A policy matching method, device, equipment and medium

Abstract

The present invention discloses a policy matching method, apparatus, equipment and medium. Since the matching score in this method is determined based on the keywords of each indicator of the policy to be matched, compared with the method of matching based on quantifiable policy indicators in the prior art, the keyword reference of each indicator of the policy in this method is more, so the determined matching score is more accurate, thereby determining whether the user to be matched and the policy to be matched are matched with higher accuracy based on the matching score and a preset threshold.

Description

Technical Field

[0001] The present invention relates to the field of computer artificial intelligence data processing technology, and in particular to a policy matching method, device, equipment and medium. Background Art

[0002] A policy intelligent matching method in the prior art adopts a policy intelligent matching system for matching, and the policy intelligent matching system includes a user platform and a background management system; the user platform includes a user category selection module, a user information filling module, and a policy matching module, and the background management system includes a policy database, a policy compilation module, and a user management module; the policy intelligent matching system is used to collect and store policy information, compile the quantifiable qualitative hard indicators of each policy information into a matching formula, input the user information and match it with each policy, and the match that meets the matching formula is successful.

[0003] This intelligent policy matching system can help users quickly and accurately match corresponding support policies, effectively solving the time-consuming and labor-intensive problems of searching, screening, evaluating and matching existing projects, and better preparing users for project application planning. However, since only quantifiable indicators are matched during the matching process, and non-quantifiable parts are not matched, the accuracy of policy matching is low. Summary of the Invention

[0004] Embodiments of the present invention provide a policy matching method, apparatus, device, and medium to solve the problem of low accuracy of policy matching in the prior art.

[0005] An embodiment of the present invention provides a policy matching method, the method comprising:

[0006] Determine each target behavior attribute information of the to-be-matched user for the to-be-matched policy, and determine each target attribute information of the to-be-matched user based on pre-stored each target basic attribute information of the to-be-matched user;

[0007] Based on the pre-trained policy matching model, the matching score between the user to be matched and the policy to be matched is obtained according to the input target attribute information of the user to be matched and the keywords of each indicator of the policy to be matched;

[0008] According to the matching score and a preset threshold, it is determined whether the user to be matched matches the policy to be matched.

[0009] Furthermore, the step of determining each target behavior attribute information of the to-be-matched user for the to-be-matched policy includes:

[0010] Collecting the behavior data of the user to be matched on the page of the policy to be matched, wherein the behavior data includes browsing time, number of browsing times within a set time period, number of downloads, and number of clicks;

[0011] Each target behavior attribute information of the to-be-matched user for the to-be-matched policy is determined according to the behavior data.

[0012] Furthermore, the training process of the policy matching model includes:

[0013] For any sample matching combination in the sample set, obtain the sample matching combination and first label information corresponding to the sample matching combination, wherein the sample matching combination includes keywords for each indicator of the policy and each attribute information of the user, and the first label information identifies the matching score between the policy and the user in the sample matching combination;

[0014] Inputting the sample matching combination into the original deep learning model, and obtaining the output second label information of the sample matching combination;

[0015] According to the first label information and the second label information, the parameter values of each parameter of the original deep learning model are adjusted to obtain the trained policy matching model.

[0016] Furthermore, inputting the sample matching combination into the original deep learning model and obtaining the output second label information of the sample matching combination includes:

[0017] Inputting the sample matching combination into the original deep learning model, and determining the correlation value between the policy in the sample matching combination and the user based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user;

[0018] The output second label information identifying the association value of the sample matching combination is obtained.

[0019] Furthermore, determining the correlation value between the policy in the sample matching combination and the user based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user includes:

[0020] According to the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user, the correlation calculation function relationship is used to perform correlation calculation to determine the correlation value between the policy in the sample matching combination and the user.

[0021] Furthermore, before obtaining, for any sample matching combination in the sample set, the sample matching combination and the first label information corresponding to the sample matching combination, the method further includes:

[0022] For each pre-saved policy, determine the target user condition of the policy based on the correspondence between the pre-saved policy and the user condition;

[0023] Determine each target user that meets the target user condition based on each pre-stored attribute information of each user and the target user condition;

[0024] For each target user, a sample matching combination in the sample set is determined based on each attribute information of the target user and the keywords of each indicator of the policy.

[0025] Accordingly, an embodiment of the present invention provides a policy matching device, comprising:

[0026] a determination module, configured to determine each target behavior attribute information of the to-be-matched user for the to-be-matched policy, and determine each target attribute information of the to-be-matched user based on pre-stored each target basic attribute information of the to-be-matched user;

[0027] The matching module is used to obtain the matching score between the user to be matched and the policy to be matched based on the pre-trained policy matching model, according to the target attribute information of the user to be matched and the keywords of each indicator of the policy to be matched; and determine whether the user to be matched and the policy to be matched are matched based on each matching score and a preset threshold.

