Information recommendation method and device, storage medium and program product

Abstract

The embodiment of the invention provides an information recommendation method and device, a storage medium and a program product. In the method, in a multi-modal recommendation model, according to service demand information and historical behavior information of a user, multi-modal demand understanding is carried out to obtain dual demand information of the user, according to multi-modal information of multiple service personnel, ability portraits of the service personnel are constructed, and the ability portraits comprise dominant skill information and recessive traits information; and according to the dual demand information and the ability portrait, performing dual demand matching between the user and the service personnel to obtain identification information of a target service personnel, and according to the identification information of the target service personnel, pushing detail information of the target service personnel to the user terminal equipment so as to recommend the target service personnel to the user. In this way, the service requirements of the user can be more accurately understood, the ability portraits of the service personnel can be more accurately constructed, the adaptation degree between the user and the service personnel is improved based on dual-requirement matching, and the service personnel are more accurately recommended to the user.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an information recommendation method, device, storage medium, and program product. Background Art

[0002] In the field of domestic services, domestic service platforms can already recommend service personnel to users, such as cleaners, nannies, blind date partners, or driving instructors. Currently, domestic service platforms maintain service tags for each service provider, which represent the service category to which they belong, such as cleaner, nanny, or driving instructor. When a user submits a request, a rules engine extracts keywords from the user's request information, matches these keywords with the service provider's tags, and recommends matching service providers to the user.

[0003] However, in actual applications, it is often the case that the service personnel recommended by the housekeeping service platform to users do not match the user's needs. That is, the housekeeping service platform is unable to recommend service personnel to users more accurately, resulting in a poor user experience. Summary of the Invention

[0004] Various aspects of the present application provide an information recommendation method, device, storage medium, and program product for more accurately recommending service personnel to users.

[0005] An embodiment of the present application provides an information recommendation method, comprising: obtaining service demand information described in natural language submitted by a user on a first application page of the user's terminal device, and obtaining historical behavior information generated by the user; obtaining multimodal information of multiple service personnel who can provide services to the user, the multimodal information reflecting the static attribute information and dynamic behavior information of the service personnel; inputting the user's service demand information, historical behavior information and the multimodal information of the multiple service personnel into a multimodal recommendation model; in the multimodal recommendation model, performing multimodal demand understanding based on the user's service demand information and the historical behavior information to obtain the user's dual demand information, the dual demand information including explicit demand information and implicit demand information; constructing a capability portrait of the service personnel based on the multimodal information of the multiple service personnel, the capability portrait including the explicit skill information and implicit trait information of the service personnel; performing dual demand matching between the user and the multiple service personnel based on the dual demand information of the user and the capability portraits of the multiple service personnel to obtain identification information of a target service personnel who is suitable for providing services to the user; and pushing detailed information of the target service personnel to a second application page on the terminal device based on the identification information of the target service personnel to recommend the target service personnel to the user.

[0006] Optionally, obtaining service demand information described in natural language and submitted by the user on the first application page includes: responding to the user's demand submission operation in any round on the application page to obtain the user's initial demand information; inputting the initial demand information into the multimodal recommendation model, performing feature extraction on the initial demand information to obtain initial features, and judging whether the initial demand information is fuzzy demand information based on the initial features; if so, predicting candidate demands of different tendencies of the user based on the initial features, generating guidance information based on the candidate demands of different tendencies, and displaying the guidance information through the first application page to guide the user to select target demands from the candidate demands of different tendencies; judging whether the target demands determined by the user in multiple rounds meet the service demand conditions, and if so, generating the service demand information based on the target demands determined by the user in multiple rounds.

[0007] Optionally, the target application to which the first application page belongs provides a personnel screening page presenting different service personnel; the historical behavior information includes: the user's historical browsing information on the personnel screening page and the historical evaluation information made for different service personnel; multimodal demand understanding is performed based on the user's service demand information and the historical behavior information to obtain the user's dual demand information, including: determining the user's explicit preference information based on the filtering tags with a click frequency higher than a first threshold and / or the page content with a stay time higher than a second threshold in the historical browsing information; generating the user's explicit demand information based on the keywords in the service demand information and the explicit preference information; performing semantic understanding on the service demand information to obtain the semantic information of the service demand information, and generating the user's implicit demand information based on the semantic information of the historical evaluation information and the semantic information of the service demand information.

