Service recommendation method and device, electronic equipment and computer program product

Abstract

The invention discloses a service recommendation method and device, electronic equipment and a computer program product. The method comprises the steps that archive information and scene information of a to-be-recommended user are collected, the archive information is used for representing the personal condition of the to-be-recommended user, and the scene information is used for representing the environment condition of the to-be-recommended user and user behaviors fed back by the to-be-recommended user; analyzing the archive information by using a preset user portrait model to obtain a to-be-recommended user portrait of the to-be-recommended user; fusing the to-be-recommended user portrait of the to-be-recommended user with the scene information to obtain a fusion feature of the to-be-recommended user; and recommendation services conforming to the fusion features are selected from the multiple preset services, each preset service has a corresponding feature tag, and the recommendation services are the preset services with the feature tags conforming to the fusion features. According to the method and the device, the technical problem that accurate recommendation cannot be carried out aiming at the service required by the user is solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a service recommendation method, device, electronic device and computer program product. Background Art

[0002] There are still only a few elderly people who have access to the Internet, so the elderly are still one of the main people who handle affairs offline. They also have problems such as "not knowing what to do, not familiar with related services, and having difficulty operating smart devices". Therefore, most government services related to the elderly still require manual assistance and support. For example, community workers, grid workers, and third-party service agency personnel are used to provide door-to-door services. However, this service model is limited by the number of personnel and the complexity of work items, the low service frequency, and the large demand for staff. It is inevitable that omissions will occur, which shows that the grassroots information support capabilities are incomplete and intelligent service means are missing. In the event of an emergency, it will be difficult to respond in a timely manner.

[0003] In addition, the physical conditions of different elderly people vary greatly. For example, some elderly people are in good health and may pay more attention to cultural and recreational activities, such as calligraphy and dance classes, while those with chronic diseases such as arthritis may need rehabilitation care services, including physical therapy and auxiliary exercise equipment.

[0004] At present, the recommendation services for the elderly are unable to make targeted recommendations based on their different needs, that is, they are unable to make accurate recommendations on the services that the elderly need.

[0005] Currently, no effective solution has been proposed to the above-mentioned problem of being unable to make accurate recommendations on services required by users. Summary of the invention

[0006] The embodiments of the present invention provide a service recommendation method, device, electronic device and computer program product to at least solve the technical problem of being unable to accurately recommend services required by users.

[0007] According to one aspect of an embodiment of the present invention, a service recommendation method is provided, comprising: collecting profile information and situational information of a user to be recommended, wherein the profile information is used to represent the personal situation of the user to be recommended, and the situational information is used to represent the environmental situation of the user to be recommended and the user behavior feedback of the user to be recommended; using a preset user portrait model to analyze the profile information to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; fusing the user portrait of the user to be recommended and the situational information to obtain a fusion feature of the user to be recommended; and selecting a recommended service that meets the fusion feature from multiple preset services, wherein each of the preset services has a corresponding feature label, and the recommended service is the preset service whose feature label meets the fusion feature.

[0008] Optionally, fusing the user portrait to be recommended and the situational information of the user to be recommended to obtain the fusion feature of the user to be recommended includes: using a preset feature fusion model to analyze the situational information of the user to be recommended to obtain a situational label, and combining the user portrait to be recommended and the situational label to obtain the fusion feature, wherein the preset feature fusion model is used to identify the situational type of the situational information, and extract the situational label from the situational information based on a feature extraction strategy corresponding to the situational type.

[0009] Optionally, the recommendation service at least includes: selecting a recommended service that meets the fusion feature from multiple preset services, including: based on the user profile to be recommended in the fusion feature, selecting at least one candidate service that meets the user profile to be recommended from the multiple preset services, wherein the candidate service is the preset service whose feature tag meets the user profile to be recommended; identifying service attributes of each candidate service, wherein the service attributes include: a first attribute indicating that the candidate service has a strong association with the user to be recommended, and a second attribute indicating that the candidate service has a weak association with the user to be recommended; using the candidate service whose service attribute is the first attribute as a first recommended service, wherein the first recommended service uses a first push method to recommend to the user to be recommended, and the first push method at least includes: sending a push message to the user to be recommended; using the candidate service whose service attribute is the second attribute as a second recommended service, wherein the second recommended service uses a second push method to recommend to the user to be recommended, and the second push method at least includes: adding the second recommended service to the recommendation list of the user to be recommended for display.

[0010] Optionally, after selecting at least one candidate service that matches the user profile to be recommended from multiple preset services based on the user profile to be recommended in the fused feature, the method further includes: based on the scenario information in the fused feature, selecting at least one push service that matches the scenario information from at least one of the candidate services, wherein the push service is the candidate service whose feature tag matches the scenario information.

[0011] Optionally, after taking the candidate service whose service attribute is the second attribute as the second recommended service, the method further includes: based on the context information in the fused feature, detecting whether the feature tag corresponding to the second recommended service conforms to the context information; if the feature tag corresponding to the second recommended service conforms to the context information, improving the arrangement order of the second recommended service in the recommendation list; if the feature tag corresponding to the second recommended service does not conform to the context information, lowering the arrangement order of the second recommended service in the recommendation list.

[0012] Optionally, in the case that there are multiple context information in the fusion feature, the method further includes: detecting the feature tag corresponding to the second recommendation service with each context information in turn to obtain the number of information matches of the feature tag that matches the context information; and determining the arrangement order of the second recommendation service in the recommendation list according to the number of information matches, wherein the larger the number of information matches of the feature tag corresponding to the second recommendation service, the higher the arrangement order of the second recommendation service in the recommendation list.

[0013] Optionally, after taking the candidate service whose service attribute is the second attribute as the second recommended service, the method further includes: collecting emotional feedback information of the user to be recommended on the second recommended service, wherein the emotional feedback information is determined at least based on physiological data of the user to be recommended, or behavioral data of the user to be recommended, and the physiological data at least includes: heart rate, blood pressure and sleep quality of the user to be recommended, and the behavioral data at least includes: facial expressions and voice information of the user to be recommended; adjusting the arrangement order of the second recommended service in the recommendation list according to the emotional feedback information, wherein, when the emotional feedback information is positive, the arrangement order of the second recommended service in the recommendation list is improved, and when the emotional feedback information is negative, the arrangement order of the second recommended service in the recommendation list is reduced.

[0014] According to another aspect of an embodiment of the present invention, a service recommendation device is also provided, including: a collection module, used to collect profile information and situational information of a user to be recommended, wherein the profile information is used to represent the personal situation of the user to be recommended, and the situational information is used to represent the environmental situation of the user to be recommended and the user behavior feedbacked by the user to be recommended; an analysis module, used to use a preset user portrait model to analyze the profile information to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; a fusion module, used to fuse the user portrait of the user to be recommended and the situational information to obtain a fusion feature of the user to be recommended; a selection module, used to select a recommended service that meets the fusion feature from multiple preset services, wherein each of the preset services has a corresponding feature label, and the recommended service is the preset service whose feature label meets the fusion feature.

[0015] According to another aspect of an embodiment of the present invention, there is further provided an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the above service recommendation method through the computer program.

[0016] According to another aspect of an embodiment of the present invention, a computer program product is provided, including computer instructions, which implement the steps of the above service recommendation method when executed by a processor.