[0028] Furthermore, the determination module is specifically used to determine each target behavior attribute information of the user to be matched for the policy to be matched, including: collecting behavior data of the user to be matched for the page of the policy to be matched, wherein the behavior data includes browsing time, number of browsing times within a set time period, number of downloads, and number of clicks; determining each target behavior attribute information of the user to be matched for the policy to be matched based on the behavior data.

[0029] Furthermore, the device further comprises:

[0030] A training module, which is used for the training process of the policy matching model, includes: for any sample matching combination in the sample set, obtaining the sample matching combination and the first label information corresponding to the sample matching combination, wherein the sample matching combination includes keywords for each indicator of the policy and each attribute information of the user, and the first label information identifies the matching score between the policy and the user in the sample matching combination; inputting the sample matching combination into the original deep learning model to obtain the output second label information of the sample matching combination; adjusting the parameter values of each parameter of the original deep learning model according to the first label information and the second label information to obtain the trained policy matching model.

[0031] Furthermore, the training module is specifically used to input the sample matching combination into the original deep learning model, and obtain the output second label information of the sample matching combination, including: inputting the sample matching combination into the original deep learning model, and determining the correlation value between the policy and the user in the sample matching combination based on the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user; and obtaining the output second label information that identifies the correlation value of the sample matching combination.

[0032] Furthermore, the training module is specifically used to determine the correlation value between the policy and the user in the sample matching combination based on the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user, including: performing correlation calculation based on the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user using a correlation calculation function relationship to determine the correlation value between the policy and the user in the sample matching combination.

[0033] Furthermore, the training module is specifically used for obtaining the sample matching combination and the first label information corresponding to the sample matching combination for any sample matching combination in the sample set. The method also includes: for each pre-saved policy, determining the target user condition of the policy based on the correspondence between the pre-saved policy and the user condition; determining each target user who meets the target user condition based on each attribute information of each user saved in advance and the target user condition; for each target user, determining the sample matching combination in the sample set based on each attribute information of the target user and the keyword of each indicator of the policy.

[0034] Accordingly, an embodiment of the present invention provides an electronic device, comprising a processor and a memory, wherein the memory is used to store program instructions, and the processor is used to implement the steps of any of the above-mentioned policy matching methods when executing the computer program stored in the memory.

[0035] Accordingly, an embodiment of the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program implements the steps of any one of the above-mentioned policy matching methods when executed by a processor.

[0036] Embodiments of the present invention provide a policy matching method, apparatus, device, and medium. In this method, each target behavior attribute information of a user to be matched for a policy to be matched is determined, and each target attribute information of the user to be matched is determined based on pre-saved basic attribute information of each target of the user to be matched; based on a pre-trained policy matching model, a matching score between the user to be matched and the policy to be matched is obtained based on the input target attribute information of each user to be matched and the keywords of each indicator of the policy to be matched; since the matching score is determined based on the keywords of each indicator of the policy to be matched, compared with the method of matching based on quantifiable policy indicators in the prior art, more indicators are referenced based on the keywords of each indicator of the policy, so the determined matching score is more accurate, and thus the accuracy of determining whether the user to be matched and the policy to be matched is higher based on the matching score and the preset threshold. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] 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 paying any creative work.

[0038] Figure 1 A process diagram of a policy matching method provided by an embodiment of the present invention;

[0039] Figure 2 A schematic diagram of the structure of a policy matching device provided by an embodiment of the present invention;

[0040] Figure 3 A schematic diagram of the structure of a policy matching device provided by an embodiment of the present invention;

[0041] Figure 4 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0042] To make the objectives, technical solutions, and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. It is apparent that the embodiments described are only some, not all, of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0043] In order to improve the accuracy of policy matching, embodiments of the present invention provide a policy matching method, apparatus, device, and medium.

[0044] Example 1:

[0045] Figure 1 A process diagram of a policy matching method provided by an embodiment of the present invention includes the following steps:

[0046] S101: Determine each target behavior attribute information of a to-be-matched user for a to-be-matched policy, and determine each target attribute information of the to-be-matched user based on pre-stored each target basic attribute information of the to-be-matched user.

[0047] A policy matching method provided in an embodiment of the present invention is applied to an electronic device, which may be a smart terminal device such as a smartphone, a tablet computer, or a PC; or a local server, a cloud server, or other device, which is not limited in the specific embodiment of the present invention.

[0048] In order to determine whether the user to be matched matches the policy to be matched, in an embodiment of the present invention, the user to be matched refers to a user for whom the applicable policy needs to be determined. The user to be matched can be a company user or an individual user; the policy to be matched can be a tax policy, a talent policy, or other policies.

