User Portrait Generation Method, Device, Electronic Device and Readable Medium

Abstract

An embodiment of the present invention provides a user portrait generation method, apparatus, electronic device, and readable medium. The method includes: obtaining user behavior information that can reflect the job hunting intention of a target user; extracting a recruitment attribute value sequence corresponding to each type of behavior from the user behavior information; the recruitment attribute value sequence includes a number of recruitment attribute values; generating a recruitment attribute value vector for the recruitment attribute values of each type of behavior; the recruitment attribute value vector is used to reflect the association relationship between the recruitment attribute values; for a target type of behavior, determining a portrait label of the target type of behavior according to the recruitment attribute value vector corresponding to the target type of behavior; and determining a portrait label of the target user according to the portrait label of the target type of behavior. The embodiment of the present invention can obtain more accurate portrait labels of the target user.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of Internet technologies, and in particular, to a method for generating a user profile, a user profile device, an electronic device, and a computer-readable medium. Background Art

[0002] A user profile is a series of profile tags related to a service and used to describe user characteristics, which are generated by statistics and algorithm calculations based on information provided by a user and the user's operation records on a service platform. It can be seen that the core work of constructing a user profile is to set profile tags for the user, and the profile tags are highly refined feature identifiers obtained by analyzing information.

[0003] In a recruitment recommendation scenario, in order to provide more accurate personalized recommendation results for a user, profile tags need to be used. Usually, profile tags for a single type of behavior are calculated, and then, through the profile tags of the single type of behavior, the positions that best match the user can be selected from a large number of candidate positions and recommended to the user, reducing the user's ineffective browsing time and improving the user experience.

[0004] However, although the existing calculation scheme for user profiles can calculate profile tags for a single type of behavior, it does not consider the correlation between various recruitment attribute values within this type of behavior, resulting in the inability to associate other recruitment needs of the user based on the profile tags, and thus unable to accurately recommend for the user based on the profile tags. Summary of the Invention

[0005] Embodiments of the present invention provide a method, a device, an electronic device, and a computer-readable storage medium for generating a user profile, so as to solve the problem that profile tags cannot reflect other recruitment needs of the user.

[0006] Embodiments of the present invention disclose a method for generating a user profile, and the method includes:

[0007] Obtain user behavior information that can reflect the job hunting intention of a target user;

[0008] Extract a sequence of recruitment attribute values corresponding to each type of behavior from the user behavior information; the sequence of recruitment attribute values includes several recruitment attribute values;

[0009] Obtain a recruitment attribute value vector of the recruitment attribute values of each type of behavior; the recruitment attribute value vector is used to reflect the correlation between the recruitment attribute values;

[0010] For a target type of behavior, determine a profile tag of the target type of behavior according to the recruitment attribute value vector corresponding to the target type of behavior;

[0011] Determine the portrait tags of the target user according to the portrait tags of the target type behavior, so as to constitute the user portrait of the target user.

[0012] Optionally, the determining the portrait tags of the target type behavior according to the recruitment attribute value vector corresponding to the target type behavior includes:

[0013] Form a recruitment attribute value vector sequence with the recruitment attribute value vectors of the target type behavior;

[0014] Sum the recruitment attribute value vector sequence of the target type behavior to obtain the recruitment attribute direction vector of the target type behavior;

[0015] Obtain the weight value of the target type behavior;

[0016] Weight the recruitment attribute direction vector of the target type behavior with the weight value;

[0017] Calculate the first cosine similarity between the recruitment attribute value vector of the target type behavior and the weighted recruitment attribute direction vector;

[0018] Take the recruitment attribute values corresponding to the recruitment attribute value vectors of the first cosine similarity before the preset number of digits as the portrait tags of the target type behavior.

[0019] Optionally, the obtaining the weight value of the target type behavior includes:

[0020] Form a recruitment attribute value vector sequence with the recruitment attribute value vectors of other type behaviors;

[0021] Calculate the second cosine value similarity between the recruitment attribute direction vector of the target type behavior and the recruitment attribute value vector sequence of the other type behaviors;

[0022] Calculate the mean value of the second cosine value similarities as the weight value of the target type behavior.

[0023] Optionally, the determining the portrait tags of the target user according to the portrait tags of the target type behavior includes:

[0024] Obtain the first cosine similarity of the portrait tags of each target type behavior; the first cosine similarity is the cosine similarity between the recruitment attribute value vectors corresponding to the recruitment attribute values of the portrait tags of each target type behavior and the weighted recruitment attribute direction vector;

[0025] Take the portrait tags of the target type behavior before the preset number of digits of the first cosine similarity as the portrait tags of the target user.

[0026] Optionally, a data logging program is set on the service platform, and the obtaining of the user behavior information capable of reflecting the job hunting intention of the target user includes:

[0027] Obtaining the user behavior information reported through the data logging program in the service platform.

[0028] Optionally, the method further includes:

[0029] Obtaining the basic profile information of the target user; the basic profile information includes biological basic information and job hunting intention information reflecting the characteristics of the target user;

[0030] Constructing the user profile of the target user with the basic profile information and / or the profile tags.

[0031] Optionally, the biological basic information at least includes gender, age, education level, and work experience; the job hunting intention information at least includes expected industry, position type, and position salary; the type of behavior at least includes applying, communicating, and clicking.

[0032] Optionally, the obtaining of the recruitment attribute value vector of each of the type of behavior includes:

[0033] Inputting the recruitment attribute values of each of the type of behavior into a pre-trained recruitment attribute value prediction model to obtain the recruitment attribute vector corresponding to the recruitment attribute value; wherein, the recruitment attribute value prediction model is trained with historical recruitment attribute value sequences of each of the type of behavior, and the historical recruitment attribute value sequences include a number of historical recruitment attribute values.

