Model training method, recruitment information recommendation method, equipment and storage medium

Abstract

The invention discloses a model training method, a recruitment information recommendation method, equipment and a storage medium. The method comprises the following steps: acquiring a first sample set in a search scene and a second sample set in a recommendation scene; training the original model by using the first sample set and the second sample set in sequence to obtain a target recommendation model; and training the original model by using the second sample set and the first sample set in sequence to obtain a target search model. According to the method provided by the invention, the samples of other scenes are utilized to perform preliminary model training, and then the samples of the scene are utilized to perform fine tuning and optimization training on the model, so that the number of training samples in a certain scene can be increased, and the model of the scene can be ensured to have relatively strong relevance with other scenes; the consistency and the strong relevance of the recruitment information of the recruitment results recommended to the user in different scenes are realized.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a model training method, a recruitment information recommendation method, a device and a storage medium. Background Art

[0002] As Internet technology matures, online recruitment has become one of the main channels for job seekers and recruiters. Online recruitment mainly includes recommendation scenarios and search scenarios. The recommendation scenario recommends recent recruitment information that users are interested in, while the search scenario is mainly used to recommend recruitment information related to the recruitment search terms provided by users.

[0003] At present, for online recruitment systems, the recommendation model corresponding to the recommendation scenario and the search model corresponding to the search scenario are trained and optimized separately during the training or optimization process due to the differences in some features between the two scenarios. The models between the two scenarios are independent of each other from training and optimization to final use.

[0004] However, this will result in a more complete online recruitment system, which requires a large number of samples when training the model corresponding to each scenario. Moreover, the models of different scenarios obtained after training may obtain different results when they have the same features. Summary of the invention

[0005] The present invention provides a model training method, a recruitment information recommendation method, a device and a storage medium to solve the problem that the models of different scenarios obtained after training will obtain recommendation results with great differences when they have the same input features. It can not only increase the number of training samples in a certain scenario, but also ensure that the model of this scenario has a strong correlation with other scenarios, so as to achieve the consistency and strong correlation of the recruitment information recommended to users in different scenarios.

[0006] According to one aspect of the present invention, a model training method is provided, the method comprising:

[0007] Obtaining a first sample set in a search scenario and a second sample set in a recommendation scenario;

[0008] The first sample set and the second sample set are used to train the original model in sequence to obtain a target recommendation model; and the second sample set and the first sample set are used to train the original model in sequence to obtain a target search model;

[0009] Among them, the target recommendation model is used to actively recommend recruitment information to users, and the target search model is used to recommend recruitment information to users based on the target search terms entered by users.

[0010] According to another aspect of the present invention, a model training device is provided, the device comprising:

[0011] An acquisition module, used to acquire a first sample set in a search scenario and a second sample set in a recommendation scenario;

[0012] A training module, used to train the original model using the first sample set and the second sample set in sequence to obtain a target recommendation model; and to train the original model using the second sample set and the first sample set in sequence to obtain a target search model;

[0013] Among them, the target recommendation model is used to actively recommend recruitment information to users, and the target search model is used to recommend recruitment information to users based on the target search terms entered by users.

[0014] According to another aspect of the present invention, a recruitment information recommendation method is provided, the method comprising:

[0015] Determine the user's target operation, and obtain a target model according to the target operation, wherein the target model is a target recommendation model or a target search model trained by the model training method of any embodiment of the present invention;

[0016] Based on the target operation and target model, job postings are identified and recommended to users.

[0017] According to another aspect of the present invention, a recruitment information recommendation device is provided, the device comprising:

[0018] A determination module, used to determine the user's target operation and obtain a target model according to the target operation, wherein the target model is a target recommendation model or a target search model trained by the model training method according to any embodiment of the present invention;

[0019] The recommendation module is used to determine and recommend job information to users based on the target operation and target model.

[0020] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0021] at least one processor; and

[0022] a memory communicatively connected to at least one processor; wherein,

[0023] The memory stores a computer program that can be executed by at least one processor, and the computer program is executed by at least one processor so that the at least one processor can execute the model training method and recruitment information recommendation method of any embodiment of the present invention.