[0008] Optionally, the multimodal information includes different unimodal information corresponding to different stages in the historical service process; any unimodal information is the resume text or interview voice in the interview stage, the service video in the service stage, the service effect image in the acceptance stage, or the user evaluation information in the evaluation stage; constructing the ability portrait of the service personnel based on the multimodal information of the multiple service personnel, the ability portrait includes explicit skill information and implicit trait information, including: for any service personnel, according to the text features and / or image features of the resume information of the service personnel, identifying the user basic information and skill information of the service personnel as explicit skill information; according to the audio features of the interview voice of the service personnel and / or the expression features of the service video, identifying the personality information of the service personnel; according to the image features of the service personnel's service effect image and / or the text features of the user evaluation information, identifying the ability information of the service personnel; using the personality information and ability information of the service personnel as implicit trait information.

[0009] Optionally, based on the dual demand information of the user and the capability portraits of the multiple service personnel, dual demand matching is performed between the user and the multiple service personnel to obtain identification information of a target service personnel suitable for providing services to the user, including: calculating a first similarity between the explicit demand information of the user and the explicit skill information of the multiple service personnel, and determining multiple candidate service personnel with different display orders from the multiple service personnel based on the first similarity; calculating a second similarity between the implicit demand information of the user and the implicit trait information of the multiple service personnel, and adjusting the display order and / or screening the multiple candidate service personnel with different display orders based on the second similarity to obtain identification information of the target service personnel.

[0010] Optionally, it also includes: obtaining service demand information samples of sample users, multimodal information samples of multiple sample service personnel, and target service personnel samples; the multimodal information samples of any sample service personnel are positive sample information or negative sample information; the similarity between the positive sample information and the service demand information sample is not lower than a similarity threshold, and the similarity between the negative sample information and the service demand information sample is lower than the similarity threshold; the service demand information samples, the multimodal information samples of the multiple sample service personnel, and the target service personnel samples are input into the contrastive learning network, and under the supervision of the target service personnel samples, with the goal of converging the joint loss function of the contrastive learning network to a specified range, the service demand information samples are compared with each other using the contrastive learning network. The contrastive learning network is trained based on the service demand information samples and the multimodal information samples of the multiple sample service personnel. The joint loss function includes: a main task loss function and a contrastive learning loss function. The training is stopped until the joint loss function converges to a specified range to obtain the multimodal recommendation model. The contrastive learning loss function is used to make the service demand information samples and the positive sample information close to each other, and to make the service demand information samples and the negative sample information far away. The main task loss function is used to calculate the error between the target service personnel and the target service personnel sample matched by the contrastive learning network based on the dual demands of the service demand information samples and the multimodal information samples of the multiple sample service personnel.

[0011] Optionally, the multimodal recommendation model pre-learns the dual demand information of different users; after pushing the detailed information of the target service personnel to the second application page on the terminal device, it also includes: determining the target dual demand information other than the dual demand information of the user from the dual demand information of the different users; the target dual demand information includes: target explicit demand information and target implicit demand information; based on the target dual demand information, a demand query statement is generated, and sent to the user through the first application page to determine whether the target dual demand information exists; if so, dual demand matching is performed between the user and the multiple service personnel based on the user's target dual demand information and the capability portraits of the multiple service personnel to determine at least one candidate service personnel from the multiple service personnel; based on the target dual demand information and the multimodal information of the at least one candidate service personnel, a reverse recommendation explanation statement is generated, and the reverse recommendation explanation statement is pushed to the second application page of the terminal device.

[0012] An embodiment of the present application also provides an electronic device, comprising: a memory and a processor; wherein the memory is used to: store one or more computer instructions; and the processor is used to execute the one or more computer instructions to: execute the steps in the information recommendation method.

[0013] An embodiment of the present application further provides a computer-readable storage medium, which, when the computer program is executed by a processor, enables the processor to implement the steps in the information recommendation method.

[0014] An embodiment of the present application further provides a computer program product, comprising a computer program / instruction. When the computer program / instruction is executed by a processor, the processor is enabled to implement the steps in the information recommendation method.