[0034] An embodiment of the present invention also discloses a user profile generation device, and the device includes:

[0035] A user behavior information acquisition module, configured to acquire user behavior information capable of reflecting the job hunting intention of the target user;

[0036] A recruitment attribute value extraction module, configured to extract a recruitment attribute value sequence corresponding to each type of behavior from the user behavior information; the recruitment attribute value sequence includes a number of recruitment attribute values;

[0037] A recruitment attribute value vector generation module, configured to obtain a recruitment attribute value vector of the recruitment attribute values of each of the type of behavior; the recruitment attribute value vector is used to reflect the association relationship between the recruitment attribute values;

[0038] A type of behavior label determination module, configured to determine the profile label of the target type of behavior according to the recruitment attribute value vector corresponding to the target type of behavior;

[0039] A user label determination module, configured to determine the portrait label of the target user according to the portrait label of the target type behavior.

[0040] Optionally, the type behavior label determination module is configured to form a recruitment attribute value vector sequence from the recruitment attribute value vectors of the target type behavior; sum the recruitment attribute value vector sequence of the target type behavior to obtain a recruitment attribute direction vector of the target type behavior; obtain a weight value of the target type behavior; the weight value of the target type behavior has an association relationship with other type behaviors; weight the recruitment attribute direction vector of the target type behavior by using the weight value; calculate a first cosine similarity between the recruitment attribute value vector of the target type behavior and the weighted recruitment attribute direction vector; take the recruitment attribute value corresponding to the recruitment attribute value vector before a preset number of digits of the first cosine similarity as the portrait label of the target type behavior.

[0041] Optionally, the type behavior label determination module is configured to form a recruitment attribute value vector sequence from the recruitment attribute value vectors of other type behaviors; calculate a second cosine similarity between the recruitment attribute direction vector of the target type behavior and the recruitment attribute value vector sequence of the other type behaviors; calculate an average value between the second cosine similarities as the weight value of the target type behavior.

[0042] Optionally, the user label determination module is configured to obtain a first cosine similarity of the portrait labels of each target type behavior; the first cosine similarity is the cosine similarity between the recruitment attribute value vector corresponding to the recruitment attribute value of the portrait label of each target type behavior and the weighted recruitment attribute direction vector; take the portrait label of the target type behavior before a preset number of digits of the first cosine similarity as the portrait label of the target user.

[0043] Optionally, a data logging program is set on the service platform, and the user behavior information acquisition module is configured to acquire user behavior information reported through the data logging program in the service platform.

[0044] Optionally, the apparatus further includes: a user portrait generation module, configured to acquire basic portrait information of the target user; the basic portrait information includes biological basic information and job hunting intention information reflecting the characteristics of the target user; and form the user portrait of the target user by using the basic portrait information and / or the portrait label of the target user.

[0045] Optionally, the biological basic information may at least include gender, age, educational background, and work experience; the job hunting intention information may at least include the expected industry, position type, and position salary; the type of behavior may at least include submission, communication, and click.

[0046] Optionally, the recruitment attribute value vector generation module is configured to input the recruitment attribute values of the respective types of behavior into a pre-trained recruitment attribute value prediction model to obtain a recruitment attribute vector corresponding to the recruitment attribute value; wherein, the recruitment attribute value prediction model is trained using historical recruitment attribute value sequences of the respective types of behavior, and the historical recruitment attribute value sequences include a number of historical recruitment attribute values.

[0047] An embodiment of the present invention also discloses an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete communication with each other through the communication bus;

[0048] The memory is used to store a computer program;

[0049] When the processor is configured to execute the program stored on the memory, it implements the method as described in the embodiment of the present invention.

[0050] An embodiment of the present invention also discloses one or more computer-readable media, on which instructions are stored, and when executed by one or more processors, cause the processors to execute the method as described in the embodiment of the present invention.

[0051] An embodiment of the present invention also discloses a computer program product, which is stored in a storage medium and is executed by at least one processor to implement the method as described in the embodiment of the present invention.

[0052] The embodiments of the present invention have the following advantages:

[0053] In an embodiment of the present invention, user behavior information that can reflect the job hunting intention of a target user is obtained, and a recruitment attribute value sequence corresponding to each type of behavior is extracted from the user behavior information. Then, a recruitment attribute value vector reflecting the association relationship between recruitment attribute values is generated based on the recruitment attribute value sequence. Next, for a target type of behavior, a portrait label of the target type of behavior is determined according to the recruitment attribute value vector corresponding to the target type of behavior. Finally, based on the portrait label of the target type of behavior, the portrait label of the target user is determined to form the user portrait of the target user. When calculating the portrait label corresponding to a single type of behavior in the embodiment of the present invention, a recruitment attribute value vector reflecting the association relationship between recruitment attribute values is generated, and the portrait label of this type of behavior is determined according to the recruitment attribute value vector of the single type of behavior. Since the association between each recruitment attribute value within a single type of behavior is considered, more accurate portrait labels for each type of behavior can be obtained. Furthermore, based on the portrait labels of each type of behavior, a more accurate portrait label of the target user can be obtained, so that when recommending content for the target user based on the portrait label of the target user, the recommended content is more accurate and rich. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a flowchart of the steps of a method for generating a user portrait provided in an embodiment of the present invention;

[0055] Figure 2 is a flowchart of the steps for determining the portrait label of a target user provided in an embodiment of the present invention;

[0056] Figure 3 is a block diagram of the structure of a user portrait device provided in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0057] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] Currently, the user portrait generation solution is to record the user behavior information of the user on the service platform, extract the corresponding recruitment attribute values in the user behavior information, and decay the recruitment attribute values of each type of behavior according to the corresponding event occurrence time to update the portrait label corresponding to the event occurrence time. For the portrait labels of each type of behavior, a preset fixed weight value is set, and then the portrait labels of each type of behavior are used to obtain the final portrait label based on the preset fixed weight value.