[0024] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions, and the computer instructions are used to enable a processor to implement the model training method and recruitment information recommendation method of any embodiment of the present invention when executed.

[0025] The model training method provided by the embodiment of the present invention obtains a first sample set in a search scenario and a second sample set in a recommendation scenario; the first sample set and the second sample set are used in sequence to train the original model to obtain a target recommendation model; and the second sample set and the first sample set are used in sequence to train the original model to obtain a target search model. In the above technical scheme, the original model is first trained with the first sample set in the search scenario, and then the model is optimized with the second sample set in the recommendation scenario to obtain the final optimized target recommendation model, which solves the problem that the current training process of the recommendation model has little correlation with the search scenario, realizes the preliminary training of the model based on samples of other scenarios, and then fine-tunes and optimizes the model based on samples of this scenario. In addition, the original model is first trained with the second sample set, and then the model is optimized with the first sample set, and finally the target search model is obtained, which solves the problem that the current training of the search model has little correlation with the recommendation scenario, realizes the training of the initial model based on samples of other scenarios, and then fine-tunes and optimizes the initial model based on samples of this scenario. Furthermore, since the above scheme implements that when training the model corresponding to a certain scenario, samples from other scenarios are first used for preliminary model training, and then the samples of this scenario are used to fine-tune and optimize the model training, and finally the model for this scenario is obtained, therefore, the above method not only enables the model of a certain scenario to use samples from other scenarios for more and more sufficient training, thereby increasing the number of samples, but also ensures that the model of this scenario has a strong correlation with other scenarios, thereby ensuring that when users use the online recruitment system, no matter what application scenario they are in, when they input the same or similar specific features, the recruitment results recommended to users can be guaranteed to be consistent and strongly correlated with recruitment information in different scenarios.

[0026] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 A flow chart of a model training method provided by an embodiment of the present invention;

[0029] Figure 2 A flowchart of a recruitment information recommendation method provided by an embodiment of the present invention;

[0030] Figure 3 A schematic diagram of the structure of a model training device provided by an embodiment of the present invention;

[0031] Figure 4 A schematic diagram of the structure of a recruitment information recommendation device provided by an embodiment of the present invention;

[0032] Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] At present, there are two application scenarios in online recruitment systems, namely search scenarios and recommendation scenarios. In the search scenario, after the user enters the search term, the online recruitment system determines the recruitment information corresponding to the search term based on the search term entered by the user, while the recommendation scenario is only the recruitment information automatically recommended to the user based on the user's recent needs. Therefore, compared with the recommendation scenario, the search scenario has the additional feature of "search term".

[0036] For example, in a modeling scheme for a search scenario, the keywords of the search scenario are "search terms", "position-related features" and "resume-related features". In a modeling scheme for a recommendation scenario, the keywords of the recommendation scenario are "position-related features" and "resume-related features". Among them, "search terms" can be search keywords entered by users, "position-related features" can be optional items pre-set by users, such as "education requirements", "age requirements" and "position requirements" and other optional items, and "resume-related features" are resume information in the resume text uploaded by job seekers to the online recruitment system.

[0037] In summary, due to the differences in the “search term” features between the search scenario and the recommendation scenario in the online recruitment system, when training models for the search scenario and the recommendation scenario, the models for the two scenarios are trained separately.

[0038] In order to solve the above problems, the embodiment of the present invention provides the following method:

[0039] Figure 1 This is a flow chart of a model training method provided by an embodiment of the present invention. This embodiment is applicable to the case where the models corresponding to the search scenario and the recommendation scenario are trained simultaneously. The method can be executed by a model training device, which can be implemented in the form of hardware and / or software, and the model training device can be configured in an electronic device. In this embodiment, the electronic device can be a computer or server of an online recruitment system. Figure 1 As shown, the method includes:

[0040] S101: Acquire a first sample set in a search scenario and a second sample set in a recommendation scenario.

[0041] Among them, the first sample set includes several first samples, the first sample includes a search term and at least one group of first information pairs corresponding to the search term; the second sample set includes several second samples, the second samples include a group of second information pairs; a group of information pairs includes a job description feature and at least one resume feature corresponding to the job description feature.