[0049] The electronic device first determines each target behavior attribute information of the user to be matched for the policy to be matched. The target behavior attribute information refers to the behavior attribute information of all behaviors of the user to be matched when viewing the page of the policy to be matched, including browsing behavior attribute information, click behavior attribute information, and download behavior attribute information.

[0050] In order to determine the target attribute information of the user to be matched, the target basic attribute information of the user to be matched can also be determined. The electronic device pre-stores each basic attribute information of each user. Based on the pre-stored basic attribute information of each user, the target basic attribute information of each user to be matched is determined. The basic attribute information is user identity-related information. The target basic attribute information includes information such as the age, gender, home address, and educational background of the user to be matched.

[0051] Based on each target behavior attribute information of the user to be matched for the policy to be matched and each target basic attribute information of the user to be matched that is saved in advance, the target attribute information of the user to be matched is determined, that is, each target behavior attribute information and each target basic attribute information are used as the target attribute information of the user to be matched, and the target attribute information includes each target behavior attribute information and each target basic attribute information.

[0052] S102: Based on the pre-trained policy matching model, the matching score between the user to be matched and the policy to be matched is obtained according to the input target attribute information of the user to be matched and the keywords of each indicator of the policy to be matched.

[0053] To determine the matching score between the user to be matched and the policy to be matched, after determining the target attribute information of the user to be matched, keywords for each indicator of the policy to be matched can also be determined. The keywords for each indicator include keywords for each quantitative indicator and keywords for each non-quantitative indicator. Each quantitative indicator refers to an indicator that can be represented by numeric characters, such as age indicators and quantity indicators. Each non-quantitative indicator refers to an indicator that cannot be represented by numeric characters, such as educational level indicators and regional indicators.

[0054] Specifically, the keywords of each indicator of the policy to be matched are pre-identified and saved, and optical character recognition (OCR) is performed on each policy on the paper document in advance to determine each character of the policy on the paper document. Based on the pre-saved indicator characters of each type of policy, the type of policy, and each character of the identified policy, the target indicator character in each character of the policy is determined, and a set number of characters after the target indicator character are used as keywords of the target indicator, and the policy and the keywords of each indicator are saved in correspondence.

[0055] For example, after identifying each character of the tax policy, based on the pre-saved indicator characters of the tax policy and each character of the identified tax policy, it is determined that each character of the tax policy includes the target non-quantitative indicator character "tax type" and the target quantitative indicator character "term", then the set number of characters after the character "tax type" are used as keywords for the target non-quantitative indicator "tax type", and the set number of characters after the character "term" are used as keywords for the target quantitative indicator "term".

[0056] After identifying each character of the talent policy, based on the pre-saved indicator characters of the talent policy and each character of the identified talent policy, it is determined that each character of the talent policy includes the target non-quantitative indicator character "education level" and the target quantitative indicator character "age". Then, the set number of characters after the character "education level" are used as keywords for the target non-quantitative indicator "education level", and the set number of characters after the character "age" are used as keywords for the target quantitative indicator "age".

[0057] The electronic device stores a pre-trained policy matching model, which is used to determine the matching score between the user to be matched and the policy to be matched. Based on the pre-trained policy matching model, the matching score between the user to be matched and the policy to be matched can be determined.

[0058] Specifically, the target attribute information of the user to be matched and the keywords for each indicator of the policy to be matched are input into the pre-trained policy matching model. The pre-trained policy matching model matches the user to be matched with the policy to be matched and determines a matching score between the user to be matched and the policy to be matched. The matching score between the user to be matched and the policy to be matched indicates the likelihood that the policy to be matched is suitable for the user to be matched.

[0059] S103: Determine whether the user to be matched matches the policy to be matched based on the matching score and a preset threshold.

[0060] In order to determine whether the user to be matched matches the policy to be matched, a preset threshold is pre-stored. Whether the user to be matched matches the policy to be matched is determined based on the comparison result between the matching score and the preset threshold. Specifically, if the matching score is not less than the preset threshold, it is determined that the user to be matched matches the policy to be matched; if the matching score is less than the preset threshold, it is determined that the user to be matched does not match the policy to be matched.

[0061] Among them, the preset threshold can be flexibly set as needed. If you want to improve the matching accuracy between the user to be matched and the policy to be matched, you can set the preset threshold to be larger; if you want to increase the matching possibility between the user to be matched and the policy to be matched, you can set the preset threshold to be smaller.

[0062] Since the matching score in the embodiment of the present invention is determined based on the keywords of each indicator of the policy to be matched and the attribute information of the user, compared with the method of matching based on quantifiable indicators of the policy in the prior art, the keyword reference indicators of each indicator of the policy in the embodiment of the present invention are more, so the determined matching score is more accurate, and thus the accuracy of determining whether the user to be matched and the policy to be matched is higher based on the matching score and the preset threshold.