[0017] In an embodiment of the present invention, the profile information and situational information of the user to be recommended are collected, wherein the profile information is used to represent the personal situation of the user to be recommended, and the situational information is used to represent the environmental situation of the user to be recommended and the user behavior fed back by the user to be recommended; the profile information is analyzed using a preset user portrait model to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; the user portrait and situational information of the user to be recommended are fused to obtain a fusion feature of the user to be recommended; from multiple preset services, a recommended service that meets the fusion feature is selected, wherein each preset service has a corresponding feature label, and the recommended service is a preset service whose feature label meets the fusion feature, so that by fusing the user portrait and the user's situational information, and recommending the service based on the fused fusion feature, a technical effect of accurately recommending the services required by the user is achieved, thereby solving the technical problem of being unable to accurately recommend the services required by the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0019] Figure 1 is a flow chart of a service recommendation method according to an embodiment of the present invention;

[0020] Figure 2 is a schematic diagram of a precise recommendation method integrating elderly care government affairs and comprehensive elderly care services according to an embodiment of the present invention;

[0021] Figure 3 is a schematic diagram of a service recommendation device according to an embodiment of the present invention;

[0022] Figure 4 It is a structural block diagram of a computer terminal according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] 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.

[0024] 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.

[0025] First, some nouns or terms that appear in the description of the embodiments of the present application are subject to the following explanations:

[0026] User portrait: A user labeling system built based on the information of a large number of target users. It characterizes the target population corresponding to a product or service, thereby achieving optimized product design and precise delivery of services.

[0027] According to an embodiment of the present invention, a service recommendation method embodiment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0028] Figure 1 is a flow chart of a service recommendation method according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0029] Step S102, collecting the profile information and context information of the user to be recommended, wherein the profile information is used to represent the personal situation of the user to be recommended, and the context information is used to represent the environmental situation of the user to be recommended and the user behavior reported by the user to be recommended;

[0030] Step S104, using a preset user portrait model to analyze the profile information to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information;

[0031] Step S106, fusing the user portrait and context information of the user to be recommended to obtain a fusion feature of the user to be recommended;

[0032] Step S108, selecting a recommended service that meets the fusion feature from a plurality of preset services, wherein each preset service has a corresponding feature tag, and the recommended service is a preset service whose feature tag meets the fusion feature.

[0033] In an embodiment of the present invention, the profile information and situational information of the user to be recommended are collected, wherein the profile information is used to represent the personal situation of the user to be recommended, and the situational information is used to represent the environmental situation of the user to be recommended and the user behavior fed back by the user to be recommended; the profile information is analyzed using a preset user portrait model to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; the user portrait and situational information of the user to be recommended are fused to obtain a fusion feature of the user to be recommended; from multiple preset services, a recommended service that meets the fusion feature is selected, wherein each preset service has a corresponding feature label, and the recommended service is a preset service whose feature label meets the fusion feature, so that by fusing the user portrait and the user's situational information, and recommending the service based on the fused fusion feature, a technical effect of accurately recommending the services required by the user is achieved, thereby solving the technical problem of being unable to accurately recommend the services required by the user.

[0034] In the above step S102, the user to be recommended may be an elderly person, and the elderly person may be a user who meets a preset age group, such as a user aged 50-100. Users in this age group are basically at or close to retirement age and need to pay attention to information about elderly care matters.

[0035] In the above step S102, the file information can be queried through the data released by the pension insurance management system. By collecting and sorting data from multiple sources, the file information of each user to be recommended is obtained, and a user database (such as an elderly portrait database) is constructed.

[0036] Optionally, the profile information includes at least: basic demographic information, health information, financial information, and capability information of the user to be recommended.

[0037] Optionally, the basic population information may at least include: age, gender, retirement status, etc. of the user to be recommended.

[0038] Optionally, the health information may at least include: historical cases of the user to be recommended, or collected physiological parameters, which may at least include: height, weight, blood pressure, blood sugar, heart rate, body temperature, etc.

[0039] Optionally, the financial information may at least indicate: the retirement salary of the user to be recommended, and whether the user to be recommended is a low-income user.

[0040] Optionally, the profile information of the user to be recommended can be identified by the ID number of the user to be recommended and can be obtained by aggregating data from multiple sources.

[0041] In the above step S102, the context information is used to represent the environment of the user to be recommended and the user behavior fed back by the user to be recommended.

[0042] Optionally, the situational information may include at least: weather data, social news data, community activity data, etc.

[0043] Optionally, the environmental conditions of the user to be recommended may at least represent the weather data of the geographical location of the user to be recommended. For example, when the temperature of the environment of the user to be recommended decreases, the probability of the user to be recommended catching a cold will increase. Therefore, the service recommended to the user to be recommended may be related information such as prevention and treatment of colds.

[0044] Optionally, the environmental conditions of the recommended user can at least represent community activity data of the geographical location of the recommended user. For example, information on health lecture activities organized by the community (such as lecture topics, number of participants, feedback from the elderly, etc.) can be jointly analyzed with the elderly's health record data to explore the potential demand characteristics of the elderly for health knowledge acquisition and medical services in specific community activity contexts.

[0045] Optionally, the user behavior feedback of the recommended user may be feedback behavior performed by the recommended user on the recommended service after receiving the recommended service. The feedback behavior may be acquired through a camera using facial expressions of the recommended user, and emotional information obtained based on facial expression analysis; the feedback behavior may also be acquired through physiological information acquisition equipment using physiological data such as heart rate, blood pressure, and sleep quality of the recommended user, and emotional information obtained based on analysis of the physiological data.

[0046] In the above step S104, the preset user portrait model can be pre-trained by machine learning using sample profile information corresponding to multiple sample users. Using the preset user portrait model, the profile information of the user to be recommended can be analyzed to obtain the user portrait of the user to be recommended.

[0047] In the above step S106, the fusion feature may be obtained by fusing the user portrait to be recommended and the context information of the user to be recommended.

[0048] Optionally, the fused feature may be a fused feature vector, which may be obtained by fusing multi-source contextual data (i.e., contextual information including weather data, social news data, community activity data, etc.) with traditional portrait data (such as the user portrait to be recommended) using a deep learning algorithm.

[0049] Optionally, based on the contextual information of the user to be recommended, knowledge graphs in fields such as medical, social and financial fields can be pre-constructed, and then the portrait of the user to be recommended can be fused with the knowledge graphs of each field to obtain a fused knowledge graph. Then, in the fused knowledge graph, recommendation services that meet the fused features can be queried based on the fused features.

[0050] In the above step S108, the recommended service may be one preset service or a combination of multiple preset services.

[0051] As an optional embodiment, the user portrait to be recommended and the situational information of the user to be recommended are fused to obtain the fusion features of the user to be recommended, including: using a preset feature fusion model to analyze the situational information of the user to be recommended to obtain a situational label, and combining the user portrait to be recommended and the situational label to obtain a fusion feature, wherein the preset feature fusion model is used to identify the situational type of the situational information, and extract the situational label from the situational information based on the feature extraction strategy corresponding to the situational type.

[0052] In the above-mentioned embodiments of the present application, the user portrait to be recommended is obtained by machine learning, and may be a feature label that can represent the user to be recommended, and may be obtained from multiple data sources. Therefore, there may be multiple forms of expression of the content of the situational information. In order to merge the situational information with the user portrait to be recommended, the user portrait to be recommended can be used to analyze the user portrait to be recommended, and the situational information represented in multiple forms can be used to extract situational labels that can represent the situational information. Then, the situational labels can be merged with the user portrait to be recommended to obtain the fused features of the user to be recommended.

[0053] It should be noted that since situational information can exist in multiple forms, situational information with different forms can have situational labels extracted through their corresponding feature extraction strategies. Therefore, when using a preset feature fusion model to determine the situational labels corresponding to each situational information, the situational type of the situational information can be first identified, and then the corresponding feature extraction strategy can be selected based on the situational type to extract situational labels from the situational information.