[0042] Specifically, the first sample set may be obtained by: in a search scenario, obtaining at least one historical search term input by a historical user. One historical search term may correspond to multiple historical resume search records, each historical resume search record may determine a job description feature, and multiple resume features may be determined based on one job description feature. Therefore, each historical search term may be used as a search term for the first sample in the first sample set, and the job description feature determined by each historical resume search record and the resume feature corresponding to the job description feature may be used as the first information pair in the first sample set. The second sample set may be obtained in a similar manner to the first sample set, except that, in a recommendation scenario, the historical job description feature may be determined based on the user's historical search record, click record, or attention, and multiple resume features may be determined based on the historical job description feature.

[0043] S102: Use the first sample set and the second sample set to train the original model in sequence to obtain a target recommendation model.

[0044] The target recommendation model is used to actively recommend recruitment information to users. The original model can be a deep learning network.

[0045] Specifically, the original model is first trained using the first sample set. Since the first sample set is a sample set in a search scenario, and there is no search term feature in the recommendation scenario, in order to ensure the consistency of the target recommendation model obtained by training the first sample set and the second sample set using different features, the original model is first pre-trained using the first sample set to obtain a first pre-trained model. Then, the search term features that did not originally exist in the second sample set are processed accordingly, and the processed second sample set is input into the first pre-trained model for model optimization training, that is, the pre-trained recommendation model is fine-tuned using the second sample set. After the training is completed, the target recommendation model can be obtained.

[0046] Exemplarily, in one implementation, the target recommendation model can be obtained according to the following steps:

[0047] (i) Use the first sample set to train the original model to obtain the first pre-trained model. That is, directly use the three features in the first sample set as input data, input them into the original model and train the original model to obtain the first pre-trained model. (ii) For each second sample in the second sample set, add the default mask value as the search term for constructing the second sample to the second sample, and use the second sample set to train the first pre-trained model to obtain the target recommendation model. That is, since the second sample set only includes two features "job description feature" and "resume feature", and does not have the feature "search term", it is necessary to add the "search term" feature to the second sample set so that the features finally input into the first pre-trained model are still three features. Therefore, the default mask value can be first added as the "search term" to the second sample set, and then the second sample set can be input into the first pre-trained model. The model will automatically identify the position that needs to be masked and mask it. Finally, the pre-trained recommendation model can be trained using the second sample set, and after the training is completed, the target recommendation model can be obtained.

[0048] In another implementation, since the target recommendation model is more dependent on the job description features and resume features than on the search terms, before the first sample set is input into the original model, the weight importance of each feature can be determined based on the pre-set weight of each feature and each feature, and the importance of the feature can be input into the original model together with the feature for training. For example, the pre-set feature weight information is obtained, and the feature weight information indicates that when training the target recommendation model, the weight corresponding to the search term in the first sample set is A1, the weight corresponding to the job description feature is B1, and the weight corresponding to the resume feature is C1. Among them, B1>C1>A1. Before inputting the first sample set into the original model, each search term in the first sample set can be multiplied by A1, each job description feature in the first sample set can be multiplied by B1, and each resume feature in the first sample set can be multiplied by C1, and then the first sample set multiplied by the corresponding weights can be input into the original model for training.

[0049] In another implementation, the search terms of the second samples can be constructed directly based on the job description features in the second information pair corresponding to each second sample in the second sample set. That is, the job description features in the second information pair are copied, and the copied job description features are used as the search terms of the second samples. The second sample set after the search terms are constructed is used as input data to train the model.

[0050] In this embodiment, the original model is first trained using the first sample set in the search scenario, and then the model is optimized using the second sample set in the recommendation scenario to obtain the final optimized target recommendation model, which solves the problem that the current training process of the recommendation model has little correlation with the search scenario, and implements preliminary training of the model based on samples from other scenarios, and then fine-tunes and optimizes the model based on samples from this scenario, so that the final target recommendation model not only has more training samples, but also ensures that the target recommendation model can be more closely related to the search scenario.

[0051] S103: Use the second sample set and the first sample set to train the original model in sequence to obtain a target search model.

