second-hand house, etc.;

the premises locations may include: a plurality of enumerated values in a second ring, between a second ring and a third ring, between a third ring and a fourth ring, between a fourth ring and a fifth ring, between a fifth ring and a sixth ring and outside the sixth ring;

the floor space may include: a plurality of enumeration values within 40 square meters, 40-60 square meters, 60-80 square meters, 80-100 square meters, 100-140 square meters, more than 140 square meters and the like;

the hall structure can include: a plurality of enumerated values such as a bay, a one-room one-hall, a two-room one-hall, a three-room one-hall, a four-room one-hall and at least a five-room one-hall;

the house types may include: at least two enumerated values such as common houses and villas;

the housing structure may include: brick-concrete structures and non-brick-concrete structures, etc.

Optionally, an example of determining the user preference of the user according to the behavior object attribute information corresponding to each of the at least one user behavior of the user and the predicted behavior value of each of the at least one user behavior of the user relative to the target behavior is shown in fig. 5.

In fig. 5, S500, behavior identifiers of user behaviors of the user are acquired.

Optionally, when the behavior value of a user behavior of the user is obtained, the behavior identifier uniquely representing the user behavior and the behavior value of the user behavior may be stored. For example, the user identification, the behavior identification of the user behavior, the behavior value of the user behavior, and the occurrence time of the user behavior may be stored as a record in the behavior price set. Thus, when the user preference of a user needs to be determined, the method and the device can search the behavior value set according to the user identification of the user, so that the behavior identifications of all user behaviors of the user and the behavior values of all user behaviors of the user can be obtained according to the search result.

And S501, searching enumeration values under each behavior object attribute element corresponding to the behavior identifier of each user behavior according to the obtained behavior identifier of each user behavior.

Optionally, for a behavior identifier of any user behavior of the user, the present disclosure may search in corresponding log information (for example, an operation log, etc.) by using the behavior identifier, that is, an enumeration value under each behavior object attribute element corresponding to the behavior identifier may be obtained. For example, the enumerated values obtained by using the behavior identifier of a user behavior of the user under the attribute elements of each behavior object include: a new house, a space between four rings and five rings, 60-80 square meters, a two-room and one-hall, a common house, a brick-concrete structure and the like. For another example, the enumerated values obtained by using the behavior identifier of another user behavior of the user under the attribute elements of the behavior objects include: a new house, four to five rings, 60-80 square meters, one room and one hall, a common house, a brick-concrete structure and the like.

S502, counting the predicted behavior values according to the enumeration values under the behavior object attribute elements corresponding to at least one user behavior of the user.

Optionally, the present disclosure may count the behavior value by taking the enumeration value under the behavior object attribute element as a unit, that is, if all user behaviors of the user relate to n different enumeration values in total, the present disclosure may perform the statistics n times, so as to obtain n statistical results, and each enumeration value corresponds to one statistical result.

As a more specific example, assuming that m1 user behaviors relate to the enumerated value a (e.g., new room) in the behavior object attribute element a and m2 user behaviors relate to the enumerated value b (e.g., second room) in the behavior object attribute element a among all the user behaviors of a user, the disclosure may accumulate the behavior values of the m1 user behaviors and accumulate the behavior values of the m2 user behaviors.

Optionally, accumulating the behavior values in the present disclosure generally means directly adding the behavior values of the user behaviors having the same enumeration value. In addition, the term "accumulating behavior values" in the present disclosure may mean adding behavior values of user behaviors having the same enumeration value based on a weight.

Optionally, the magnitude of the weight corresponding to the behavior value of each user behavior may be related to the time interval between the occurrence time of the user behavior and the current time. That is, the present disclosure may adopt a time cooling manner to count the behavior value of each user behavior of the user. That is, according to the present disclosure, the currently predicted behavior values may be cooled according to the occurrence time of each user behavior, and then, according to the enumeration values under the behavior object attribute elements corresponding to each user behavior of the user, the behavior values after cooling may be counted. The cooling process may be a cooling process based on a cooling coefficient, and the cooling coefficient may be considered as the above weight. A more specific example is as follows:

suppose a user has m user behaviors, namely a first user behavior, a second user behavior, … …, an m-1 st user behavior, and an m-th user behavior, and the first user behavior and the second user behavior relate to an enumerated value a (e.g., a new room) in a behavior object attribute element a, and the m-1 st user behavior and the m-th user behavior relate to an enumerated value b (e.g., a second room) in the behavior object attribute element a.