[0054] Optionally, the fused feature may be a combination of the user portrait to be recommended and context information.

[0055] As an optional embodiment, among multiple preset services, selecting a recommended service that meets the fusion feature includes: based on the portrait of the user to be recommended in the fusion feature, selecting at least one candidate service that meets the portrait of the user to be recommended among the multiple preset services, wherein the candidate service is a preset service whose feature label meets the portrait of the user to be recommended; identifying service attributes of each candidate service, wherein the service attributes include: a first attribute indicating that the candidate service has a strong association with the user to be recommended, and a second attribute indicating that the candidate service has a weak association with the user to be recommended; taking the candidate service whose service attribute is the first attribute as the first recommended service, wherein the first recommended service uses a first push method to recommend to the user to be recommended, and the first push method at least includes: sending a push message to the user to be recommended; taking the candidate service whose service attribute is the second attribute as the second recommended service, wherein the second recommended service uses a second push method to recommend to the user to be recommended, and the second push method at least includes: adding the second recommended service to the recommendation list of the user to be recommended for display.

[0056] In the above-mentioned embodiment of the present application, the preset service may include a first recommended service that needs to be used by the user to be recommended first, and a second recommended service that is not required to be used by the user to be recommended first. Therefore, when selecting a recommended service for the user to be recommended from multiple preset services, at least one candidate service that meets the profile of the user to be recommended can be selected from the multiple preset services first, and then the service attributes of each candidate service can be identified, and the candidate service that has a strong correlation with the user to be recommended can be used as the first recommended service, and the candidate service that has a weak correlation with the user to be recommended can be used as the second recommended service. Then, the first recommended service and the second recommended service can be pushed to the user to be recommended in different ways.

[0057] Optionally, the first recommended service may be a service that provides government affairs, such as retirement benefit certification and other related matters for the user to be recommended.

[0058] Optionally, the first recommendation service may be pushed to the user to be recommended via SMS or APP push messages, ensuring that the user to be recommended can obtain the push service first.

[0059] Optionally, the second recommendation service can be displayed in a recommendation list within the APP, and the user to be recommended needs to actively check the recommendation list to obtain push messages from the second recommendation service, thereby preventing the push messages from the second recommendation service from harassing the user to be recommended.

[0060] As an optional embodiment, after selecting at least one candidate service that matches the user portrait to be recommended from multiple preset services based on the user portrait to be recommended in the fused features, the method also includes: based on the context information in the fused features, selecting at least one push service that matches the context information from at least one candidate service, wherein the push service is a candidate service whose feature tag matches the context information.

[0061] In the above-mentioned embodiments of the present application, at least one candidate service can be obtained after preliminary screening based on the portrait of the user to be recommended. After obtaining at least one candidate service, it is necessary to screen at least one candidate service again based on situational information to obtain at least one push service for the user to be recommended, and then push the push service to the user to be recommended based on the screened push service.

[0062] Optionally, the push service may be obtained by screening from at least one candidate service based on a context label of the context information.

[0063] As an optional embodiment, after the candidate service whose service attribute is the second attribute is used as the second recommended service, the method also includes: based on the context information in the fused feature, detecting whether the feature label corresponding to the second recommended service is consistent with the context information; when the feature label corresponding to the second recommended service is consistent with the context information, improving the arrangement order of the second recommended service in the recommendation list; when the feature label corresponding to the second recommended service is not consistent with the context information, lowering the arrangement order of the second recommended service in the recommendation list.

[0064] In the above-mentioned embodiment of the present application, for the second recommendation service that has a weak correlation with the user to be recommended, the context information can be further used for screening, thereby improving the arrangement order of the second recommendation service that meets the context information in the recommendation list, and reducing the arrangement order of the second recommendation service that does not meet the context information in the recommendation list, so that the user can get the second recommendation service that meets the context information first, thereby improving the accuracy of the push service.

[0065] As an optional embodiment, when there are multiple context information in the fusion feature, the method also includes: detecting the feature tag corresponding to the second recommendation service with each context information in turn to obtain the number of information matches for the feature tag that matches the context information; determining the arrangement order of the second recommendation service in the recommendation list according to the number of information matches, wherein the larger the number of information matches for the feature tag corresponding to the second recommendation service, the higher the arrangement order of the second recommendation service in the recommendation list.

[0066] In the above-mentioned embodiment of the present application, each user to be recommended may have multiple situational information, so there may also be multiple situational information in the fusion features of the user to be recommended. Furthermore, when providing the second recommendation service to the user to be recommended, each second recommendation service may also have multiple preset feature tags. All the situational information of the user to be recommended can be matched and detected with all the feature tags of each second recommendation service in turn to obtain the number of information matches between the feature tags and the situational information in each second recommendation service. Then, the multiple second recommendation services are arranged in descending order according to their corresponding number of information matches to obtain a recommendation list, so that the user can get the second recommendation service with the largest number of matching information matches first, thereby improving the accuracy of the push service.

[0067] As an optional embodiment, after the candidate service whose service attribute is the second attribute is used as the second recommended service, the method also includes: collecting emotional feedback information of the user to be recommended on the second recommended service, wherein the emotional feedback information is determined at least based on the physiological data of the user to be recommended, or the behavioral data of the user to be recommended, and the physiological data at least includes: the heart rate, blood pressure and sleep quality of the user to be recommended, and the behavioral data at least includes: the facial expression and voice information of the user to be recommended; adjusting the arrangement order of the second recommended service in the recommendation list according to the emotional feedback information, wherein when the emotional feedback information is positive, the arrangement order of the second recommended service in the recommendation list is improved, and when the emotional feedback information is negative, the arrangement order of the second recommended service in the recommendation list is reduced.

[0068] In the above-mentioned embodiment of the present application, after the candidate service whose service attribute is the second attribute is taken as the second recommended service, the emotional feedback information of the user to be recommended on the second recommended service is collected to obtain the preference of the recommended object for the second recommended service, thereby improving the arrangement order of the second recommended service with positive emotional feedback information in the recommendation list, and reducing the arrangement order of the second recommended service with negative emotional feedback information in the recommendation list, so that the user can preferentially obtain the second recommended service with the largest number of matching information, thereby improving the accuracy of the push service.

[0069] The present invention also provides a preferred embodiment, which provides a precise recommendation method integrating government affairs and comprehensive services for the elderly. By maintaining the elements of government service items, collecting data related to the elderly, and building a portrait model for elderly users, precise recommendations of government affairs and comprehensive services for the elderly can be achieved.

[0070] As an optional example, maintenance of government service item elements at least includes: obtaining item information and item management.

[0071] Optionally, obtaining the matter information includes: regularly obtaining shared and decentralized matter information from a preset matter management system, and generating a matter change log record as a matter data source for subsequent business applications.

[0072] Optionally, matter management includes: management of matter information including association between major and minor matters, matter application information, matter review, matter release, matter version and other management.

[0073] As an optional example, the collection of elderly-related data at least includes: data collection, data cleaning, data conversion, and data loading.

[0074] Optionally, data collection includes: according to the needs of government affairs recommendation, completing the collection of various types of data related to the portrait of the elderly, aggregating basic data of the elderly, ability assessment data, health data of the elderly, elderly service data, government service data, etc., and storing the data in a unified manner at the interface layer.