[0052] The target search model is used to recommend recruitment information to users based on the target search terms input by users. The original model can be a deep learning network.

[0053] Specifically, the second sample set is first used to train the original model. Since the second sample set is a sample set in the recommendation scenario, and the search scenario has an additional feature of search terms. Therefore, in order to ensure the compatibility of the target search model finally trained with the second sample set, search term features that are not currently in the second sample set will be constructed or added, and the second sample set with constructed or added search term features will be used to train the original model to obtain a pre-trained search model. And, the pre-trained recommendation model is then optimized and trained using the first sample set, that is, the pre-trained search model is fine-tuned using the first sample set. After the training is completed, the target search model can be obtained.

[0054] Exemplarily, in one implementation, the target search model can be obtained according to the following steps:

[0055] (1) For each second sample in the second sample set, construct the search term for the second sample based on the job description features in the second information pair. That is, since the job description features and the search term features are highly correlated, the job description features in the second information pair can be directly used as the search term corresponding to each second sample. After the job description features are used as search terms to construct the search terms for the second sample, a second sample can include a set of second information pairs and search terms corresponding to the second information pairs. (2) Use the second sample set after constructing the search terms to train the original model to obtain a second pre-trained model. (3) Use the first sample set to train the second pre-trained model to obtain a target search model.

[0056] In another implementation, a corresponding "blank" may be directly added to each information pair, and the "blank" may be used to replace the search term feature, and the original model may be trained using the added second sample set.

[0057] In another implementation, after taking the job description feature as the search term, the original model is trained using the second sample set after constructing the search term, and the second pre-trained model can also be obtained according to the following steps:

[0058] (I) Obtaining feature weight information. The feature weight information is used to indicate the feature weight ratio of the search term of the second sample in the second sample set after the search term is constructed, the job description feature in the second information pair, and at least one resume feature corresponding to the job description feature. That is, in the search model, the more critical feature is the search term. Therefore, after the job description feature is used as the search term, the feature weight ratio of the search term, the job description feature, and the resume feature in the search scenario can be obtained.

[0059] (ii) The second sample set after constructing the search terms and the feature weight information are used to train the original model to obtain a second pre-trained model. Specifically, the search terms, job description features, and resume features can be readjusted according to the weight ratio of the search terms, the weight ratio of the job description features, and the weight ratio of the resume features, and then the original model is trained. For example, the weight corresponding to the newly constructed search terms in the second sample set is A2, the weight corresponding to the job description features is B2, and the weight corresponding to the resume features is C2. Among them, A2>B2>C2. Before inputting the second sample set into the original model, each search term in the second sample set can be multiplied by A2, each job description feature in the second sample set can be multiplied by B2, and each resume feature in the second sample set can be multiplied by C2, and then the second sample set multiplied by the corresponding weights can be input into the original model for training.

[0060] In this embodiment, the second sample set is first used to train the original model, and then the first sample set is used to optimize the model, and finally the target search model is obtained, which solves the problem of low correlation with the recommended scene when the search model is currently trained, and realizes the initial model training based on samples of other scenes, and then the initial model is fine-tuned and optimized based on the samples of this scene, so that the final target search model not only has more training samples, but also can ensure that the target search model can be more closely related to the recommended scene.

[0061] It is worth noting that S102 and S103 are parallel steps. In the specific implementation process, they can be executed simultaneously or separately, which is not limited here.