[0059] The defects of the above user portrait generation solution include: 1) When calculating the portrait tags of a single type of behavior, the event occurrence time is used for decay. Although timeliness is considered, the correlation between recruitment attribute values within this type of behavior is not considered; 2) When determining the final portrait tags by integrating the portrait tags of various types of behavior, preset fixed weight values are adopted, without adjustment according to the correlation between various types of behavior of the user, which is not flexible enough. When making recommendations based on the portrait tags, other associated information cannot be further provided for the user. For example, when applying for jobs, more associated job information cannot be provided for the user.

[0060] To address the above problems, an embodiment of the present invention provides a user portrait generation method. When calculating the portrait tags of a single type of behavior, a recruitment attribute value vector reflecting the correlation between recruitment attribute values is used to determine the portrait tags of various types of behavior, considering the correlation between recruitment attribute values within a single type of behavior; in addition, for the recruitment attribute value vector of a certain type of behavior, a weight value reflecting the correlation between this type of behavior and other types of behavior is calculated, making the fusion of portrait tags of various types of behavior more flexible and adjustable according to the correlation between types of behavior.

[0061] Refer to Figure 1 , which shows the step flow chart of a user portrait generation method provided in an embodiment of the present invention, and specifically may include the following steps:

[0062] Step 102, obtain user behavior information that can reflect the job hunting intention of the target user.

[0063] Among them, the service platform may refer to an application program (APP, Application). The target user refers to a user in the service platform with a job hunting intention. For example, a user who has submitted a resume in the service platform, a user who has browsed and clicked on recruitment information in the service platform, or a user who has communicated with a recruitment position publisher in the service platform, etc. Specifically, the application program may include a lifestyle application program, and the lifestyle application program can provide functions such as job hunting, renting a house, and second-hand cars for the target user.

[0064] In an embodiment of the present invention, user behavior information that can reflect the job hunting intention of the user generated during the process of the user using the recruitment function of the service platform is obtained. Exemplarily, when the user uses the recruitment function of the service platform, the generated user behavior information may include [click behavior information of the target user on the recruitment position], [resume submission behavior information for the recruitment position], [communication behavior information with the recruitment position publisher], etc.

[0065] It should be noted that the user behavior information and other privacy information involved in the embodiments of the present invention are all information authorized by the user or authorized by all parties.

[0066] Step 104: Extract the recruitment attribute value sequences corresponding to each type of behavior from the user behavior information; the recruitment attribute value sequences include several recruitment attribute values.

[0067] Among them, the type of behavior can at least include [Delivery], [Communication], [Click], etc.; each type of behavior can include one or more recruitment attributes. For example, the recruitment attributes can at least include [Recruitment Industry], [Position Type], [Position Salary], etc.; each recruitment attribute can include multiple recruitment attribute values, and the recruitment attribute values of multiple recruitment attributes can form a set of recruitment attribute value sequences. Exemplarily, a set of recruitment attribute value sequences can be [Internet, Product Manager, 10K - 20K].

[0068] In the embodiments of the present invention, the recruitment attribute value sequences corresponding to the recruitment attributes of each type of behavior can be extracted from the user behavior information. For example, when the user behavior information is [The target user's click behavior on the product manager position of xx Internet company], then based on this user behavior information, the type of behavior [Click] and the recruitment attribute value sequence [Internet, Product Manager] can be obtained, and the recruitment attribute value sequence includes the recruitment attribute values [Internet] and [Product Manager].

[0069] Step 106: Obtain the recruitment attribute value vectors of the recruitment attribute values of each type of behavior; the recruitment attribute value vectors are used to reflect the association relationship between the recruitment attribute values.

[0070] As an example of the present invention, the recruitment attribute values of each type of behavior are respectively input into the trained recruitment attribute value prediction model, so as to obtain the recruitment attribute value vectors corresponding to each type of behavior. For example, inputting the recruitment attribute value [Product Manager] of the type of behavior [Delivery] into the trained recruitment attribute value prediction model, the recruitment attribute value vector corresponding to the recruitment attribute value [Product Manager] of the type of behavior [Delivery] can be obtained.

[0071] Step 108: For the target type of behavior, determine the portrait label of the target type of behavior according to the recruitment attribute value vector corresponding to the target type of behavior.

[0072] Among them, the target type of behavior refers to the type of behavior for which the corresponding portrait label needs to be calculated. In the embodiments of the present invention, the type of behavior for which the corresponding portrait label needs to be calculated can be all types of behavior or some types of behavior, and no limitation is made thereto.

[0073] In an embodiment of the present invention, for a target type of behavior, portrait tags of the target type of behavior can be determined according to the recruitment attribute value vector corresponding to the target type of behavior. For example, according to the recruitment attribute value vector corresponding to the recruitment attribute value of the target type of behavior [Click] of the target user, the portrait tag corresponding to the target type of behavior [Click] of the target user can be determined as the recruitment attribute value [Internet], or, according to the recruitment attribute value vector corresponding to the recruitment attribute value of the target type of behavior [Delivery] of the target user, the portrait tag corresponding to the target type of behavior [Delivery] of the target user can be determined as the recruitment attribute value [Product Manager], or, according to the recruitment attribute value vector corresponding to the recruitment attribute value of the target type of behavior [Communication] of the target user, the portrait tag corresponding to the target type of behavior [Communication] of the target user can be determined as the recruitment attribute value [15K - 20K].