[0063] Example 2:

[0064] In order to determine each target behavior attribute information of the user to be matched, based on the above embodiment, in an embodiment of the present invention, the step of determining each target behavior attribute information of the user to be matched for the policy to be matched includes:

[0065] Collecting the behavior data of the user to be matched on the page of the policy to be matched, wherein the behavior data includes browsing time, number of browsing times within a set time period, and number of clicks;

[0066] Each target behavior attribute information of the to-be-matched user for the to-be-matched policy is determined according to the behavior data.

[0067] In order to determine the target behavior attribute information of each user to be matched, in an embodiment of the present invention, the electronic device collects the behavior data of the user to be matched on the page of the policy to be matched. The behavior data includes the browsing time of the user to be matched on the page of the policy to be matched, the number and frequency of clicks of the user to be matched on the page of the policy to be matched, the number of downloads of the policy to be matched by the user to be matched when browsing the page of the policy to be matched, and the number of browsings of the user to be matched on the page of the policy to be matched within a set time period. The set time period can be one day, one week, or one month, and the embodiment of the present invention does not impose any limitation on this.

[0068] Specifically, it can be to pre-set a database of main nodes of the page of the policy to be matched, that is, to use the embedding technology to collect the behavioral data of the users to be matched on the page of the policy to be matched; or it can be to place a listener on the page of the policy to be matched through the basic code, that is, to use the non-embedding technology to collect the behavioral data of the users to be matched on the page of the policy to be matched.

[0069] Based on the collected behavioral data, determine each target behavioral attribute information of the user to be matched for the policy to be matched. Specifically, based on the browsing time information, click number information, click frequency information, download number information, and browsing number information within a set time period in the behavioral data, determine the browsing behavior attribute information, click behavior attribute information, and download behavior attribute information of the user to be matched for the policy to be matched.

[0070] Example 3:

[0071] In order to train the policy matching model, based on the above embodiments, in an embodiment of the present invention, the training process of the policy matching model includes:

[0072] For any sample matching combination in the sample set, obtain the sample matching combination and first label information corresponding to the sample matching combination, wherein the sample matching combination includes keywords for each indicator of the policy and each attribute information of the user, and the first label information identifies the matching score between the policy and the user in the sample matching combination;

[0073] Inputting the sample matching combination into the original deep learning model, and obtaining the output second label information of the sample matching combination;

[0074] According to the first label information and the second label information, the parameter values of each parameter of the original deep learning model are adjusted to obtain the trained policy matching model.

[0075] In order to realize the training of the policy matching model, an embodiment of the present invention stores a sample set for training, which includes a sample matching combination. The sample matching combination includes keywords for each indicator of the policy and each attribute information of the user, and also includes the first label information corresponding to the sample matching combination.

[0076] Among them, the first label information is used to identify the matching score between the policy and the user in the sample matching combination. The matching score is the score value received in advance when matching the policy of the sample matching combination with the user. The higher the matching score, the more matching the policy of the sample matching combination with the user.

[0077] In this embodiment of the present invention, after obtaining any sample matching combination in the sample set and the first label information for that sample matching combination, the sample matching combination is input into the original deep learning model, which then outputs the second label information for that sample matching combination. The second label information identifies the matching score assigned by the original deep learning model to the policy and user in that sample matching combination.

[0078] After determining the second label information of the sample matching combination based on the original deep learning model, the original deep learning model is trained based on the second label information and the first label information of the sample matching combination to adjust the parameter values of various parameters of the original deep learning model to obtain a trained policy matching model.

[0079] The above operation is performed on each sample matching combination included in the sample set used to train the deep learning model. When a preset condition is met, a trained deep learning model is obtained. The preset condition may be that the number of sample matching combinations in the sample set whose second label information, obtained after training with the original deep learning model, is consistent with the first label information exceeds a set number; or that the number of iterations of training the original deep learning model reaches a set maximum number of iterations. Specifically, the embodiments of the present invention do not impose any restrictions on this.

[0080] As a possible implementation method, when training the original deep learning model, the sample matching combinations in the sample set can be divided into training sample matching combinations and test sample matching combinations. The original deep learning model is first trained based on the training sample matching combinations, and then the reliability of the trained policy matching model is tested based on the test sample matching combinations.

[0081] Example 4:

[0082] In order to obtain the second label information of the output sample matching combination, based on the above embodiments, in an embodiment of the present invention, the inputting of the sample matching combination into the original deep learning model to obtain the output second label information of the sample matching combination includes:

[0083] Inputting the sample matching combination into the original deep learning model, and determining the correlation value between the policy in the sample matching combination and the user based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user;

[0084] The output second label information identifying the association value of the sample matching combination is obtained.