[0062] The model training method provided by the embodiment of the present invention obtains a first sample set in a search scenario and a second sample set in a recommendation scenario; the first sample set and the second sample set are used in sequence to train the original model to obtain a target recommendation model; and the second sample set and the first sample set are used in sequence to train the original model to obtain a target search model. In the above technical scheme, the original model is first trained with the first sample set in the search scenario, and then the model is optimized with the second sample set in the recommendation scenario to obtain the final optimized target recommendation model, which solves the problem that the current training process of the recommendation model has little correlation with the search scenario, realizes the preliminary training of the model based on samples of other scenarios, and then fine-tunes and optimizes the model based on samples of this scenario. In addition, the original model is first trained with the second sample set, and then the model is optimized with the first sample set, and finally the target search model is obtained, which solves the problem that the current training of the search model has little correlation with the recommendation scenario, realizes the training of the initial model based on samples of other scenarios, and then fine-tunes and optimizes the initial model based on samples of this scenario. Furthermore, since the above scheme implements that when training the model corresponding to a certain scenario, samples from other scenarios are first used for preliminary model training, and then the samples of this scenario are used to fine-tune and optimize the model training, and finally the model for this scenario is obtained, therefore, the above method not only enables the model of a certain scenario to use samples from other scenarios for more and more sufficient training, thereby increasing the number of samples, but also ensures that the model of this scenario has a strong correlation with other scenarios, thereby ensuring that when users use the online recruitment system, no matter what application scenario they are in, when they input the same or similar specific features, the recruitment results recommended to users can be guaranteed to be consistent and strongly correlated with recruitment information in different scenarios.

[0063] Figure 2 This is a flow chart of a recruitment information recommendation method provided by an embodiment of the present invention. The target model mentioned in this embodiment is a target recommendation model or a target search model trained by the model training method mentioned in the above embodiment. Figure 2 As shown, the method includes:

[0064] S201: Determine the user's target operation, and obtain a target model according to the target operation.

[0065] The target operation includes at least one of a search operation and an option operation. In this embodiment, the search operation is a user-defined search term independently input by the user based on the online recruitment system. The option operation is an operation selected by the user based on the option features pre-set in the online recruitment system.

[0066] Specifically, when a user logs into the online recruitment system and enters the main page of the system, the user's target operation can be determined first, and the target scenario required by the user can be determined according to the target operation, and the target model can be determined according to the target scenario.

[0067] For example, in one implementation, the user enters a search term in the search box on the main page. At this time, it can be determined that the target operation is a search operation. Therefore, it can be determined that the user is in a search scenario based on the search operation, and the target model is the target search model. In another implementation, the user only selects an optional item in the preset option box on the main page. At this time, it can be determined that the target operation is an option operation. Therefore, it can be determined that the user is in a recommendation scenario based on the option operation, and the target model is the target recommendation model.

[0068] Therefore, obtaining the target model according to the target operation may include:

[0069] If the target operation includes a search operation, the target model is a target search model; if the target operation only includes an option operation, the target model is a target recommendation model. And, if the target operation includes a search operation and an option operation, the target model is a target search model.

[0070] S202: Determine and recommend recruitment information to users based on the target operation and the target model.

[0071] Specifically, the target operation is used to indicate the user's current urgent need for recruitment information, and the target model is used to determine whether the recruitment scenario required by the user is a recommendation scenario or a search scenario. Therefore, the online recruitment system can extract the information in the target operation and input it into the target model. After the target model outputs the recruitment information, it can recommend the recruitment information to the user.

[0072] Exemplarily, determining the job posting may include:

[0073] If the target operation includes a search operation, the target search term corresponding to the search operation is obtained, and the target search term is input into the target search model to obtain the recruitment information corresponding to the target search term. If the target operation only includes an option operation, the target job description feature corresponding to the option operation is obtained, and the target job description feature is input into the target recommendation model to obtain the recruitment information corresponding to the target job description feature. And, if the target operation includes a search operation and an option operation, the corresponding target search term is determined according to the search operation, and the corresponding target job description feature is determined according to the option operation, and then the target search term and the target job description feature are input into the target search model to obtain the recruitment information corresponding to the target search term and the target job description feature.

[0074] Take a certain scenario as an example. When a user opens an online recruitment system, first, the target operation is an option operation, that is, the user selects "education-bachelor's degree" in the system's preset options. At this time, the online recruitment system inputs the job description feature of "education-bachelor's degree" into the target recommendation model, and the first target recruitment information can be obtained. And, the user continues to enter "bachelor's degree" in the search box, then the online recruitment system switches the model to the target search model, and inputs the job description feature of "education-bachelor's degree" and the search term "bachelor's degree" as input data into the target search model, and the second target recruitment information can be obtained. If the user does not perform other operations afterwards, the second target recruitment information can be recommended to the user in the end. Among them, since the target search model and the target recommendation model are trained together using samples in the other scenario and samples in this scenario, to a certain extent, the first target recruitment information and the second target recruitment information are highly correlated.