Suppose that the behavior value of the first user behavior of the user is v1, the date of occurrence of the first user behavior is t1 days from the current date, the behavior value of the second user behavior of the user is v2, the date of occurrence of the second user behavior is t2 days from the current date, the behavior value of the m-1 th user behavior of the user is vm-1, the date of occurrence of the m-1 th user behavior is tm-1 days from the current date, the behavior value of the m-th user behavior of the user is vm, and the date of occurrence of the m-th user behavior is tm days from the current date. Wherein t1 is greater than t2, tm-1 is greater than t2, and tm is greater than tm-1.

Under the above assumptions, the present disclosure may perform cooling processing on v1 with a cooling coefficient corresponding to t1 (e.g., a cooling coefficient corresponding to v1 × t 1), obtaining v 1'. V2 is cooled by the cooling coefficient corresponding to t2 (such as the cooling coefficient corresponding to v2 × t 2), and v 2' is obtained. And performing cooling treatment on vm-1 by using the cooling coefficient corresponding to tm-1 (such as the cooling coefficient corresponding to vm-1 × tm-1) to obtain vm-1'. And performing cooling treatment on vm by using the cooling coefficient corresponding to tm (such as the cooling coefficient corresponding to vm multiplied by tm), and obtaining vm'.

Thereafter, the present disclosure may calculate the sum of v1 'and v 2', obtaining s 1; and the sum of vm-1 'and vm' is calculated to obtain s 2. s1 is the statistical result of the enumerated value a, and s2 is the statistical result of the enumerated value b.

Alternatively, the cooling coefficient may be calculated based on the number of days from the current date of the occurrence date of the user behavior.

According to the method and the device, the behavior value of the user behavior is cooled based on the time interval between the occurrence time of the user behavior and the current time, statistics is carried out on the behavior value after the cooling, and finally determined user preference can change along with the interest change of the user, so that the user preference of the user at present can be accurately determined.

S503, determining the user preference of the user according to the statistical result.

Optionally, the present disclosure may obtain the preference degree of the user on each enumerated value by using the statistical result corresponding to each enumerated value. In one example, the present disclosure may directly use the statistical result corresponding to each enumerated value as the preference degree of the user on each enumerated value. In another example, for any enumerated value of all enumerated values, the present disclosure may divide the statistical result corresponding to the enumerated value by the accumulated statistical results corresponding to all enumerated values included in the behavior object attribute element to which the enumerated value belongs, and take the quotient corresponding to the enumerated value as the preference degree of the user on the enumerated value. The accumulated result of the statistical results corresponding to all enumeration values included in the behavior object attribute element to which the enumeration value belongs may be the sum of the behavior values after cooling processing corresponding to all enumeration values included in the behavior object attribute element.

For a continuation example, the present disclosure may calculate the sum s of v1 ', v 2', … …, vm-1 ', and vm', and take s1/s as the user's preference for the enumerated value a and s2/s as the user's preference for the enumerated value b.

Optionally, the present disclosure may obtain one or more preference degrees of the user, and the present disclosure may sort all the obtained preference degrees in a descending order, where the sorting may reflect the user preference order of the user. For example, the present disclosure may determine an enumerated value corresponding to the highest preference degree first, and use the enumerated value corresponding to the highest preference degree as the user preference with the highest preference degree; secondly, the method can determine the enumeration value corresponding to the second highest preference degree, and take the enumeration value corresponding to the second highest preference degree as the preference of the user with the second highest preference degree.

Because the enumeration value under each object attribute element contained in the object attribute information can reflect the target object related to the user behavior more finely, the user preference can be reflected more accurately by counting the value of each behavior accumulated currently based on the enumeration value.

Exemplary devices

Fig. 6 is a schematic structural diagram of an embodiment of the apparatus for determining user preferences according to the present disclosure. The apparatus of this embodiment may be used to implement the method embodiments of the present disclosure described above.