[0085] In order to obtain the second label information of the output sample matching combination, in an embodiment of the present invention, the sample matching combination is input into the original deep learning model. The original deep learning model performs correlation calculation based on the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user to determine the correlation value between the policy in the sample matching combination and the user.

[0086] The correlation value represents the degree of correlation between the policy and the user in the sample matching combination. Specifically, the correlation value between the policy and the user is determined based on the degree of correlation between each attribute information and the keyword of each indicator. The correlation value reflects the possibility of matching between the policy and the user. The higher the correlation value, the higher the possibility of matching between the policy and the user.

[0087] After determining the correlation value between the policy and the user in the sample matching combination, since the correlation value reflects the possibility of matching the policy and the user, the correlation value of the sample matching combination is obtained as the output and used as the second label information of the sample matching combination.

[0088] To determine the relevance value between a policy and a user, in an embodiment of the present invention, determining the relevance value between the policy and the user in the sample matching combination based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user includes:

[0089] According to the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user, the correlation calculation function relationship is used to perform correlation calculation to determine the correlation value between the policy in the sample matching combination and the user.

[0090] In order to determine the correlation value between the policy and the user, in an embodiment of the present invention, based on the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user, since there is a correlation measurement function relationship in the prior art, the correlation measurement function relationship is directly used to perform correlation measurement to determine the correlation value between the policy and the user in the sample matching combination.

[0091] The correlation calculation function relationship may be an existing Apriori algorithm function relationship or other existing correlation calculation function relationship. Calculating correlation according to the existing correlation calculation function relationship belongs to the prior art and will not be elaborated in detail in the embodiment of the present invention.

[0092] Example 5:

[0093] In order to improve the training efficiency of the policy matching model, based on the above embodiments, in an embodiment of the present invention, before obtaining, for any sample matching combination in the sample set, the sample matching combination and the first label information corresponding to the sample matching combination, the method further includes:

[0094] For each pre-saved policy, determine the target user condition of the policy based on the correspondence between the pre-saved policy and the user condition;

[0095] Determine each target user that meets the target user condition based on each pre-stored attribute information of each user and the target user condition;

[0096] For each target user, a sample matching combination in the sample set is determined based on each attribute information of the target user and the keywords of each indicator of the policy.

[0097] In order to improve the training efficiency of the policy matching model, in the embodiment of the present invention, the sample matching combination is not determined randomly, but policies and users with certain correlations are first determined and used as the sample matching combination in the sample set.

[0098] For each pre-stored policy, first, users who have a certain correlation with the policy are screened out. The electronic device pre-stores a correspondence between policies and user conditions, wherein the user conditions of each policy are predetermined.

[0099] According to the pre-stored correspondence between policies and user conditions, the user condition corresponding to the policy is determined in the correspondence, and the user condition corresponding to the policy is used as the target user condition of the policy.

[0100] After determining the target user condition for the policy, each target user that meets the target user condition is determined based on the target user condition. Specifically, the electronic device pre-stores attribute information for each user, and based on the pre-stored attribute information for each user and the target user condition, each target user that meets the target user condition is determined.

[0101] The electronic device screens out each target user that has a certain association with the policy. For each target user, based on the target user and the policy, it determines each attribute information of the target user and the keyword of each indicator of the policy, and uses it as a sample matching combination in the sample set. For each target user, a corresponding sample matching combination is determined, and for each policy, each sample matching combination corresponding to each policy is determined to comprehensively determine the sample set.

[0102] Example 6:

[0103] Based on the above embodiments, Figure 2 A schematic diagram of the structure of a policy matching device provided in an embodiment of the present invention, the device comprising:

[0104] The determination module 201 is used to determine each target behavior attribute information of the to-be-matched user for the to-be-matched policy, and determine each target attribute information of the to-be-matched user based on the pre-stored each target basic attribute information of the to-be-matched user;

[0105] The matching module 202 is used to obtain the matching score between the user to be matched and the policy to be matched based on the pre-trained policy matching model, according to the input of each target attribute information of the user to be matched and the keyword of each indicator of the policy to be matched; and determine whether the user to be matched matches the policy to be matched based on each matching score and a preset threshold.

[0106] Furthermore, the determination module is specifically used to determine each target behavior attribute information of the user to be matched for the policy to be matched, including: collecting behavior data of the user to be matched for the page of the policy to be matched, wherein the behavior data includes browsing time, number of browsing times within a set time period, number of downloads, and number of clicks; determining each target behavior attribute information of the user to be matched for the policy to be matched based on the behavior data.