[0074] Step 110: Determine the portrait tag of the target user according to the portrait tag of the target type of behavior, so as to form the user portrait of the target user.

[0075] In an embodiment of the present invention, after obtaining the portrait tag of the target type of behavior, the portrait tag that can most express the characteristics of the target user can be further selected from the portrait tags of the target type of behavior as the portrait tag of the target user, so as to form the user portrait of the target user.

[0076] Exemplarily, assuming that the portrait tags of the target type of behavior of the target user are [Internet], [Product Manager], [15K - 20K], etc., [Internet] and [Product Manager] among them can be used as the portrait tags of the target user.

[0077] It can be understood that determining the portrait tag of the target type of behavior based on the recruitment attribute value vector of the target user that can reflect the correlation relationship between recruitment attribute values, and then determining the portrait tag of the target user based on the portrait tag of the target type of behavior can fully explore the relevant information of the target user, which is beneficial to improving the recommendation effect when making recommendations based on the portrait tag of the target user.

[0078] In the above user portrait generation method, user behavior information that can reflect the job hunting intention of the target user is obtained, and recruitment attribute value sequences corresponding to various types of behaviors are extracted from the user behavior information, so as to generate a recruitment attribute value vector reflecting the association relationship between recruitment attribute values according to the recruitment attribute value sequences. Then, for the target type of behavior, according to the recruitment attribute value vector corresponding to the target type of behavior, the portrait label of the target type of behavior is determined, and then, according to the portrait label of the target type of behavior, the portrait label of the target user is finally determined to constitute the user portrait of the target user. In the embodiment of the present invention, when calculating the portrait label corresponding to a single type of behavior, a recruitment attribute value vector reflecting the association relationship between recruitment attribute values is generated, and the portrait label of this type of behavior is determined according to the recruitment attribute value vector of the single type of behavior. Since the relevance between each recruitment attribute value within a single type of behavior is considered, more accurate portrait labels of various types of behaviors can be obtained. Furthermore, based on the portrait labels of various types of behaviors, a more accurate portrait label of the target user can be obtained, so that when recommending content for the target user based on the portrait label of the target user, the recommended content is more accurate and rich.

[0079] Based on the above embodiment, a variant embodiment of the above embodiment is proposed. Here, it should be noted that for the sake of brief description, only the differences from the above embodiment are described in the variant embodiment.

[0080] In an exemplary embodiment, the step 108 of determining the portrait label of the target type of behavior according to the recruitment attribute value vector corresponding to the target type of behavior may include the following steps:

[0081] Form a recruitment attribute value vector sequence composed of the recruitment attribute value vectors of the target type of behavior;

[0082] Sum the recruitment attribute value vector sequence of the target type of behavior to obtain a recruitment attribute direction vector of the target type of behavior; obtain the weight value of the target type of behavior; the weight value of the target type of behavior has an association relationship with the weight values of other types of behaviors;

[0083] Weight the recruitment attribute direction vector of the target type of behavior by using the weight value;

[0084] Calculate the first cosine similarity between the recruitment attribute value vector of the target type of behavior and the weighted recruitment attribute direction vector;

[0085] Take the recruitment attribute values corresponding to the recruitment attribute value vector before the preset number of digits of the first cosine similarity as the portrait label of the target type of behavior.

[0086] Among them, cosine similarity, also known as cosine likeness, is used to evaluate the similarity between vectors by calculating the cosine value of the angle between two vectors.

[0087] In a specific implementation, user behavior information generated by a target user when using a service platform is obtained, recruitment attribute values corresponding to each type of behavior are extracted from the user behavior information, and then, the recruitment attribute values of each type of behavior can be input into a pre-trained recruitment attribute value prediction model to obtain a recruitment attribute value vector that can reflect the correlation relationship between recruitment attribute values.

[0088] For a target type of behavior, a recruitment attribute value vector of the recruitment attribute value of the target type of behavior can be obtained, the recruitment attribute value vectors are combined into a recruitment attribute value vector sequence, and the sum of the recruitment attribute value vector sequence is calculated to obtain a recruitment attribute direction vector of the target type of behavior. Exemplarily, assume that the target type of behavior is [submission], and there are 3 recruitment attribute value vectors corresponding to the target type of behavior [submission]. Then, the 3 recruitment attribute value vectors can be combined to obtain a recruitment attribute value vector sequence, and then the sum of the recruitment attribute value vector sequence is calculated to obtain a recruitment attribute direction vector corresponding to the target type of behavior [submission] of the target user.

[0089] In an embodiment of the present invention, a weight value of a target type of behavior is obtained, where the weight value is used to represent the importance of constructing a user portrait. Then, the weight value is used to weight the recruitment attribute direction vector of the target type of behavior, and then the cosine similarity between the recruitment attribute value vector of the target type of behavior and the weighted recruitment attribute direction vector is calculated. Then, the recruitment attribute value corresponding to the recruitment attribute value vector before a preset number of digits of the first cosine similarity is taken as the portrait label of the target type of behavior.

[0090] Exemplarily, assume that the weight value corresponding to the target type of behavior [submission] is 0.3, and the preset number of digits N can be 2. The weight value is used to weight the recruitment attribute direction vector of the target type of behavior [submission], and then the cosine similarity with the recruitment attribute value vector of the target type of behavior is calculated. If the recruitment attribute values in the first 2 digits of the cosine similarity are [product manager], [10K - 20K], then [product manager], [10K - 20K] can be used as the portrait labels of the target type of behavior [submission].