As shown in fig. 6, the apparatus of the present embodiment may include: an acquisition module 600, a prediction module 601, and a determine preferences module 602. Optionally, the apparatus of this embodiment may further include: a training module 603.

The obtaining module 600 is configured to obtain behavior feature information of a user behavior. The behavior feature information may include: a plurality of behavioral characteristic elements.

Optionally, the obtaining module 600 includes: a first sub-module 6001 and a second sub-module 6002. The first sub-module 6001 is configured to obtain incremental service data based on user behavior. The second sub-module 6002 is configured to obtain all behavior feature elements of the user behavior according to the field content of the predetermined field included in the incremental service data obtained by the first sub-module 6001. The behavior characteristic elements comprise: discrete behavior feature elements, and/or continuous behavior feature elements.

The predicting module 601 is configured to predict a behavior value of a user behavior relative to a target behavior according to each behavior feature element included in the behavior feature information acquired by the acquiring module 600.

Optionally, the prediction module 601 may include: a third submodule 6011, a fourth submodule 6012, and a fifth submodule 6013. The third submodule 6011 is configured to determine, according to the behavior type of the user behavior, a value prediction model corresponding to the user behavior; and the user behaviors of different behavior types correspond to different value prediction models. The fourth submodule 6012 is configured to use behavior feature elements included in the behavior feature information of the user behavior as model inputs, provide the model for the value prediction model corresponding to the user behavior, and perform prediction processing on the probability that the user behavior is converted into the target behavior according to the model inputs through the value prediction model corresponding to the user behavior. The fifth submodule 6013 is configured to obtain a behavior value of the user behavior with respect to the target behavior according to the probability value output by the value prediction model.

The preference determining module 602 is configured to determine the user preference of the user according to the behavior object attribute information corresponding to each of the at least one user behavior of the user and the behavior value, relative to the target behavior, of each of the at least one user behavior predicted by the predicting module.

Optionally, the determining the preference module 602 includes: a ninth sub-module 6021, a tenth sub-module 6022, an eleventh sub-module 6023, and a twelfth sub-module 6024. The ninth sub-module 6021 is configured to count the predicted values of the behaviors according to the enumerated values of the behavior object attribute elements corresponding to at least one user behavior of the user. For example, the ninth sub-module 6021 may firstly cool each behavior value predicted by the prediction module 601 according to the occurrence time of each user behavior, and then the ninth sub-module 6021 may count each behavior value after cooling according to the enumerated value under the behavior object attribute element corresponding to each at least one user behavior of the user. The tenth submodule 6022 is configured to determine the user preference of the user according to the result statistically obtained by the ninth submodule 6021. The eleventh submodule 6023 is used for obtaining the behavior identification of at least one user behavior. The twelfth submodule 6024 is configured to obtain, according to the behavior identifier of the at least one user behavior obtained by the eleventh submodule 6023, an enumeration value under each behavior object attribute element corresponding to the behavior identifier of each user behavior.

The training module 603 is configured to train the value prediction model to be trained. The training module 603 may include: a sixth sub-module 6031, a seventh sub-module 6032, and an eighth sub-module 6033. The sixth sub-module 6031 is configured to obtain behavior feature information samples of the user behaviors of the multiple users for each user behavior type according to the stock business data based on the user behaviors. For example, the sixth sub-module 6031 may first obtain, for each user behavior type, a plurality of behavior feature elements of the user behaviors of the plurality of users according to field contents of a predetermined field included in the stock quantity service data, and then, the sixth sub-module 6031 determines, according to the stock quantity service data, whether a target behavior exists in a predetermined time range after each of the user behaviors of the plurality of users occurs; for each user behavior with a target behavior, the sixth sub-module 6031 sets a first label for representing and converting into the target behavior for corresponding user behavior feature information, and obtains a positive behavior feature information sample; for each user behavior without the target behavior, the sixth sub-module 6031 sets a second label for representing the non-target behavior for the corresponding user behavior feature information, and obtains a negative behavior feature information sample. The seventh sub-module 6032 is configured to, for any behavior type, respectively use a plurality of behavior feature elements included in behavior feature information samples of a plurality of user behaviors having the behavior type as model inputs, provide the model with a value prediction model to be trained corresponding to the user behavior having the behavior type, and perform prediction processing on probabilities that the corresponding user behaviors are respectively converted into target behaviors according to the respective model inputs through the value prediction models to be trained. The eighth submodule 6033 is configured to adjust the model parameters of each to-be-trained value prediction model according to the label used for representing whether the user behavior is converted into the target behavior in each behavior feature information sample and the corresponding probability value output by each to-be-trained value prediction model.