[0107] Furthermore, the device further comprises:

[0108] A training module, which is used for the training process of the policy matching model, includes: for any sample matching combination in the sample set, obtaining the sample matching combination and the first label information corresponding to the sample matching combination, wherein the sample matching combination includes keywords for each indicator of the policy and each attribute information of the user, and the first label information identifies the matching score between the policy and the user in the sample matching combination; inputting the sample matching combination into the original deep learning model to obtain the output second label information of the sample matching combination; adjusting the parameter values of each parameter of the original deep learning model according to the first label information and the second label information to obtain the trained policy matching model.

[0109] Furthermore, the training module is specifically used to input the sample matching combination into the original deep learning model, and obtain the output second label information of the sample matching combination, including: inputting the sample matching combination into the original deep learning model, and determining the correlation value between the policy and the user in the sample matching combination based on the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user; and obtaining the output second label information that identifies the correlation value of the sample matching combination.

[0110] Furthermore, the training module is specifically used to determine the correlation value between the policy and the user in the sample matching combination based on the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user, including: performing correlation calculation based on the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user using a correlation calculation function relationship to determine the correlation value between the policy and the user in the sample matching combination.

[0111] Furthermore, the training module is specifically used for obtaining the sample matching combination and the first label information corresponding to the sample matching combination for any sample matching combination in the sample set. The method also includes: for each pre-saved policy, determining the target user condition of the policy based on the correspondence between the pre-saved policy and the user condition; determining each target user who meets the target user condition based on each attribute information of each user saved in advance and the target user condition; for each target user, determining the sample matching combination in the sample set based on each attribute information of the target user and the keyword of each indicator of the policy.

[0112] The structure of the policy matching device of the embodiment of the present invention is described below through a specific embodiment. Figure 3 A schematic diagram of the structure of a policy matching device provided by an embodiment of the present invention is shown in FIG. Figure 3As shown, the policy matching device includes a matching module 301 , a condition setting module 302 , a feedback module 303 , and a training module 304 .

[0113] The matching module 301 is connected to the condition setting module 302, the feedback module 303, and the training module 304 respectively. The matching module 301 is used to determine the target user condition of each policy according to the correspondence between the policy and the user condition pre-stored in the condition setting module 302; determine each target user who meets the target user condition according to the target user condition, and for each target user, send each attribute information of the target user and the keyword of each indicator of the policy as a sample matching combination in the sample set to the training module 304; and determine the target user to be matched. Match the user's target behavior attribute information for the policy to be matched, and determine each target attribute information of the user to be matched based on the pre-saved basic attribute information of each target of the user to be matched; based on the policy matching model pre-trained in the training module 304, obtain the matching score between the user to be matched and the policy to be matched according to the input keywords of each target attribute information of the user to be matched and each indicator of the policy to be matched; determine whether the user to be matched matches the policy to be matched based on the matching score and the preset threshold; equivalent to the determination module 201 and the matching module 202 of the policy matching device in Example 6.

[0114] The condition setting module 302 is connected to the matching module 301 and is used to receive the user conditions for each policy, determine the corresponding relationship between the policy and the user conditions, and save the corresponding relationship.

[0115] The feedback module 303 is connected to the matching module 301, and is used to receive the scoring value of the policy and user in each sample matching combination determined by the matching module 301 and use it as the matching score of the policy and user, and send the matching score to the matching module 301 and save it in the sample matching combination.

[0116] The training module 304 is connected to the matching module 301 and is used to train the original deep learning model according to the sample matching combination in the sample set sent by the matching module 301 to obtain a trained policy matching model, which is equivalent to the training module of the policy matching device in Example 6.

[0117] Example 7:

[0118] Figure 4This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Based on the above embodiments, an electronic device is further provided in an embodiment of the present invention, including a processor 401, a communication interface 402, a memory 403, and a communication bus 404. The processor 401, the communication interface 402, and the memory 403 communicate with each other via the communication bus 404.

[0119] The memory 403 stores a computer program. When the program is executed by the processor 401, the processor 401 performs the following steps:

[0120] Determine each target behavior attribute information of the to-be-matched user for the to-be-matched policy, and determine each target attribute information of the to-be-matched user based on pre-stored each target basic attribute information of the to-be-matched user;

[0121] Based on the pre-trained policy matching model, the matching score between the user to be matched and the policy to be matched is obtained according to the input target attribute information of the user to be matched and the keywords of each indicator of the policy to be matched;

[0122] According to the matching score and a preset threshold, it is determined whether the user to be matched matches the policy to be matched.

[0123] Furthermore, the processor 401 is specifically configured to determine each target behavior attribute information of the to-be-matched user for the to-be-matched policy, including:

[0124] Collecting the behavior data of the user to be matched on the page of the policy to be matched, wherein the behavior data includes browsing time, number of browsing times within a set time period, number of downloads, and number of clicks;

[0125] Each target behavior attribute information of the to-be-matched user for the to-be-matched policy is determined according to the behavior data.