[0091] In the above exemplary embodiments, when calculating the portrait tags of a single type of behavior, the cosine similarity between the recruitment attribute value vectors of each recruitment attribute value and the recruitment attribute direction vector of this type of behavior weighted by the weight value is calculated to determine the recruitment attribute value as the portrait tag of this type of behavior based on the cosine similarity. Among them, the weight value of a single type of behavior has an association relationship with other types of behaviors, so that the integration with other types of behaviors is more flexible when calculating the portrait tags of a single type of behavior and can be flexibly adjusted continuously with the user behavior information.

[0092] In one exemplary embodiment, the obtaining of the weight value of the target type of behavior may include the following steps:

[0093] Form a recruitment attribute value vector sequence with the recruitment attribute value vectors of other types of behaviors;

[0094] Calculate the second cosine value similarity between the recruitment attribute direction vector of the target type of behavior and the recruitment attribute value vector sequence of the other types of behaviors;

[0095] Calculate the mean value among the second cosine value similarities as the weight value of the target type of behavior.

[0096] In the embodiments of the present invention, for each type of behavior of the target user, the corresponding recruitment attribute value vector can be obtained respectively, and then the recruitment attribute value vectors are formed into a recruitment attribute value vector sequence, and the recruitment attribute value vectors of each type of behavior are summed to obtain the recruitment attribute direction vector of each type of behavior.

[0097] For the target type of behavior, calculate the cosine similarity between the recruitment attribute direction vector of the target type of behavior and the recruitment attribute value vector sequence of other types of behaviors. For example, assume that the behavior types of the target user include S1, S2, S3, and S4. If the target type of behavior is S1, then it is necessary to calculate the cosine similarity between the recruitment attribute direction vector of S1 and the recruitment attribute value vector sequences of S2, S3, and S4 respectively. Then, take the mean value of the cosine similarities as the weight value of the target type of behavior.

[0098] As an optional example, the mathematical expression of the weight value corresponding to the target type of behavior is as follows:

[0099]

[0100]

[0101] Where proj(A', D A ) is the recruitment attribute value vector sequence of the A' type of behavior In the recruitment attribute direction vector D of the A type of behavior ACosine value projection, cos(D A , V i ) is the cosine value projection, is the sequence length of, A s is all types of behaviors of the target user, is A s the quantity of, w(A) is the weight value corresponding to the A-type behavior of the target user.

[0102] In the above exemplary embodiment, the mean value of the cosine similarity between the recruitment attribute direction vector of the target type behavior and the recruitment attribute value vector sequence of other type behaviors is used as the weight value of the target type behavior. Therefore, the weight value reflects the association relationship between the target type behavior and other type behaviors. Furthermore, the portrait label of the target type behavior determined based on the weight value has fusion with the portrait labels of other type behaviors, and moreover, the weight value can also be flexibly adjusted according to the user behavior information.

[0103] In an exemplary embodiment, the step 110 of determining the portrait label of the target user according to the portrait label of the target type behavior may include the following steps:

[0104] Obtain the first cosine similarity of the portrait labels of each of the target type behaviors; the first cosine similarity is the cosine similarity between the recruitment attribute value vector of the recruitment attribute value corresponding to the portrait label of each of the target type behaviors and the weighted recruitment attribute direction vector;

[0105] Take the portrait labels of the target type behaviors before the preset number of digits of the first cosine similarity as the portrait label of the target user.

[0106] In the embodiment of the present invention, after obtaining the portrait labels of multiple target type behaviors, the cosine similarity corresponding to the portrait label of the target type behavior can be obtained, that is, determine the cosine similarity between the recruitment attribute value vector corresponding to the recruitment attribute value determined as the portrait label of the target type behavior and the weighted recruitment attribute direction vector. Then, take the portrait labels of the target type behaviors before the preset number of digits of the cosine similarity as the portrait label of the target user.

[0107] Exemplarily, assume that the cosine similarity C1 of portrait label 1 of the target type behavior [Delivery] of the target user, the cosine similarity C2 of portrait label 2 of the target type behavior [Click], and the cosine similarity C3 of portrait label 3 of the target type behavior [Communication] have been calculated. Assume that the preset number of digits is 1, and moreover, the cosine similarity C2 of portrait label 2 is the largest. Then, the portrait label can be used as the portrait label of the target user.

[0108] In the above exemplary embodiments, after obtaining the portrait tags of the target user's target type behavior, the final portrait tags of the target user can be further determined from the portrait tags of the target type behavior based on the cosine similarity corresponding to the portrait tags, so that the user portrait of the target user is more accurate.

[0109] In one exemplary embodiment, a data tracking program is set on the service platform. The step 102, obtaining user behavior information that can reflect the job hunting intention of the target user, may include the following steps:

[0110] Obtain the user behavior information reported by the data tracking program in the service platform.

[0111] Specifically, a data tracking program can be set for the service platform. Setting the data tracking program means inserting some probes into the service platform on the basis of ensuring the integrity of the original logic of the service platform. Among them, the probe is essentially a code segment for information collection, which can be an assignment statement or a function call for collecting coverage information, etc. Then, through the data tracking program in the service platform, the operation behaviors of the target user on the service platform can be collected to obtain user behavior information that can reflect the user's job hunting intention, such as [click behavior information of the target user on the recruitment position], [resume submission behavior information for the recruitment position], [communication behavior information with the recruitment position publisher], etc. Then, recruitment attribute values, such as recruitment industry, position type, and position salary, etc., can be extracted from the user behavior information to facilitate the subsequent generation of the portrait tags of the target user.