[0015] In this embodiment, in a multimodal recommendation model, a multimodal demand understanding can be performed based on the user's service demand information and historical behavior information to obtain the user's dual demand information. Based on the multimodal information of multiple service personnel, a capability profile of the service personnel can be constructed. The capability profile includes explicit skill information and implicit trait information. Based on the dual demand information and capability profile, a dual demand matching is performed between the user and the service personnel to obtain the identification information of the target service personnel. Based on the identification information of the target service personnel, the detailed information of the target service personnel is pushed to the user terminal device to recommend the target service personnel to the user. In this way, the user's service demand can be understood more accurately and the capability profile of the service personnel can be constructed. Based on the dual demand matching, the compatibility between the user and the service personnel can be improved, and service personnel can be recommended to the user more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0017] Figure 1 A flowchart of an information recommendation method provided by an exemplary embodiment of the present application;

[0018] Figure 2 A schematic diagram of an electronic device provided as an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0019] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0020] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation portals for users to choose to authorize or refuse. In addition, the various models involved in this application (including but not limited to language models or large models) are in compliance with relevant laws and standards.

[0021] In the field of domestic services, domestic service platforms can already recommend service personnel to users, such as cleaners, nannies, blind date partners, or driving instructors. Currently, domestic service platforms maintain service tags for each service provider, which represent the service category to which they belong, such as cleaner, nanny, or driving instructor. When a user submits a request, a rules engine extracts keywords from the user's request information, matches these keywords with the service provider's tags, and recommends matching service providers to the user.

[0022] However, in actual applications, due to the strong ambiguity and subjectivity of the needs input by users, it is usually impossible to accurately identify the core needs of users; on the other hand, the preset labels cannot accurately reflect the actual service capabilities of service personnel; based on the above two aspects, it is often the case that the service personnel recommended to users by the housekeeping service platform do not match the user's needs. The housekeeping service platform cannot accurately recommend service personnel to users, resulting in a poor user experience.

[0023] In response to the above technical problems, a technical solution is provided in the embodiments of the present application, which can more accurately understand the multimodal needs based on the user's service demand information and historical behavior information, and more accurately build the service personnel's ability portrait based on the service personnel's multimodal information, so that dual demand matching can be performed between the user and multiple service personnel to obtain the target service personnel and recommend them to the user. In this way, the compatibility between the user and the service personnel can be improved, and service personnel can be recommended to the user more accurately. The following is a detailed description of the technical solutions provided by the various embodiments of the present application in conjunction with the accompanying drawings.

[0024] Figure 1 The information recommendation method provided by the exemplary embodiment of this application is as follows: Figure 1 As shown, the method may include the following steps:

[0025] Step 11: Obtain service demand information described in natural language submitted by the user on the first application page of the terminal device, and obtain historical behavior information generated by the user.

[0026] Step 12: Acquire multimodal information of multiple service personnel who can provide services to the user. The multimodal information reflects the static attribute information and dynamic behavior information of the service personnel.

[0027] Step 13: Input the user's service demand information, historical behavior information, and multimodal information of multiple service personnel into the multimodal recommendation model.

[0028] Step 14. In the multimodal recommendation model, multimodal demand understanding is performed based on the user's service demand information and historical behavior information to obtain the user's dual demand information, which includes explicit demand information and implicit demand information; based on the multimodal information of multiple service personnel, a capability profile of the service personnel is constructed, which includes the service personnel's explicit skill information and implicit trait information; based on the user's dual demand information and the capability profiles of multiple service personnel, dual demand matching is performed between the user and multiple service personnel to obtain the identification information of the target service personnel suitable for providing services to the user.

[0029] Step 15: According to the identification information of the target service personnel, detailed information of the target service personnel is pushed to the second application page on the terminal device to recommend the target service personnel to the user.

[0030] The execution subject in this embodiment can be any electronic device, such as a server or a terminal device. The terminal device can be a mobile phone, a tablet computer, a computer, etc. This embodiment does not limit the specific implementation method of the electronic device.