[0126] Furthermore, the processor 401 is further configured to train the policy matching model in a process including:

[0127] For any sample matching combination in the sample set, obtain the sample matching combination and first label information corresponding to the sample matching combination, wherein the sample matching combination includes keywords for each indicator of the policy and each attribute information of the user, and the first label information identifies the matching score between the policy and the user in the sample matching combination;

[0128] Inputting the sample matching combination into the original deep learning model, and obtaining the output second label information of the sample matching combination;

[0129] According to the first label information and the second label information, the parameter values of each parameter of the original deep learning model are adjusted to obtain the trained policy matching model.

[0130] Furthermore, the processor 401 is specifically configured to input the sample matching combination into the original deep learning model, and obtain the output second label information of the sample matching combination, including:

[0131] Inputting the sample matching combination into the original deep learning model, and determining the correlation value between the policy in the sample matching combination and the user based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user;

[0132] The output second label information identifying the association value of the sample matching combination is obtained.

[0133] Furthermore, the processor 401 is specifically configured to determine the correlation value between the policy in the sample matching combination and the user based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user, including:

[0134] According to the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user, the correlation calculation function relationship is used to perform correlation calculation to determine the correlation value between the policy in the sample matching combination and the user.

[0135] Furthermore, the processor 401 is further configured to, for any sample matching combination in the sample set, obtain the sample matching combination and the first label information corresponding to the sample matching combination, and the method further includes:

[0136] For each pre-saved policy, determine the target user condition of the policy based on the correspondence between the pre-saved policy and the user condition;

[0137] Determine each target user that meets the target user condition based on each pre-stored attribute information of each user and the target user condition;

[0138] For each target user, a sample matching combination in the sample set is determined based on each attribute information of the target user and the keywords of each indicator of the policy.

[0139] The communication bus mentioned in the electronic device mentioned above may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is used in the figure, but this does not mean that there is only one bus or only one type of bus.

[0140] The communication interface 402 is used for communication between the electronic device and other devices.

[0141] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk memory. Alternatively, the memory may be at least one storage device located away from the processor.

[0142] The above-mentioned processor can be a general-purpose processor, including a central processing unit, a network processor (NP), etc.; it can also be a digital signal processing processor (DSP), an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc.

[0143] Example 8:

[0144] Based on the above embodiments, an embodiment of the present invention further provides a computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to perform the following steps:

[0145] Determine each target behavior attribute information of the to-be-matched user for the to-be-matched policy, and determine each target attribute information of the to-be-matched user based on pre-stored each target basic attribute information of the to-be-matched user;

[0146] Based on the pre-trained policy matching model, the matching score between the user to be matched and the policy to be matched is obtained according to the input target attribute information of the user to be matched and the keywords of each indicator of the policy to be matched;

[0147] According to the matching score and a preset threshold, it is determined whether the user to be matched matches the policy to be matched.

[0148] Furthermore, the step of determining each target behavior attribute information of the to-be-matched user for the to-be-matched policy includes:

[0149] Collecting the behavior data of the user to be matched on the page of the policy to be matched, wherein the behavior data includes browsing time, number of browsing times within a set time period, number of downloads, and number of clicks;

[0150] Each target behavior attribute information of the to-be-matched user for the to-be-matched policy is determined according to the behavior data.

[0151] Furthermore, the training process of the policy matching model includes:

[0152] For any sample matching combination in the sample set, obtain the sample matching combination and first label information corresponding to the sample matching combination, wherein the sample matching combination includes keywords for each indicator of the policy and each attribute information of the user, and the first label information identifies the matching score between the policy and the user in the sample matching combination;

[0153] Inputting the sample matching combination into the original deep learning model, and obtaining the output second label information of the sample matching combination;

[0154] According to the first label information and the second label information, the parameter values of each parameter of the original deep learning model are adjusted to obtain the trained policy matching model.

[0155] Furthermore, inputting the sample matching combination into the original deep learning model and obtaining the output second label information of the sample matching combination includes:

[0156] Inputting the sample matching combination into the original deep learning model, and determining the correlation value between the policy in the sample matching combination and the user based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user;

[0157] The output second label information identifying the association value of the sample matching combination is obtained.

[0158] Furthermore, determining the correlation value between the policy in the sample matching combination and the user based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user includes:

[0159] According to the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user, the correlation calculation function relationship is used to perform correlation calculation to determine the correlation value between the policy in the sample matching combination and the user.