[0075] Optionally, the elderly portrait database is designed to use ID card numbers as identifiers and supports the aggregation of data from the following three data sources:

[0076] 1) Data from the big data platform: such as basic population information, health data of the elderly, etc.

[0077] 2) Batch data obtained through correspondence and other means: such as the list of elderly people receiving subsistence allowances and elderly people in extreme poverty provided through EXCLE and other forms.

[0078] 3) Data generated by business systems: such as data on elderly people’s ability assessment, risk assessment, and elderly care services.

[0079] Optionally, data collection methods include: full extraction and incremental extraction, wherein full extraction is used when data is initialized to extract the data in the data source from the database intactly; incremental extraction uses incremental extraction based on timestamp filtering or database transaction logs.

[0080] Optionally, for data cleaning, users filter out data that does not meet the requirements and confirm whether it has been corrected by the business unit before extraction. Among them, data that does not meet the requirements mainly includes three categories: incomplete data, erroneous data and duplicate data.

[0081] Optionally, incomplete data: due to the lack of some necessary information or incomplete association and carrying relationships of the data, such as the ID number and name of the population, it is necessary to complete it within the specified time and extract it again.

[0082] Optionally, erroneous data: due to an imperfect system or human error, when erroneous data is input, it is directly written into the database without judgment, or the impact range analysis of the project cutover is not comprehensive, etc. It needs to be corrected within the specified time and extracted again.

[0083] Optionally, duplicate data: is caused by deficiencies in the system's data model, defects in the primary key or constraint, or errors in the data extraction process. It is necessary to determine whether it is an error in the extraction process or an error in the source system. If the extraction process is wrong, re-extract directly. If the source system is wrong, it needs to be corrected within the specified time and extracted again.

[0084] Optionally, data conversion is used to generate new data from the data source according to certain conversion rules and store it in the destination data source. Among them, the data collection of elderly data involves data from multiple data sources, and different data sources have different data structure definitions. Therefore, it is necessary to perform data conversion processing on the collected data, such as standardizing the data format, standardizing the data naming, standardizing the data encoding, and standardizing the data identification.

[0085] Optionally, data conversion includes: format conversion, data translation, data derivation, simple data aggregation, etc. In most cases, the main conversion between the data source and the information resource base is format conversion, data translation, and data derivation, while complex data aggregation and other complex calculations mainly appear when the data is summarized.

[0086] Optionally, the data conversion function supports data conversion between different source systems, supports different data source system platforms, and supports data definition, data structure and conversion processing of erroneous data.

[0087] Optionally, data loading is used to load the extracted, cleaned and converted data into the information resource base library, including data row loading and data block loading.

[0088] Optionally, the data loading cycle and data loading strategy are determined based on comprehensive consideration of factors such as efficiency and business implementation.

[0089] It should be noted that the loading cycle should take into account the business analysis needs and the cost of system loading, and different loading cycles should be used for data of different business systems, but the real-time and integrity of business data should be maintained as much as possible at the same time, so as to establish a population portrait database that includes basic data of the elderly, ability assessment data, health data of the elderly, elderly service data, government service data, and comprehensive elderly care service data.

[0090] As an optional example, the elderly user portrait model is used to construct a proactive government service recommendation type portrait based on the smallest granularity of government service items based on the application conditions and service scope of elderly-related government affairs / services and the elderly's archival information.

[0091] Optionally, the archival information of the elderly includes: basic information of the elderly (age, gender, registered address, etc.), classification of the elderly (living alone, empty nester, single parent, etc.), physical condition (disability, disability, etc.) and other field information.

[0092] Optionally, when the information data of the elderly changes, the portrait label model will be automatically matched to generate a unique label portrait for the elderly, forming a dynamically updated label portrait system for the elderly.

[0093] Optionally, it supports users to add new tags and generate calculation logic through various methods such as drag-and-drop configuration, NLP method, batch import of tag data, etc. to guide back-end tag generation; it supports users to delete, modify, copy, query, publish and other management operations on existing tag configuration data; it supports users to query tag configuration applications initiated by themselves and tag configuration applications that require approval.

[0094] Optionally, users can withdraw applications that have been initiated but not approved, or review applications that have been initiated for approval. If the application fails the review, it will be returned to the applicant for revision.

[0095] As an optional example, the recommendation of comprehensive government affairs and elderly care services is used to match corresponding government affairs for different types of elderly users through the data association module after building a user portrait model, and make accurate recommendations.

[0096] Optionally, government affairs can be divided into strongly related matters and weakly related matters according to business needs. Strong correlation refers to matters that must be handled next, and weak correlation refers to matters that are recommended to be handled next. Strongly related matters can be actively reminded by outbound calls or mini-program message notifications; weakly related matters are displayed on the mini-program, and when the elderly use the mini-program to query government affairs, they can view the government affairs recommended by the system. At the same time, other related government affairs matters after the handling of the government affairs matter can be configured, and welfare matters can be associated. For example, when it is recognized that the elderly have handled retirement and obtained a retirement identification, matters related to retirement benefits certification can be recommended and displayed on the mini-program.

[0097] In order to ensure the accuracy and timeliness of government affairs recommendation, the technical solution provided by this application needs to timely update user portraits and re-associate affairs according to the handling of affairs by the elderly. By constructing a cyclic user portrait iteration and precise service implementation mechanism of "acquisition of elderly-related data - data analysis and processing - portrait model construction - association of government affairs / elderly care needs - government affairs / comprehensive service push - result feedback - elderly-related data update - user portrait correction - government affairs / comprehensive service push", the elderly can know what they should know about government services, enjoy all the welfare services they should enjoy, and obtain the comprehensive elderly care services they need.

[0098] As an optional example, when recommending service items based on user portraits, accurate recommendations based on contextual awareness and dynamic portrait construction can be made, including at least: feature fusion and similarity calculation driven by contextual awareness, and portrait matching based on cross-domain knowledge fusion.

[0099] Optionally, the context-aware driven feature fusion and similarity calculation includes at least: context feature extraction and fusion and dynamic weight adaptive similarity calculation.

[0100] Optionally, context feature extraction and fusion are used to introduce context-aware factors on the basis of traditional elderly care government portrait features, such as the impact of seasonal changes on the health needs and elderly care service preferences of the elderly, changes in elderly care service attention caused by social hot events, etc. For example, in winter, the elderly may have an increased demand for elderly care institutions with complete heating facilities. The system extracts the "winter heating demand" context feature by monitoring seasonal data and analyzing the historical selection behavior of the elderly in the same period, and integrates it into the elderly portrait.

[0101] In the above embodiment of the present application, a deep learning algorithm is used to fuse multi-source context data (including weather data, social news data, community activity data, etc.) with traditional portrait data to generate a fusion feature vector. For example, a convolutional neural network is used to jointly analyze the health lecture activity information organized by the community (such as lecture topics, number of participants, feedback from the elderly, etc.) and the health file data of the elderly to discover the potential demand characteristics of the elderly for health knowledge acquisition and medical services in a specific community activity context.

[0102] Optionally, dynamic weight adaptive similarity calculation is used to construct a similarity calculation model based on reinforcement learning, and dynamically adjust the weight of each feature in the similarity calculation according to the behavioral feedback of the elderly in different situations.

[0103] For example, when the elderly frequently inquire about health care services during the peak influenza season, the system automatically increases the similarity weight of health-related characteristics (such as vaccination status, knowledge of disease prevention, etc.) when matching with elderly care services or services.