[0112] In one exemplary embodiment, the method may further include the following steps:

[0113] Obtain the basic portrait information of the target user; the basic portrait information includes biological basic information and job hunting intention information reflecting the characteristics of the target user;

[0114] Construct the user portrait of the target user with the basic portrait information and / or the portrait tags of the target user.

[0115] Among them, the biological basic information may at least include the gender, age, education background, and work experience of the target user, etc.; the job hunting intention information may at least include the expected industry, position type, and position salary, etc. Optionally, the biological basic information and the job hunting intention information can be obtained from the resume of the target user or from the registration information in the service platform. The embodiments of the present invention do not need to be limited thereto.

[0116] In an embodiment of the present invention, the basic portrait information of the target user, that is, the biological basic information and the job hunting intention information, is obtained, and then the biological basic information and the job hunting intention information are combined with the portrait tags of the target user to form the user portrait of the target user. Subsequently, based on the user portrait, information such as recruitment information can be recommended for the target user.

[0117] Exemplarily, assume that the portrait tags in the user portrait of the target user include [Product Manager], and the basic portrait information includes education background [Bachelor's degree] and work experience [Five years]. If a recruitment information is released by a recruitment position publisher: recruiting a product manager with a bachelor's degree and more than three years of work experience, then this recruitment information can be accurately recommended to the target user.

[0118] In the above exemplary embodiment, the basic portrait information of the target user is combined with the portrait tags of the target user determined based on the recruitment attribute value vector that can reflect the association relationship between recruitment attribute values and the weight value that can reflect the association relationship between type behaviors as the user portrait of the target user, so that when information is recommended based on the user portrait, the recommended content is more personalized.

[0119] In an exemplary embodiment, the step of obtaining the recruitment attribute value vectors of the recruitment attribute values of each of the type behaviors in 106 may include the following steps:

[0120] Input the recruitment attribute values of each of the type behaviors into a pre-trained recruitment attribute value prediction model to obtain the recruitment attribute vectors corresponding to the recruitment attribute values; wherein, the recruitment attribute value prediction model is trained using the historical recruitment attribute value sequences of each of the type behaviors, and the historical recruitment attribute value sequences include several historical recruitment attribute values.

[0121] In a specific implementation, the historical user behavior information that can reflect the job hunting intention of the user on the service platform is recorded. For the recruitment attribute of each type behavior, the corresponding recruitment attribute value is extracted from the historical user behavior information respectively, and the extracted recruitment attribute values are sorted according to time in user groups to obtain multiple groups of recruitment attribute value sequences. Then, using the multiple groups of recruitment attribute value sequences as the data training set, a recruitment attribute value prediction model is trained to obtain a trained recruitment attribute value prediction model. Based on the trained recruitment attribute value prediction model, a recruitment attribute value vector that can reflect the association relationship between recruitment attribute values can be output.

[0122] In an embodiment of the present invention, the recruitment attribute values of each type behavior are respectively input into the trained recruitment attribute value prediction model, so as to obtain the recruitment attribute value vectors corresponding to each type behavior. For example, inputting the recruitment attribute value sequence of the type behavior [Delivery] into the trained recruitment attribute value prediction model can obtain the recruitment attribute value vector corresponding to the recruitment attribute value of the type behavior [Delivery].

[0123] To enable those skilled in the art to better understand the technical solutions of the embodiments of the present invention, the following uses a specific example to illustrate the generation process of the portrait tags of the target user. Refer to Figure 2 , and the specific steps include:

[0124] Obtain the user behavior information of the job hunting intention of the target user, and extract the behavior sequences (recruitment attribute value sequences) corresponding to the three types of behaviors A1, A2, and A3 from the user behavior information, namely the A1 behavior sequence, the A2 behavior sequence, and the A3 behavior sequence; based on the recruitment attribute value sequence, obtain the direction sequences (recruitment attribute direction vectors) corresponding to A1, A2, and A3, namely the D1 direction vector, the D2 direction vector, and the D3 direction vector; for the D1 direction vector, the D2 direction vector, and the D3 direction vector, respectively use the corresponding weight values for weighting to obtain the corresponding weighted direction vector D user， Then, all the recruitment attribute values of the recruitment attributes of each type of behavior are combined with the corresponding direction vector D user Calculate the cosine similarity, and then take the top N recruitment attribute values with the largest cosine similarity as the portrait tags of this type of behavior. Finally, combine the portrait tags of each type of behavior, and take the top N portrait tags of the type of behavior with the largest cosine similarity as the portrait tags of the target user, where N is a positive integer.

[0125] In summary, in the embodiments of the present invention, when calculating the portrait tags of the target type behavior of the target user, the correlation relationship between the recruitment attribute values within the target type behavior is considered by using the recruitment attribute value vector, and the correlation relationship between the target type behavior and other types of behaviors is considered by using the weight value, so that the portrait tags of each target type, that is, have a relationship association with the recruitment attribute values within the type behavior, and have a relationship with other types of behaviors, and then the portrait tags of the target type behavior are combined to obtain the portrait tags of the target user, which is more accurate and can better perform information recommendation.