[0075] The recruitment information recommendation method provided by the embodiment of the present invention determines the user's target operation and obtains a target model based on the target operation; based on the target operation and the target model, determines and recommends recruitment information to the user. In the above technical scheme, the user's target operation is obtained, and the target search model or target recommendation model trained by the method of the above embodiment corresponding to the user's target operation is determined based on the target operation, and the recruitment information corresponding to the user's target operation is determined and recommended to the user based on the target operation and the target model. On the one hand, based on the user's target operation, the user's current more urgent recruitment information needs and recruitment scenarios can be determined, and the user can be matched with the recruitment information that is closest to the recruitment information needs in the recruitment scenario, so that the online recruitment system can automatically switch recruitment scenarios and recruitment needs according to the user's wishes, thereby improving the user's usage experience. On the other hand, since the target search model and the target recommendation model are first trained based on samples of the other scenario and then optimized based on samples in this scenario, when the user uses the online recruitment system, if the user's target operation changes, the target model will be switched. When the user's search terms are similar to the job description features, no matter which model is used as the target model, the recruitment information recommended to the user will be highly relevant, so that during the user's use, no matter what recruitment scenario the user is in, the matching accuracy of the recruitment information recommended to the user is high.

[0076] Figure 3 A schematic diagram of the structure of a model training device provided by an embodiment of the present invention. Figure 3 As shown, the device comprises:

[0077] The acquisition module 301 is used to acquire a first sample set in a search scenario and a second sample set in a recommendation scenario.

[0078] The training module 302 is used to train the original model using the first sample set and the second sample set in sequence to obtain a target recommendation model; and to train the original model using the second sample set and the first sample set in sequence to obtain a target search model.

[0079] Among them, the target recommendation model is used to actively recommend recruitment information to users, and the target search model is used to recommend recruitment information to users based on the target search terms entered by users.

[0080] Optionally, the training module 302 is specifically used for:

[0081] The search term of each first sample in the first sample set is masked, and the masked first sample set is used to train the original model to obtain a first pre-trained model; the first pre-trained model is trained with the second sample set to obtain a target recommendation model.

[0082] Optionally, the training module 302 is specifically used for:

[0083] For each second sample in the second sample set, the search term of the second sample is constructed according to the job description features in the second information pair; the original model is trained using the second sample set after the search term is constructed to obtain a second pre-trained model; the second pre-trained model is trained using the first sample set to obtain a target search model.

[0084] Optionally, the original model is trained using the second sample set after the search term is constructed to obtain a second pre-trained model, and the training module 302 is specifically used for:

[0085] Acquire feature weight information, wherein the feature weight information is used to indicate the feature weight ratio of the search term of the second sample in the second sample set after the search term is constructed, the job description feature in the second information pair, and at least one resume feature corresponding to the job description feature; use the second sample set after the search term is constructed and the feature weight information to train the original model to obtain a second pre-trained model.

[0086] The model training device provided in the embodiment of the present invention can execute the model training method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0087] Figure 4 FIG. 1 is a schematic diagram of a recruitment information recommendation device provided by an embodiment of the present invention. Figure 4 As shown, the device comprises:

[0088] The determination module 401 is used to determine the user's target operation and obtain a target model according to the target operation, wherein the target model is a target recommendation model or a target search model trained using the model training method in the above embodiment.

[0089] The recommendation module 402 is used to determine and recommend recruitment information to the user based on the target operation and the target model.

[0090] Optionally, the target operation includes at least one of a search operation and an option operation; the target model is obtained according to the target operation, and the determination module 401 is specifically used to:

[0091] If the target operation includes a search operation, the target model is a target search model; if the target operation only includes an option operation, the target model is a target recommendation model.

[0092] Optionally, according to the target operation and the target model, the recruitment information is determined, and the recommendation module 402 is specifically used to:

[0093] If the target operation includes a search operation, obtain the target search term corresponding to the search operation, and input the target search term into the target search model to obtain the recruitment information corresponding to the target search term; if the target operation only includes an option operation, obtain the target position description feature corresponding to the option operation, and input the target position description feature into the target recommendation model to obtain the recruitment information corresponding to the target position description feature.