[0104] The above embodiment of the present application uses transfer learning technology to migrate the similarity calculation model parameters trained in similar situations to the current situation, speeding up the model adaptation speed and improving the accuracy. For example, the similarity calculation model parameters optimized in the winter elderly care service recommendation in a certain area are migrated to other areas with similar climates, and fine-tuned in combination with local special situational factors (such as the impact of local winter customs on elderly care services).

[0105] Optionally, the cross-domain knowledge fusion portrait matching includes at least: multi-domain knowledge graph construction and fusion and recommendation reasoning based on the fused knowledge graph.

[0106] Optionally, multi-domain knowledge graphs are constructed and integrated to construct and integrate multi-domain knowledge graphs such as elderly care government affairs and medical care, social networking, and finance.

[0107] For example, in the medical field knowledge graph, the disease information of the elderly is associated with medical resources (such as hospital departments, doctor expertise, drug information, etc.); in the social field knowledge graph, the social relationship network of the elderly is connected with retirement community activities, volunteer services, etc.; in the financial field knowledge graph, the economic status of the elderly is mapped with retirement financial products, insurance services, etc.

[0108] The above embodiment of the present application realizes the integration of entities and relationships in knowledge graphs of different fields through semantic mapping and entity alignment technology. For example, the "pension subsidy service" entity in the pension government affairs knowledge graph is semantically associated and aligned with the "pension financial product preferential service" entity in the financial field knowledge graph, so that multi-field factors can be comprehensively considered when matching portraits.

[0109] Optionally, recommendation reasoning based on the fused knowledge graph is used to develop a recommendation reasoning algorithm based on a graph neural network, perform path search and relationship reasoning on the fused knowledge graph, and recommend cross-domain comprehensive elderly care services and service combinations for the elderly.

[0110] For example, when an elderly person suffers from a chronic disease and is financially well-off, the system uses knowledge graph reasoning to recommend a comprehensive plan that includes professional medical care services, living in a high-end retirement community, and a suitable retirement financial investment portfolio, thereby achieving all-round and accurate matching of retirement services.

[0111] The technical solution provided in this application utilizes multimodal feedback data fusion and in-depth analysis to integrate biometric feedback data.

[0112] Optionally, biometric feedback data of the elderly is integrated, such as physiological data such as heart rate, blood pressure, sleep quality, etc. collected through wearable devices, as well as non-physiological biometric data such as facial expression recognition and voice emotion analysis.

[0113] For example, facial expression recognition technology can be used to analyze the emotional state of the elderly when they are receiving elderly care services, and combined with voice emotion analysis to determine the elderly’s satisfaction with the services and potential needs.

[0114] Optionally, the biometric feedback data can be integrated and analyzed with traditional feedback data (such as service evaluation, user experience, etc.) to explore deeper service needs and problems.

[0115] For example, when the elderly are receiving rehabilitation services, their heart rate fluctuates greatly and their voice reveals anxiety. Combined with the service evaluation data, the system can infer that the intensity or method of rehabilitation services may need to be adjusted, and further analyze the reasons, such as the communication method of the rehabilitation therapist and the comfort level of the rehabilitation equipment.

[0116] The technical solution provided in this application can also be based on the sharing and analysis of feedback data of federated learning.

[0117] Optionally, federated learning technology can be used to achieve feedback data sharing and joint analysis among different elderly care service agencies and platforms, thereby expanding the scale and diversity of feedback data while protecting data privacy.

[0118] For example, multiple retirement communities, medical institutions, and elderly care service platforms form a federated learning alliance to share feedback data from the elderly, but the data is always retained locally, and only model parameter update information is shared. Through joint training and analysis models, more universal and targeted elderly care service optimization strategies can be discovered.

[0119] Optionally, a personalized federated averaging algorithm in federated learning is utilized to generate personalized feedback analysis results and service recommendation adjustment strategies for each elderly person based on the shared model.

[0120] For example, different service improvement suggestions can be customized for different elderly people based on factors such as the elderly’s region, age, and health status. For example, more convenient home medical services can be recommended for elderly people in mountainous areas, and more diverse social elderly care activities can be recommended for elderly people in cities.

[0121] The technical solution provided in this application can simulate and optimize elderly care services based on generative adversarial networks through adaptive dynamic adjustment and personalized customization.

[0122] Optionally, the generative adversarial network (GAN) technology is used to generate simulated elderly care service scenarios based on the portraits and feedback data of the elderly, and the optimal elderly care service configuration plan is found by continuously optimizing the generator and discriminator.

[0123] For example, the generator generates different nursing home layouts, service processes, and staffing plans, and the discriminator evaluates them based on the needs and satisfaction standards of the elderly. Through multiple iterative training, it obtains the nursing service simulation plan that best meets the personalized needs of the elderly, and applies it to actual service recommendations and institutional operation optimization.

[0124] Optionally, in the formulation of elderly care services, GAN is used to simulate the benefits and social impacts of the elderly under different service parameters (such as subsidy amounts, application conditions, etc.), providing decision support and optimization suggestions for service makers.

[0125] For example, by simulating the improvement in the quality of life of elderly people with different income levels brought about by different pension subsidy services, a more precise and fair pension service adjustment plan can be formulated.

[0126] The technical solution provided in this application can dynamically customize and update personalized service packages.

[0127] Optionally, personalized elderly care service packages can be dynamically customized and updated in real time based on the elderly's real-time feedback, status changes and preference adjustments.

[0128] For example, when the interests and hobbies of the elderly change from calligraphy and painting to music and dance, the system automatically adjusts their cultural and entertainment service packages, adds recommendations for music courses and dance activities, and promptly updates the package content and service arrangements.

[0129] Optionally, blockchain technology can be used to record and trace the customization and update process of personalized service packages to ensure the transparency and traceability of services. The elderly and their families can check the customization basis, update history and service provision of service packages at any time to enhance trust and satisfaction. At the same time, blockchain technology can also be used for service quality certification and data security protection to ensure the authenticity and privacy of data in the elderly care service process.

[0130] Figure 2 is a schematic diagram of a precise recommendation method integrating elderly care government affairs and comprehensive elderly care services according to an embodiment of the present invention. Figure 2 As shown, the method comprises the following steps:

[0131] Step S21, data processing, is used to collect user data of the elderly, which includes: basic data, ability assessment data, health data of the elderly, elderly service data and government service data, etc.

[0132] Step S22, data processing, is used to perform data cleaning, data conversion, and data loading on the collected user data.

[0133] Step S23, label establishment, adding labels to user data through NLP method, drag-and-drop configuration, and batch import.

[0134] Step S24, portrait generation, is used to use the user portrait model to generate a user portrait of each elderly person according to the tagged user data.

[0135] Step S25, matter recommendation, is used to make related recommendations of government affairs matters based on the constructed user portrait.

[0136] The above-mentioned embodiments of the present application improve the efficiency of resource allocation for elderly care services. Through accurate recommendations based on elderly care government portraits, the needs of the elderly can be accurately matched with elderly care service resources (including elderly care institutions, community elderly care service facilities, medical resources, elderly care services, etc.). The waste of elderly care service resources is avoided, so that resources can accurately flow to the elderly groups that need them most. For example, limited medical care resources are allocated preferentially to severely disabled elderly people, which improves the efficiency of medical resource utilization and ensures that the elderly can obtain timely and appropriate services. It helps elderly care service agencies to better understand the needs of potential customers and optimize service content and operation strategies. For example, elderly care institutions can adjust the configuration of service facilities, personnel training direction, etc. according to the customer characteristics brought by the recommendation system, improve their own competitiveness and service quality, and achieve supply and demand balance and optimization in the elderly care service market.