[0126] It should be noted that for the method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the embodiments of the present invention are not limited by the described action sequence, because according to the embodiments of the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0127] Refer to Figure 3 , which shows a structural block diagram of a user portrait device provided in the embodiments of the present invention. The device may specifically include the following modules:

[0128] The user behavior information acquisition module 302 is used to acquire user behavior information that can reflect the job hunting intention of the target user;

[0129] The recruitment attribute value extraction module 304 is used to extract the recruitment attribute value sequences corresponding to each type of behavior from the user behavior information; several recruitment attribute values are included in the recruitment attribute value sequence;

[0130] The recruitment attribute value vector generation module 306 is used to obtain the recruitment attribute value vectors of the recruitment attribute values of each type of behavior; the recruitment attribute value vector is used to reflect the association relationship between the recruitment attribute values;

[0131] The type behavior label determination module 308 is used to determine the portrait label of the target type behavior according to the recruitment attribute value vector corresponding to the target type behavior;

[0132] The user label determination module 310 is used to determine the portrait label of the target user according to the portrait label of the target type behavior.

[0133] In an exemplary embodiment, the type behavior label determination module 308 is used to form a recruitment attribute value vector sequence from the recruitment attribute value vectors of the target type behavior; sum the recruitment attribute value vector sequence of the target type behavior to obtain the recruitment attribute direction vector of the target type behavior; obtain the weight value of the target type behavior; the weight value of the target type behavior is associated with other types of behaviors; weight the recruitment attribute direction vector of the target type behavior with the weight value; calculate the first cosine similarity between the recruitment attribute value vector of the target type behavior and the weighted recruitment attribute direction vector; take the recruitment attribute values corresponding to the recruitment attribute value vectors before the preset number of digits of the first cosine similarity as the portrait label of the target type behavior.

[0134] In an exemplary embodiment, the type behavior label determination module 308 is used to form a recruitment attribute value vector sequence from the recruitment attribute value vectors of other types of behaviors; calculate the second cosine similarity between the recruitment attribute direction vector of the target type behavior and the recruitment attribute value vector sequence of other types of behaviors; calculate the mean value between the second cosine similarities as the weight value of the target type behavior.

[0135] In an exemplary embodiment, the user tag determination module 310 is configured to obtain the first cosine similarity of the portrait tags of each of the target type behaviors; the first cosine similarity is the cosine similarity between the recruitment attribute value vector of the recruitment attribute values corresponding to the portrait tags of each of the target type behaviors and the weighted recruitment attribute direction vector; the portrait tags of the target type behaviors before a preset number of digits of the first cosine similarity are taken as the portrait tags of the target user.

[0136] In an exemplary embodiment, a data logging program is set on the service platform, and the user behavior information acquisition module 302 is configured to obtain user behavior information reported through the data logging program in the service platform.

[0137] In an exemplary embodiment, the device further includes: a user portrait construction module, configured to obtain the basic portrait information of the target user; the basic portrait information includes biological basic information and job hunting intention information reflecting the characteristics of the target user; the basic portrait information, the portrait tags and / or the portrait tags of the target user are used to construct the user portrait of the target user.

[0138] In an exemplary embodiment, the biological basic information at least includes gender, age, education level and work experience; the job hunting intention information at least includes expected industry, position type and position salary; the type behaviors at least include submission, communication, click.

[0139] In an exemplary embodiment, the recruitment attribute value vector generation module 306 is configured to input the recruitment attribute values of each of the type behaviors into a pre-trained recruitment attribute value prediction model to obtain a recruitment attribute vector corresponding to the recruitment attribute value; wherein, the recruitment attribute value prediction model is trained by using historical recruitment attribute value sequences of each of the type behaviors, and the historical recruitment attribute value sequences include a plurality of historical recruitment attribute values.

[0140] In summary, in the embodiments of the present invention, user behavior information capable of reflecting the job hunting intention of the target user is obtained, and the recruitment attribute value sequences corresponding to various types of behaviors are extracted from the user behavior information, so as to generate a recruitment attribute value vector reflecting the association relationship between the recruitment attribute values according to the recruitment attribute value sequences. Then, for the target type of behavior, according to the recruitment attribute value vector corresponding to the target type of behavior, the portrait label of the target type of behavior is determined, and then, according to the portrait label of the target type of behavior, the portrait label of the target user is finally determined to constitute the user portrait of the target user. When calculating the portrait label corresponding to a single type of behavior in the embodiments of the present invention, a recruitment attribute value vector reflecting the association relationship between the recruitment attribute values is generated, and the portrait label of this type of behavior is determined according to the recruitment attribute value vector of the single type of behavior. Since the relevance between each recruitment attribute value within a single type of behavior is considered, more accurate portrait labels of various types of behaviors can be obtained. Furthermore, based on the portrait labels of various types of behaviors, a more accurate portrait label of the target user can be obtained, so that when recommending content for the target user based on the portrait label of the target user, the recommended content is more accurate and rich.

[0141] For the apparatus embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, refer to the partial description of the method embodiments.

[0142] Preferably, the embodiments of the present invention further provide an electronic device, including: a processor, a memory, and a computer program stored on the memory and executable on the processor. When the computer program is executed by the processor, it implements each process of the above-mentioned method embodiment for generating a user portrait and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0143] The embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by the processor, it implements each process of the above-mentioned method embodiment for generating a user portrait and can achieve the same technical effect. To avoid repetition, it will not be elaborated here. Among them, the computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc.

[0144] The embodiments of the present invention provide a computer program product, which is stored in a storage medium. The program product is executed by at least one processor to implement each process of the above-mentioned method embodiment for generating a user portrait and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0145] It should be noted that in this text, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or device including a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "including one..." does not exclude the existence of additional identical elements in the process, method, article or device including such element.

[0146] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described method of the embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0147] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the spirit and scope protected by the claims of the present invention, and all of them belong to the protection scope of the present invention.

[0148] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in connection with the embodiments disclosed in the embodiments of the present invention can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0149] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be repeated here.