[0031] In this embodiment, a user can submit service request information on a first application page. Service request information can be described in natural language, which refers to the language used by humans for daily communication, such as Chinese, English, or French. The first application page can be any page on the target application on the first user's terminal device. The target application can be a housekeeping service application, another instant messaging application, or any other type of application, without limitation in this embodiment. The first application page can be a chat page or any other page on the target application, without limitation in this embodiment. If the first application page is implemented as a chat page on the target application, the user can submit service request information in the form of a chat. In this embodiment, user behavior information generated when using the target application at historical moments, namely historical behavior information, can include historical browsing information on the target application's personnel screening page and historical evaluation information for different service personnel. It can also include browsing behavior information or consultation information on other application pages before submitting the service request information, without limitation in this embodiment. Other application pages can be any page, such as a customer service consultation page or homepage, without limitation in this embodiment.

[0032] In this embodiment, multimodal information can be obtained for multiple service personnel who can provide services to the user. The services here can be housekeeping services, security services, or any other services, and this embodiment does not limit this. The multimodal information can reflect the service personnel's static attribute information and dynamic behavior information. The static attribute information may include resume information and static tags, while the dynamic behavior information may include service videos and interview audio.

[0033] In this embodiment, the user's service demand information, historical behavior information, and multimodal information of multiple service personnel can be input into the multimodal recommendation model. In the multimodal recommendation model, multimodal demand understanding can be performed based on the user's service demand information and historical behavior information to obtain the user's dual demand information. Among them, the dual demand information may include explicit demand information and implicit demand information. Explicit demand information refers to the demand content that is clearly expressed by the user during the communication process and is easy to identify and understand, and implicit demand information refers to the potential demand content that the user has not directly / clearly expressed. For example, explicit demand information may be "parenting", "English teaching ability" and "first aid knowledge", and implicit demand information may be "strong education awareness" and "emotional stability".

[0034] In a multimodal recommendation model, a service provider's competency profile can be constructed based on the multimodal information of multiple service providers. This profile can include both explicit skill information and implicit trait information. Explicit skill information refers to the specific knowledge and operational capabilities clearly expressed by the service provider, such as "parenting," "English teaching ability," and "first aid knowledge." Implicit trait information refers to the service provider's inherent qualities in personality, emotional management, interpersonal communication, and values, which are not explicit but have a significant impact on service quality, such as "strong educational awareness" and "emotional stability."

[0035] It should be noted that the multimodal recommendation model can first perform multimodal demand understanding and then construct the service personnel's capability portrait, or it can first construct the service personnel's capability portrait and then perform multimodal demand understanding, or it can perform multimodal demand understanding and construct the service personnel's capability portrait at the same time. This embodiment does not impose any restrictions.

[0036] In the multimodal recommendation model, dual demand matching can be performed between the user and multiple service personnel based on the user's dual demand information and the capability portraits of multiple service personnel, so as to determine the identification information of the target service personnel from the identification information of multiple service personnel maintained in advance. The identification information can be used to identify the identity of the target service personnel, and can be an ID (Identity Document), name or name abbreviation, etc., which is not limited in this embodiment. Dual demand matching refers to the initial matching at the explicit level and the secondary matching at the implicit level. The target service personnel can be one or more service personnel, which is not limited in this embodiment.

[0037] Subsequently, detailed information about the target service personnel can be pushed to a second application page on the terminal device based on the target service personnel's identification information. The target service personnel's detailed information can be determined from pre-maintained personnel information for multiple service personnel based on the target service personnel's identification information. The detailed information may include, but is not limited to, various basic personal information, such as place of origin, name, and gender, as well as service information, such as the type of service provided, service duration, years of service, and service price, etc., although this embodiment does not impose any restrictions on this.