[0160] Furthermore, before obtaining, for any sample matching combination in the sample set, the sample matching combination and the first label information corresponding to the sample matching combination, the method further includes:

[0161] For each pre-saved policy, determine the target user condition of the policy based on the correspondence between the pre-saved policy and the user condition;

[0162] Determine each target user that meets the target user condition based on each pre-stored attribute information of each user and the target user condition;

[0163] For each target user, a sample matching combination in the sample set is determined based on each attribute information of the target user and the keywords of each indicator of the policy.

[0164] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0165] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0166] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0167] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0168] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A policy matching method, characterized in that: The method comprises: Determine each target behavior attribute information of the to-be-matched user for the to-be-matched policy, and determine each target attribute information of the to-be-matched user based on pre-stored each target basic attribute information of the to-be-matched user; Based on the pre-trained policy matching model, the matching score between the user to be matched and the policy to be matched is obtained according to the input target attribute information of the user to be matched and the keywords of each indicator of the policy to be matched; Determining whether the user to be matched matches the policy to be matched based on the matching score and a preset threshold; The training process of the policy matching model includes: For any sample matching combination in the sample set, obtain the sample matching combination and first label information corresponding to the sample matching combination, wherein the sample matching combination includes keywords for each indicator of the policy and each attribute information of the user, and the first label information identifies the matching score between the policy and the user in the sample matching combination; Inputting the sample matching combination into the original deep learning model, and obtaining the output second label information of the sample matching combination; Adjusting the parameter values of each parameter of the original deep learning model according to the first label information and the second label information to obtain the trained policy matching model; Inputting the sample matching combination into the original deep learning model and obtaining the output second label information of the sample matching combination includes: Inputting the sample matching combination into the original deep learning model, and determining the correlation value between the policy in the sample matching combination and the user based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user; The output second label information identifying the association value of the sample matching combination is obtained.

2. The method according to claim 1, characterized in that Determining each target behavior attribute information of the to-be-matched user for the to-be-matched policy includes: Collecting the behavior data of the user to be matched on the page of the policy to be matched, wherein the behavior data includes browsing time, number of browsing times within a set time period, number of downloads, and number of clicks; Each target behavior attribute information of the to-be-matched user for the to-be-matched policy is determined according to the behavior data.

3. The method according to claim 1, characterized in that Determining the correlation value between the policy in the sample matching combination and the user based on the keyword of each indicator of the policy in the sample matching combination and each attribute information of the user includes: According to the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user, the correlation calculation function relationship is used to perform correlation calculation to determine the correlation value between the policy in the sample matching combination and the user.

4. The method according to claim 1, wherein Before obtaining, for any sample matching combination in the sample set, the sample matching combination and the first label information corresponding to the sample matching combination, the method further includes: For each pre-saved policy, determine the target user condition of the policy based on the correspondence between the pre-saved policy and the user condition; Determine each target user that meets the target user condition based on each pre-stored attribute information of each user and the target user condition; For each target user, a sample matching combination in the sample set is determined based on each attribute information of the target user and the keywords of each indicator of the policy.

5. A policy matching device, characterized in that: The device comprises: a determination module, configured to determine each target behavior attribute information of the to-be-matched user for the to-be-matched policy, and determine each target attribute information of the to-be-matched user based on pre-stored each target basic attribute information of the to-be-matched user; A matching module is configured to obtain a matching score between the user to be matched and the policy to be matched based on the target attribute information of the user to be matched and the keywords of each indicator of the policy to be matched, based on a pre-trained policy matching model; and determine whether the user to be matched and the policy to be matched are matched based on each matching score and a preset threshold; A training module for the policy matching model training process includes: for any sample matching combination in the sample set, obtaining the sample matching combination and first label information corresponding to the sample matching combination, wherein the sample matching combination includes keywords for each indicator of the policy and each attribute information of the user, and the first label information identifies the matching score between the policy and the user in the sample matching combination; inputting the sample matching combination into the original deep learning model to obtain second label information of the output sample matching combination; adjusting the parameter values of each parameter of the original deep learning model according to the first label information and the second label information to obtain the trained policy matching model; The training module is specifically used to input the sample matching combination into the original deep learning model, and obtain the output second label information of the sample matching combination, including: inputting the sample matching combination into the original deep learning model, and determining the correlation value between the policy and the user in the sample matching combination based on the keywords of each indicator of the policy in the sample matching combination and each attribute information of the user; and obtaining the output second label information that identifies the correlation value of the sample matching combination.

6. An electronic device, characterized in that: The electronic device includes a processor and a memory, wherein the memory is used to store program instructions, and the processor is used to implement the steps of the policy matching method according to any one of claims 1 to 4 when executing the computer program stored in the memory.

7. A computer-readable storage medium, characterized in that The device stores a computer program, which, when executed by a processor, implements the steps of the policy matching method according to any one of claims 1 to 4.

Images