[0150] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections between each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical or other forms.

[0151] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0152] In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, can also exist separately physically for each unit, or two or more units can be integrated in one unit.

[0153] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs that can store program codes.

[0154] As described above, the above are only specific implementation manners of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for generating a user profile, characterized in that, Including: Obtaining user behavior information that can reflect the job hunting intention of the target user; Extracting a recruitment attribute value sequence corresponding to each type of behavior from the user behavior information; The recruitment attribute value sequence includes several recruitment attribute values; Obtaining a recruitment attribute value vector of the recruitment attribute values of each of the types of behavior; the recruitment attribute value vector is used to reflect the association relationship between the recruitment attribute values; For a target type of behavior, determining a portrait label of the target type of behavior according to the recruitment attribute value vector corresponding to the target type of behavior; Determining a portrait label of the target user according to the portrait label of the target type of behavior to form a user portrait of the target user; Among them, the determining the portrait label of the target type of behavior according to the recruitment attribute value vector corresponding to the target type of behavior includes: Forming a recruitment attribute value vector sequence from the recruitment attribute value vectors of the target type of behavior; Summing the recruitment attribute value vector sequence of the target type of behavior to obtain a recruitment attribute direction vector of the target type of behavior; Obtaining a weight value of the target type of behavior; the weight value of the target type of behavior has an association relationship with other types of behavior; Weighting the recruitment attribute direction vector of the target type of behavior by using the weight value; Calculating a first cosine similarity between the recruitment attribute value vector of the target type of behavior and the weighted recruitment attribute direction vector; Taking the recruitment attribute values corresponding to the recruitment attribute value vectors before a preset number of digits of the first cosine similarity as the portrait label of the target type of behavior.

2. The method according to claim 1, characterized in that, The obtaining the weight value of the target type of behavior includes: Forming a recruitment attribute value vector sequence from the recruitment attribute value vectors of other types of behavior; Calculating a second cosine value similarity between the recruitment attribute direction vector of the target type of behavior and the recruitment attribute value vector sequence of the other types of behavior; Calculating the mean value between the second cosine value similarities as the weight value of the target type of behavior.

3. The method according to claim 1, wherein The determining the portrait label of the target user according to the portrait label of the target type of behavior includes: Obtaining a first cosine similarity of the portrait labels of each of the target types of behavior; the first cosine similarity is the cosine similarity between the recruitment attribute value vectors of the recruitment attribute values corresponding to the portrait labels of each of the target types of behavior and the weighted recruitment attribute direction vector; Taking the portrait labels of the target types of behavior before a preset number of digits of the first cosine similarity as the portrait label of the target user.

4. The method according to claim 1, wherein There is a data logging program set on the service platform, and the obtaining user behavior information that can reflect the job hunting intention of the target user includes: Obtaining user behavior information reported through the data logging program in the service platform.

5. The method according to claim 1, wherein The method further includes: Obtaining basic portrait information of the target user; the basic portrait information includes biological basic information and job hunting intention information reflecting the characteristics of the target user; Constituting the user portrait of the target user with the basic portrait information and / or the portrait label of the target user.

6. The method according to claim 5, wherein the biological basic information at least includes gender, age, education level, and work experience; the job hunting intention information at least includes the expected industry, position type, and position salary; the type of behavior at least includes submission, communication, and click.

7. The method according to claim 1, wherein The recruitment attribute value vector for obtaining the recruitment attribute values of each of the type of behavior includes: inputting the recruitment attribute values of each of the type of behavior into a pre-trained recruitment attribute value prediction model to obtain a recruitment attribute vector corresponding to the recruitment attribute value; wherein, the recruitment attribute value prediction model is trained using historical recruitment attribute value sequences of each of the type of behavior, and the historical recruitment attribute value sequences include a number of historical recruitment attribute values.

8. A user portrait generation device, characterized in that, including: a user behavior information acquisition module, configured to acquire user behavior information that can reflect the job hunting intention of a target user; a recruitment attribute value extraction module, configured to extract a recruitment attribute value sequence corresponding to each type of behavior from the user behavior information; the recruitment attribute value sequence includes a number of recruitment attribute values; a recruitment attribute value vector generation module, configured to obtain a recruitment attribute value vector of the recruitment attribute values of each of the type of behavior; the recruitment attribute value vector is used to reflect the correlation relationship between the recruitment attribute values; a type of behavior label determination module, configured to determine a portrait label of the target type of behavior according to the recruitment attribute value vector corresponding to the target type of behavior; a user label determination module, configured to determine a portrait label of the target user according to the portrait label of the target type of behavior; wherein, the type of behavior label determination module is configured to form a recruitment attribute value vector sequence from the recruitment attribute value vectors of the target type of behavior; sum the recruitment attribute value vector sequence of the target type of behavior to obtain a recruitment attribute direction vector of the target type of behavior; obtain a weight value of the target type of behavior; the weight value of the target type of behavior has a correlation relationship with other types of behavior; weight the recruitment attribute direction vector of the target type of behavior using the weight value; calculate a first cosine similarity between the recruitment attribute value vector of the target type of behavior and the weighted recruitment attribute direction vector; take the recruitment attribute value corresponding to the recruitment attribute value vector before a preset number of digits of the first cosine similarity as the portrait label of the target type of behavior.

9. An electronic device, characterized in that, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete communication with each other through the communication bus; the memory is used to store a computer program; when the processor is configured to execute the program stored on the memory, the method according to any one of claims 1-7 is implemented.

10. One or more computer-readable media, on which instructions are stored, and when executed by one or more processors, cause the processors to execute the method according to any one of claims 1-7.

Images