[0137] The above embodiments of the present application improve the quality of life and satisfaction of the elderly, enable the elderly to obtain elderly care services and service recommendations that meet their own characteristics and needs, and improve the pertinence and effectiveness of services. For example, appropriate rehabilitation services and medical insurance reimbursement services are recommended based on the health status and economic conditions of the elderly, so that the elderly can obtain high-quality medical rehabilitation support within an affordable range, improve their health status, and improve their quality of life. Personalized recommendations reduce the difficulties and confusion of the elderly in self-selection among many elderly care services and services, saving time and energy. The elderly can more easily find the elderly care methods and support that suit them, which enhances their satisfaction and happiness in their elderly care life.

[0138] The above-mentioned embodiments of the present application, based on the data collected by the elderly care government portrait and precise recommendation system, can deeply understand the demand characteristics and service utilization of different elderly groups, and provide strong data support for the formulation and adjustment of elderly care services. For example, by analyzing the portrait data, it is found that the demand for elderly care subsidies of a large number of low-income elderly people in a certain area is not met. The scope and standards of subsidy services are adjusted accordingly to improve the scientificity and fairness of services. It is helpful to more effectively supervise and manage the elderly care service market. By mastering the distribution of elderly care service resources and the feedback information of the elderly, problems in the market (such as substandard service quality, unreasonable prices, etc.) can be discovered in time, and corresponding measures can be taken to standardize and rectify them, protect the legitimate rights and interests of the elderly, and promote the healthy development of the elderly care service industry.

[0139] The above-mentioned embodiments of the present application enhance the synergy of social elderly care services, and promote the collaborative cooperation among multiple subjects such as medical institutions, elderly care service institutions, and social welfare organizations through accurate recommendations. It is possible to better coordinate the resources of all parties and realize the integration of medical, elderly care, social service and other resources. For example, through the recommendation system, the elderly with medical needs are connected with the surrounding medical institutions and elderly care institutions. Medical institutions can provide medical support for elderly care institutions, and elderly care institutions can provide rehabilitation and nursing services for the elderly, forming a good model of medical and elderly care integration. Social welfare organizations can carry out elderly care activities in a more targeted manner based on the recommendation information, and provide more intimate volunteer services for the elderly. At the same time, elderly care service institutions can also achieve resource complementarity and cooperation by sharing portrait information and recommendation results, jointly improve the level of elderly care services, and build a more complete and coordinated social elderly care service system.

[0140] According to an embodiment of the present invention, a service recommendation device embodiment is also provided. It should be noted that the service recommendation device can be used to execute the service recommendation method in the embodiment of the present invention, and the service recommendation method in the embodiment of the present invention can be executed in the service recommendation device.

[0141] Figure 3is a schematic diagram of a service recommendation device according to an embodiment of the present invention. Figure 3 As shown, the device may include: a collection module 32, used to collect profile information and situational information of the user to be recommended, wherein the profile information is used to represent the personal situation of the user to be recommended, and the situational information is used to represent the environmental situation of the user to be recommended and the user behavior feedback of the user to be recommended; an analysis module 34, used to use a preset user portrait model to analyze the profile information to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; a fusion module 36, used to fuse the user portrait of the user to be recommended and the situational information to obtain a fusion feature of the user to be recommended; a selection module 38, used to select a recommended service that meets the fusion feature from multiple preset services, wherein each preset service has a corresponding feature label, and the recommended service is a preset service whose feature label meets the fusion feature.

[0142] It should be noted that the acquisition module 32 in this embodiment can be used to execute step S102 in the embodiment of the present application, the analysis module 34 in this embodiment can be used to execute step S104 in the embodiment of the present application, the fusion module 36 in this embodiment can be used to execute step S106 in the embodiment of the present application, and the selection module 38 in this embodiment can be used to execute step S108 in the embodiment of the present application. The examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the contents disclosed in the above embodiments.

[0143] In an embodiment of the present invention, the profile information and situational information of the user to be recommended are collected, wherein the profile information is used to represent the personal situation of the user to be recommended, and the situational information is used to represent the environmental situation of the user to be recommended and the user behavior fed back by the user to be recommended; the profile information is analyzed using a preset user portrait model to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; the user portrait and situational information of the user to be recommended are fused to obtain a fusion feature of the user to be recommended; from multiple preset services, a recommended service that meets the fusion feature is selected, wherein each preset service has a corresponding feature label, and the recommended service is a preset service whose feature label meets the fusion feature, so that by fusing the user portrait and the user's situational information, and recommending the service based on the fused fusion feature, a technical effect of accurately recommending the services required by the user is achieved, thereby solving the technical problem of being unable to accurately recommend the services required by the user.

[0144] As an optional embodiment, the fusion module includes: a combination unit, which uses a preset feature fusion model to analyze the context information of the user to be recommended, obtain a context label, and combine the user portrait to be recommended and the context label to obtain a fusion feature, wherein the preset feature fusion model is used to identify the context type of the context information, and extract the context label from the context information based on the feature extraction strategy corresponding to the context type.

[0145] As an optional embodiment, the selection module includes: a first selection unit, used to select at least one candidate service that meets the profile of the user to be recommended from multiple preset services based on the profile of the user to be recommended in the fusion features, wherein the candidate service is a preset service whose feature label meets the profile of the user to be recommended; an identification unit, used to identify the service attributes of each candidate service, wherein the service attributes include: a first attribute indicating that the candidate service has a strong association with the user to be recommended, and a second attribute indicating that the candidate service has a weak association with the user to be recommended; a first determination unit, used to use the candidate service whose service attribute is the first attribute as the first recommended service, wherein the first recommended service uses a first push method to recommend to the user to be recommended, and the first push method at least includes: sending a push message to the user to be recommended; a second determination unit, used to use the candidate service whose service attribute is the second attribute as the second recommended service, wherein the second recommended service uses a second push method to recommend to the user to be recommended, and the second push method at least includes: adding the second recommended service to the recommendation list of the user to be recommended for display.

[0146] As an optional embodiment, the device also includes: a second selection unit, which is used to select at least one candidate service that meets the user portrait to be recommended from multiple preset services based on the user portrait to be recommended in the fused features, and then, based on the context information in the fused features, select at least one push service that meets the context information from at least one candidate service, wherein the push service is a candidate service whose feature tag meets the context information.

[0147] As an optional embodiment, the device also includes: a first detection unit, which is used to detect whether the feature label corresponding to the second recommended service conforms to the scenario information based on the scenario information in the fused feature after the candidate service whose service attribute is the second attribute is used as the second recommended service; a first sorting unit, which is used to improve the arrangement order of the second recommended service in the recommendation list when the feature label corresponding to the second recommended service conforms to the scenario information; and a second sorting unit, which is used to lower the arrangement order of the second recommended service in the recommendation list when the feature label corresponding to the second recommended service does not conform to the scenario information.

[0148] As an optional embodiment, when there are multiple pieces of context information in the fused features, the device also includes: a second detection unit, used to detect the feature label corresponding to the second recommendation service with each piece of context information in turn, to obtain the number of information matches that the feature label matches the context information; a third determination unit, used to determine the arrangement order of the second recommendation service in the recommendation list according to the number of information matches, wherein the larger the number of information matches for the feature label corresponding to the second recommendation service, the higher the arrangement order of the second recommendation service in the recommendation list.