[0094] The recruitment information recommendation device provided in the embodiment of the present invention can execute the recruitment information recommendation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0095] Figure 5 A schematic diagram of the structure of an electronic device 10 provided for an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0096] like Figure 5As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0097] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0098] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the model training method and the recruitment information recommendation method.

[0099] In some embodiments, the model training method and the recruitment information recommendation method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the model training method and the recruitment information recommendation method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the model training method and the recruitment information recommendation method in any other appropriate manner (e.g., by means of firmware).

[0100] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0101] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0102] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0103] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0104] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0105] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0106] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0107] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A model training method, characterized in that: The method comprises: Obtaining a first sample set in a search scenario and a second sample set in a recommendation scenario; The first sample set and the second sample set are used in sequence to train the original model to obtain a target recommendation model; and the second sample set and the first sample set are used in sequence to train the original model to obtain a target search model; The target recommendation model is used to actively recommend recruitment information to the user, and the target search model is used to recommend recruitment information to the user based on the target search term input by the user.

2. The model training method according to claim 1, characterized in that: The first sample set includes a plurality of first samples, wherein the first sample includes a search term and at least one set of first information pairs corresponding to the search term; the second sample set includes a plurality of second samples, wherein the second samples include a set of second information pairs; A set of information pairs includes a job description feature and at least one resume feature corresponding to the job description feature.

3. The model training method according to claim 2, characterized in that: The step of sequentially training an original model using the first sample set and the second sample set to obtain a target recommendation model includes: Using the first sample set to train the original model to obtain a first pre-trained model; For each of the second samples in the second sample set, a default mask value is added to the second sample as a search term for constructing the second sample, and the first pre-trained model is trained using the second sample set to obtain the target recommendation model.

4. The model training method according to claim 2, characterized in that: The method of sequentially training the original model using the second sample set and the first sample set to obtain a target search model includes: For each of the second samples in the second sample set, constructing a search term for the second sample according to the job description features in the second information pair; The original model is trained using the second sample set after the search term is constructed to obtain a second pre-trained model; The second pre-trained model is trained using the first sample set to obtain the target search model.

5. The model training method according to claim 4, characterized in that: The method of training the original model using the second sample set after constructing the search term to obtain a second pre-trained model includes: Acquire feature weight information, wherein the feature weight information is used to indicate the feature weight ratio of the search term of the second sample in the second sample set after the search term is constructed, the job description feature in the second information pair, and at least one resume feature corresponding to the job description feature; The original model is trained using the second sample set after constructing the search term and the feature weight information to obtain a second pre-trained model.

6. A recruitment information recommendation method, characterized in that: include: Determine a target operation of the user, and acquire a target model according to the target operation, wherein the target model is a target recommendation model or a target search model trained by the model training method according to any one of claims 1 to 5; According to the target operation and the target model, job information is determined and recommended to the user.

7. The recruitment information recommendation method according to claim 6, characterized in that: The target operation includes at least one of a search operation and an option operation; The step of obtaining a target model according to the target operation includes: If the target operation includes the search operation, the target model is the target search model; If the target operation only includes the option operation, the target model is the target recommendation model.

8. The recruitment information recommendation method according to claim 7, characterized in that: The step of determining recruitment information according to the target operation and the target model includes: If the target operation includes the search operation, obtaining a target search term corresponding to the search operation, and inputting the target search term into the target search model to obtain recruitment information corresponding to the target search term; If the target operation only includes the option operation, the target job description feature corresponding to the option operation is obtained, and the target job description feature is input into the target recommendation model to obtain the recruitment information corresponding to the target job description feature.

9. An electronic device, characterized in that: include: one or more processors; a memory for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the model training method as described in any one of claims 1 to 5, or implement the recruitment information recommendation method as described in any one of claims 6 to 8.

10. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, it implements the model training method as described in any one of claims 1 to 5, or implements the recruitment information recommendation method as described in any one of claims 6 to 8.

Images