The operations specifically executed by the modules and the sub-modules and units included in the modules may be referred to in the description of the method embodiments with reference to fig. 2 to 5, and are not described in detail here.

Exemplary electronic device

An electronic device according to an embodiment of the present disclosure is described below with reference to fig. 7. FIG. 7 shows a block diagram of an electronic device in accordance with an embodiment of the disclosure. As shown in fig. 7, the electronic device 71 includes one or more processors 711 and memory 712.

The processor 711 may be a Central Processing Unit (CPU) or other form of processing unit having the capability to determine user preferences and/or instruction execution capabilities, and may control other components in the electronic device 71 to perform desired functions.

Memory 712 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory, for example, may include: random Access Memory (RAM) and/or cache memory (cache), etc. The nonvolatile memory, for example, may include: read Only Memory (ROM), hard disk, flash memory, and the like. One or more computer program instructions may be stored on the computer-readable storage medium and executed by the processor 711 to implement the methods of determining user preferences for the various embodiments of the present disclosure described above and/or other desired functions. Various contents such as an input signal, a signal component, a noise component, etc. may also be stored in the computer-readable storage medium.

In one example, the electronic device 71 may further include: input devices 713 and output devices 714, among other components, interconnected by a bus system and/or other form of connection mechanism (not shown). The input device 713 may also include, for example, a keyboard, a mouse, and the like. The output device 714 can output various information to the outside. The output devices 714 may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, among others.

Of course, for simplicity, only some of the components of the electronic device 71 relevant to the present disclosure are shown in fig. 7, omitting components such as buses, input/output interfaces, and the like. In addition, the electronic device 71 may include any other suitable components, depending on the particular application.

Exemplary computer program product and computer-readable storage Medium

In addition to the methods and apparatus described above, embodiments of the present disclosure may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the method of determining user preferences according to various embodiments of the present disclosure described in the "exemplary methods" section above of this specification.

The computer program product may write program code for carrying out operations for embodiments of the present disclosure in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. Furthermore, embodiments of the present disclosure may also be a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, cause the processor to perform steps in a method of determining user preferences according to various embodiments of the present disclosure described in the "exemplary methods" section above of this specification.

The computer-readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium may include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

The foregoing describes the general principles of the present disclosure in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present disclosure are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present disclosure. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the disclosure is not intended to be limited to the specific details so described.

In the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts in the embodiments are referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and for the relevant points, reference may be made to the partial description of the method embodiment.

The block diagrams of devices, apparatuses, systems referred to in this disclosure are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, and systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," comprising, "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".

The methods and apparatus of the present disclosure may be implemented in a number of ways. For example, the methods and apparatus of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order for the steps of the method is for illustration only, and the steps of the method of the present disclosure are not limited to the order specifically described above unless specifically stated otherwise. Further, in some embodiments, the present disclosure may also be embodied as programs recorded in a recording medium, the programs including machine-readable instructions for implementing the methods according to the present disclosure. Thus, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.

It is also noted that in the devices, apparatuses, and methods of the present disclosure, each component or step can be decomposed and/or recombined. These decompositions and/or recombinations are to be considered equivalents of the present disclosure.

The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects, and the like, will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the disclosure. Thus, the present disclosure is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the disclosure to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.

Claims

1. A method of determining user preferences, comprising:

acquiring behavior characteristic information of user behavior, wherein the behavior characteristic information comprises: a plurality of behavioral characteristic elements;

predicting the behavior value of the user behavior relative to the target behavior according to each behavior feature element contained in the behavior feature information;

and determining the user preference of the user according to the behavior object attribute information corresponding to the at least one user behavior of the user and the predicted behavior value of the at least one user behavior relative to the target behavior.