[0149] As an optional embodiment, the device also includes: a collection unit, which is used to collect emotional feedback information of the user to be recommended on the second recommended service after the candidate service whose service attribute is the second attribute is used as the second recommended service, wherein the emotional feedback information is determined at least based on the physiological data of the user to be recommended, or the behavioral data of the user to be recommended, and the physiological data at least includes: the heart rate, blood pressure and sleep quality of the user to be recommended, and the behavioral data at least includes: facial expressions and voice information of the user to be recommended; a third sorting unit, which is used to adjust the arrangement order of the second recommended service in the recommendation list according to the emotional feedback information, wherein when the emotional feedback information is positive, the arrangement order of the second recommended service in the recommendation list is improved, and when the emotional feedback information is negative, the arrangement order of the second recommended service in the recommendation list is reduced.

[0150] An embodiment of the present invention may provide an electronic device, which may be a computer terminal, and the computer terminal may be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the computer terminal may also be replaced by a terminal device such as a mobile terminal.

[0151] Optionally, in this embodiment, the computer terminal may be located in at least one network device among a plurality of network devices of the computer network.

[0152] In this embodiment, the above-mentioned computer terminal can execute the program code of the following steps in the service recommendation method: collecting the profile information and situational information of the user to be recommended, wherein the profile information is used to represent the personal situation of the user to be recommended, and the situational information is used to represent the environmental situation of the user to be recommended and the user behavior feedback of the user to be recommended; using a preset user portrait model to analyze the profile information to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; fusing the user portrait and situational information of the user to be recommended to obtain a fusion feature of the user to be recommended; and selecting a recommended service that meets the fusion feature from multiple preset services, wherein each preset service has a corresponding feature label, and the recommended service is a preset service whose feature label meets the fusion feature.

[0153] Figure 4 is a structural block diagram of a computer terminal according to an embodiment of the present invention. Figure 4 As shown, the computer terminal 40 may include: one or more (only one is shown in the figure) processors 42 , and a memory 44 .

[0154] Among them, the memory can be used to store software programs and modules, such as program instructions / modules corresponding to the service recommendation method and device in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned service recommendation method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0155] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: collect the profile information and situational information of the user to be recommended, wherein the profile information is used to represent the personal situation of the user to be recommended, and the situational information is used to represent the environmental situation of the user to be recommended and the user behavior feedback of the user to be recommended; use a preset user portrait model to analyze the profile information to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; fuse the user portrait and situational information of the user to be recommended to obtain a fusion feature of the user to be recommended; and select a recommended service that meets the fusion feature from multiple preset services, wherein each preset service has a corresponding feature label, and the recommended service is a preset service whose feature label meets the fusion feature.

[0156] Optionally, the processor may also execute the program code of the following steps: using a preset feature fusion model to analyze the context information of the user to be recommended, obtaining a context label, and combining the user portrait to be recommended and the context label to obtain a fusion feature, wherein the preset feature fusion model is used to identify the context type of the context information, and extract the context label from the context information based on a feature extraction strategy corresponding to the context type.

[0157] Optionally, the processor may also execute the program code of the following steps: based on the portrait of the user to be recommended in the fusion features, select at least one candidate service that meets the portrait of the user to be recommended from multiple preset services, wherein the candidate service is a preset service whose feature tag meets the portrait of the user to be recommended; identify the service attributes of each candidate service, wherein the service attributes include: a first attribute indicating that the candidate service has a strong association with the user to be recommended, and a second attribute indicating that the candidate service has a weak association with the user to be recommended; use the candidate service whose service attribute is the first attribute as the first recommended service, wherein the first recommended service is recommended to the user to be recommended using a first push method, and the first push method at least includes: sending a push message to the user to be recommended; use the candidate service whose service attribute is the second attribute as the second recommended service, wherein the second recommended service is recommended to the user to be recommended using a second push method, and the second push method at least includes: adding the second recommended service to the recommendation list of the user to be recommended for display.

[0158] Optionally, the processor may also execute program code of the following steps: based on the context information in the fused feature, selecting at least one push service that matches the context information from at least one candidate service, wherein the push service is a candidate service whose feature tag matches the context information.

[0159] Optionally, the processor may also execute the program code of the following steps: based on the context information in the fused features, detecting whether the feature tag corresponding to the second recommended service is consistent with the context information; when the feature tag corresponding to the second recommended service is consistent with the context information, improving the arrangement order of the second recommended service in the recommendation list; when the feature tag corresponding to the second recommended service is not consistent with the context information, lowering the arrangement order of the second recommended service in the recommendation list.

[0160] Optionally, in the case that there are multiple context information in the fusion feature, the processor may also execute the following program code: sequentially detecting the feature tag corresponding to the second recommendation service with each context information to obtain the number of information matches for which the feature tag matches the context information; and determining the arrangement order of the second recommendation service in the recommendation list according to the number of information matches, wherein the larger the number of information matches for the feature tag corresponding to the second recommendation service, the higher the arrangement order of the second recommendation service in the recommendation list.

[0161] Optionally, the processor may also execute the program code of the following steps: collecting emotional feedback information of the user to be recommended on the second recommended service, wherein the emotional feedback information is determined at least based on the physiological data of the user to be recommended, or the behavioral data of the user to be recommended, and the physiological data at least include: the heart rate, blood pressure and sleep quality of the user to be recommended, and the behavioral data at least include: the facial expression and voice information of the user to be recommended; adjusting the arrangement order of the second recommended service in the recommendation list according to the emotional feedback information, wherein when the emotional feedback information is positive, the arrangement order of the second recommended service in the recommendation list is increased, and when the emotional feedback information is negative, the arrangement order of the second recommended service in the recommendation list is decreased.

[0162] It can be understood by those skilled in the art that Figure 4 The structure shown is for illustration only, and the computer terminal may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a handheld computer, a mobile Internet device (Mobile Internet Devices, MID), a PAD, or other terminal devices. Figure 4 The structure of the electronic device is not limited. For example, the computer terminal 40 may also include Figure 4 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 4 Different configurations shown.

[0163] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a computer program. The computer program can be stored in a non-volatile medium. The non-volatile storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0164] The embodiment of the present invention further provides a non-volatile storage medium. Optionally, in this embodiment, the non-volatile storage medium can be used to store the program code executed by the service recommendation method provided in the embodiment.

[0165] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0166] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: collecting profile information and scenario information of the user to be recommended, wherein the profile information is used to represent the personal situation of the user to be recommended, and the scenario information is used to represent the environmental situation of the user to be recommended and the user behavior feedback of the user to be recommended; using a preset user portrait model, analyzing the profile information to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; fusing the user portrait and scenario information of the user to be recommended to obtain a fusion feature of the user to be recommended; and selecting a recommended service that meets the fusion feature from multiple preset services, wherein each preset service has a corresponding feature label, and the recommended service is a preset service whose feature label meets the fusion feature.

[0167] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: using a preset feature fusion model, analyzing the context information of the user to be recommended to obtain a context label, and combining the user portrait to be recommended and the context label to obtain a fusion feature, wherein the preset feature fusion model is used to identify the context type of the context information, and extract the context label from the context information based on a feature extraction strategy corresponding to the context type.

[0168] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: based on the portrait of the user to be recommended in the fusion features, selecting at least one candidate service that matches the portrait of the user to be recommended from multiple preset services, wherein the candidate service is a preset service whose feature tag matches the portrait of the user to be recommended; identifying service attributes of each candidate service, wherein the service attributes include: a first attribute indicating that the candidate service has a strong association with the user to be recommended, and a second attribute indicating that the candidate service has a weak association with the user to be recommended; taking the candidate service whose service attribute is the first attribute as the first recommended service, wherein the first recommended service is recommended to the user to be recommended using a first push method, and the first push method at least includes: sending a push message to the user to be recommended; taking the candidate service whose service attribute is the second attribute as the second recommended service, wherein the second recommended service is recommended to the user to be recommended using a second push method, and the second push method at least includes: adding the second recommended service to the recommendation list of the user to be recommended for display.