2. The method of claim 1, wherein the obtaining behavior feature information of the user behavior comprises:

obtaining incremental business data based on user behaviors;

acquiring a plurality of behavior characteristic elements of the user behavior according to the field content of a preset field contained in the incremental business data;

wherein the behavior feature information of the user behavior comprises: discrete behavior feature elements, and/or continuous behavior feature elements.

3. The method according to claim 1 or 2, wherein the predicting the behavior value of the user behavior relative to the target behavior according to each behavior feature element included in the behavior feature information includes:

determining a value prediction model corresponding to the user behavior according to the behavior type of the user behavior; the user behaviors of different behavior types correspond to different value prediction models;

behavior characteristic elements contained in the behavior characteristic information of the user behaviors are used as model input and provided for a value prediction model corresponding to the user behaviors;

performing prediction processing on the probability of converting the user behavior into the target behavior according to the model input of the value prediction model corresponding to the user behavior;

and acquiring the behavior value of the user behavior relative to the target behavior according to the probability value output by the value prediction model.

4. The method of claim 3, wherein the method further comprises:

respectively acquiring behavior characteristic information samples of the user behaviors of a plurality of users aiming at each user behavior type according to stock business data based on the user behaviors;

aiming at any behavior type, respectively taking a plurality of behavior characteristic elements contained in behavior characteristic information samples of a plurality of user behaviors with the behavior type as model input and providing the model input to a value prediction model to be trained corresponding to the user behaviors with the behavior type;

respectively inputting the value prediction models to be trained according to the respective models, and performing prediction processing on the probability of converting the corresponding user behaviors into target behaviors;

and adjusting the model parameters of the value prediction models to be trained according to the labels which are used for representing whether the user behaviors are converted into the target behaviors in the behavior characteristic information samples and the corresponding probability values output by the value prediction models to be trained.

5. The method of claim 4, wherein the obtaining behavior feature information samples of the user behaviors of the plurality of users for each user behavior type according to the inventory business data based on the user behaviors comprises:

respectively acquiring a plurality of behavior characteristic elements of the user behaviors of a plurality of users aiming at each user behavior type according to the field content of a preset field contained in the stock business data;

judging whether target behaviors exist in the user behaviors of the plurality of users within a preset time range after the user behaviors occur or not according to the stock business data;

for each user behavior with a target behavior, setting a first label for representing and converting the corresponding user behavior characteristic information into the target behavior, and obtaining a positive behavior characteristic information sample;

and for each user behavior without the target behavior, setting a second label for representing the non-converted target behavior for the corresponding user behavior characteristic information, and obtaining a negative behavior characteristic information sample.

6. The method according to any one of claims 1 to 5, wherein the determining the user preference of the user according to behavior object attribute information corresponding to each of at least one user behavior of the user and the predicted behavior value of each of at least one user behavior relative to a target behavior comprises:

counting the predicted behavior values according to enumeration values under the behavior object attribute elements corresponding to at least one user behavior of the user;

and determining the user preference of the user according to the statistical result.

7. The method according to claim 6, wherein the performing statistics on the predicted behavior values according to enumerated values under behavior object attribute elements corresponding to at least one user behavior of the user includes:

cooling the predicted behavior values according to the occurrence time of each user behavior;

and counting the behavior values after the cooling treatment according to the enumeration values under the behavior object attribute elements corresponding to the at least one user behavior of the user.

8. An apparatus for determining user preferences, wherein the apparatus comprises:

an obtaining module, configured to obtain behavior feature information of a user behavior, where the behavior feature information includes: a plurality of behavioral characteristic elements;

the prediction module is used for predicting the behavior value of the user behavior relative to the target behavior according to each behavior feature element contained in the behavior feature information acquired by the acquisition module;

and the preference determining module is used for determining the user preference of the user according to the behavior object attribute information corresponding to the at least one user behavior of the user and the behavior value of the at least one user behavior predicted by the predicting module relative to the target behavior.

9. A computer-readable storage medium, the storage medium storing a computer program for performing the method of any of the preceding claims 1-7.

10. An electronic device, the electronic device comprising:

a processor;

a memory for storing the processor-executable instructions;

the processor is configured to read the executable instructions from the memory and execute the instructions to implement the method of any one of claims 1-7.