[0169] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: based on the context information in the fused features, selecting at least one push service that matches the context information from at least one candidate service, wherein the push service is a candidate service whose feature tag matches the context information.

[0170] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: based on the context information in the fused features, detecting whether the feature tag corresponding to the second recommended service is consistent with the context information; when the feature tag corresponding to the second recommended service is consistent with the context information, improving the arrangement order of the second recommended service in the recommendation list; when the feature tag corresponding to the second recommended service is not consistent with the context information, lowering the arrangement order of the second recommended service in the recommendation list.

[0171] Optionally, in this embodiment, when there are multiple context information in the fused features, the non-volatile storage medium is configured to store program code for executing the following steps: detecting the feature tag corresponding to the second recommendation service with each context information in turn to obtain the number of information matches for which the feature tag matches the context information; determining the arrangement order of the second recommendation service in the recommendation list according to the number of information matches, wherein the larger the number of information matches for the feature tag corresponding to the second recommendation service, the higher the arrangement order of the second recommendation service in the recommendation list.

[0172] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: collecting emotional feedback information of the user to be recommended on the second recommended service, wherein the emotional feedback information is determined at least based on the physiological data of the user to be recommended, or the behavioral data of the user to be recommended, and the physiological data at least includes: the heart rate, blood pressure and sleep quality of the user to be recommended, and the behavioral data at least includes: facial expressions and voice information of the user to be recommended; adjusting the arrangement order of the second recommended service in the recommendation list according to the emotional feedback information, wherein when the emotional feedback information is positive, the arrangement order of the second recommended service in the recommendation list is increased, and when the emotional feedback information is negative, the arrangement order of the second recommended service in the recommendation list is decreased.

[0173] The embodiment of the present invention further provides a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, the steps of the service recommendation method provided in the above embodiment are implemented.

[0174] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0175] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0176] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0177] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0178] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0179] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a non-volatile storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned non-volatile storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0180] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A service recommendation method, characterized in that: include: Collecting profile information and context information of the user to be recommended, wherein the profile information is used to represent the personal situation of the user to be recommended, and the context information is used to represent the environmental situation of the user to be recommended and the user behavior reported by the user to be recommended; Using a preset user portrait model, analyzing the profile information to obtain a user portrait of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; fusing the user portrait of the user to be recommended and the context information to obtain a fusion feature of the user to be recommended; Among multiple preset services, a recommended service that meets the fusion feature is selected, wherein each of the preset services has a corresponding feature tag, and the recommended service is the preset service whose feature tag meets the fusion feature.

2. The method according to claim 1, characterized in that The user portrait of the user to be recommended and the context information are integrated to obtain the integrated features of the user to be recommended, including: Using a preset feature fusion model, the context information of the user to be recommended is analyzed to obtain a context label, and the portrait of the user to be recommended and the context label are combined to obtain the fusion feature, wherein the preset feature fusion model is used to identify the context type of the context information, and extract the context label from the context information based on a feature extraction strategy corresponding to the context type.

3. The method according to claim 1, characterized in that Among the multiple preset services, the recommended service that meets the fusion characteristics includes: Based on the user profile to be recommended in the fusion feature, selecting at least one candidate service that matches the user profile to be recommended from the plurality of preset services, wherein the candidate service is the preset service whose feature label matches the user profile to be recommended; Identifying service attributes of each of the candidate services, wherein the service attributes include: a first attribute indicating that the candidate service has a strong association relationship with the user to be recommended, and a second attribute indicating that the candidate service has a weak association relationship with the user to be recommended; The candidate service whose service attribute is the first attribute is used as a first recommended service, wherein the first recommended service is recommended to the to-be-recommended user using a first push method, and the first push method at least includes: sending a push message to the to-be-recommended user; The candidate service whose service attribute is the second attribute is used as a second recommended service, wherein the second recommended service is recommended to the to-be-recommended user using a second push method, and the second push method at least includes: adding the second recommended service to the to-be-recommended user's recommendation list for display.

4. The method according to claim 3, characterized in that After selecting at least one candidate service that meets the user profile to be recommended from the plurality of preset services based on the user profile to be recommended in the fusion feature, the method further includes: Based on the context information in the fused feature, at least one push service that meets the context information is selected from at least one of the candidate services, wherein the push service is the candidate service whose feature tag meets the context information.

5. The method according to claim 3, characterized in that: After taking the candidate service whose service attribute is the second attribute as a second recommended service, the method further includes: Based on the context information in the fused feature, detecting whether the feature label corresponding to the second recommended service conforms to the context information; When the feature tag corresponding to the second recommended service matches the context information, improving the arrangement order of the second recommended service in the recommendation list; When the feature tag corresponding to the second recommended service does not conform to the context information, the arrangement order of the second recommended service in the recommendation list is lowered.

6. The method according to claim 5, characterized in that In the case where there are multiple pieces of context information in the fusion feature, the method further includes: Detecting the feature tag corresponding to the second recommended service with each of the context information in turn to obtain the number of pieces of information matching the feature tag with the context information; According to the number of information matches, the arrangement order of the second recommended service in the recommendation list is determined, wherein the greater the number of information matches of the feature tag corresponding to the second recommended service, the higher the arrangement order of the second recommended service in the recommendation list.

7. The method according to claim 3, characterized in that After taking the candidate service whose service attribute is the second attribute as a second recommended service, the method further includes: Collecting emotional feedback information of the user to be recommended on the second recommendation service, wherein the emotional feedback information is determined at least based on physiological data of the user to be recommended or behavioral data of the user to be recommended, the physiological data at least includes: heart rate, blood pressure and sleep quality of the user to be recommended, and the behavioral data at least includes: facial expression and voice information of the user to be recommended; Adjust the arrangement order of the second recommendation service in the recommendation list according to the emotional feedback information, wherein, when the emotional feedback information is positive, the arrangement order of the second recommendation service in the recommendation list is improved, and when the emotional feedback information is negative, the arrangement order of the second recommendation service in the recommendation list is reduced.

8. A service recommendation device, characterized in that: include: A collection module, used to collect profile information and context information of the user to be recommended, wherein the profile information is used to represent the personal situation of the user to be recommended, and the context information is used to represent the environmental situation of the user to be recommended and the user behavior reported by the user to be recommended; An analysis module is used to analyze the profile information using a preset user portrait model to obtain a user portrait to be recommended of the user to be recommended, wherein the preset user portrait model is trained by machine learning using multiple sets of training data, and each set of training data in the multiple sets of training data includes: sample profile information of the same sample user, and a sample user portrait corresponding to the sample profile information; A fusion module, used for fusing the user portrait to be recommended and the context information of the user to be recommended to obtain a fusion feature of the user to be recommended; The selection module is used to select a recommended service that meets the fusion feature from a plurality of preset services, wherein each of the preset services has a corresponding feature tag, and the recommended service is the preset service whose feature tag meets the fusion feature.

9. An electronic device, comprising a memory and a processor, characterized in that: The memory stores a computer program, and the processor is configured to execute the service recommendation method according to any one of claims 1 to 7 through the computer program.

10. A computer program product comprising computer instructions, characterized in that When the computer instructions are executed by a processor, the steps of the service recommendation method described in any one of claims 1 to 7 are implemented.

Images