[0038] Among them, the second application page and the first application page mentioned above can be the same page or different pages; the second application page can be any page such as an email page or a message notification page. Optionally, when the second application page and the first application page are different pages, the second application page can be placed on the first application page for display in any display method such as a floating layer or a pop-up window, or the second application page can be displayed separately. This embodiment does not limit the specific implementation method and display method of the second application page. Optionally, when the second application page and the first application page are the same page, for example, both are chat pages, then for the user, the user can enter service demand information in the form of a chat on the chat page, and obtain detailed information of the recommended target service personnel on the chat page. In this embodiment, in a multimodal recommendation model, a multimodal demand understanding can be performed based on the user's service demand information and historical behavior information to obtain the user's dual demand information. Based on the multimodal information of multiple service personnel, a capability profile of the service personnel can be constructed. The capability profile includes explicit skill information and implicit trait information. Based on the dual demand information and capability profile, a dual demand matching is performed between the user and the service personnel to obtain the identification information of the target service personnel. Based on the identification information of the target service personnel, the detailed information of the target service personnel is pushed to the user terminal device to recommend the target service personnel to the user. In this way, the user's service demand can be understood more accurately and the capability profile of the service personnel can be constructed. Based on the dual demand matching, the compatibility between the user and the service personnel can be improved, and service personnel can be recommended to the user more accurately.

[0039] In some optional embodiments, the user may submit service demand information through multiple rounds of demand submission operations on the application page. When obtaining the service demand information described in natural language submitted by the user on the first application page, the user's initial demand information may be obtained in response to any round of demand submission operations on the application page; the initial demand information is input into the multimodal recommendation model, and the initial demand information is subjected to feature extraction to obtain initial features, and the initial demand information is judged based on the initial features to determine whether it is fuzzy demand information, that is, whether the initial demand information can directly reflect the user's service demand. If so, it can be considered that the user's expression is fuzzy, and the user's candidate demands with different tendencies can be predicted based on the initial features. Guidance information is generated based on the candidate demands with different tendencies, and the guidance information is displayed through the first application page to guide the user to select the target demand from the candidate demands with different tendencies.

[0040] For example, if a user's initial requirement is "I need a cook," and the multimodal recommendation model determines this as fuzzy, it can predict the user's candidate needs for fast cooking and a wide variety of dishes, and generate guidance information asking, "Do you value cooking speed or dish variety more?" Guided by this guidance, the user can select "a wide variety of dishes" as their target requirement.

[0041] The multimodal recommendation model can determine whether the target requirements determined by the user in multiple rounds meet the service requirement conditions. If so, service requirement information is generated based on the target requirements determined by the user in multiple rounds. Among them, the service requirement conditions can be set to any conditions according to the actual design requirements, and can be the number conditions of the target requirements, etc., which are not limited in this embodiment. For example, after determining that the target requirements determined by the user in multiple rounds meet the "number of target requirements is greater than 2", service requirement information can be generated based on the target requirements determined by the user in multiple rounds. Among them, the target requirements determined by the user in multiple rounds can be directly used as service requirement information, or context understanding can be performed on the basis of the target requirements determined by the user in multiple rounds to obtain service requirement information, and sentence generation or paragraph generation can also be performed on the basis of the target requirements determined by the user in multiple rounds to obtain service requirement information, which is not limited in this embodiment.

[0042] In this way, the user's service needs can be more efficiently mined by guiding information generation, so as to more accurately obtain the service demand information described in natural language submitted by the user on the first application page.

[0043] In some optional embodiments, the target application to which the first application page belongs provides a personnel screening page presenting different service personnel. The historical behavior information may include: the user's historical browsing information on the personnel screening page and historical evaluation information for different service personnel. Based on this, the aforementioned embodiment of "performing multimodal demand understanding based on the user's service demand information and historical behavior information to obtain the user's dual demand information" can be implemented based on the following steps:

[0044] Step R1, based on the filter tags with a click frequency higher than the first threshold and / or the page content with a dwell time higher than the second threshold in the historical browsing information, determine the user's explicit preference information. Among them, when the user browses for different service personnel on the personnel screening page, he can click at least one filter tag to filter the service personnel, and the selection of the filter tag can reflect the user's preference for the service personnel to a certain extent. Among them, the click frequency corresponding to any filter tag in the historical browsing information is higher than the first threshold, which reflects the user's preference for the tag content corresponding to the filter tag. The user's dwell time on any page content on the personnel screening page is higher than the second threshold, which reflects the user's preference for the page content. Among them, the first threshold and the second threshold can be set to any value according to actual design requirements, and this embodiment does not impose any restrictions.

[0045] Step R2: Generate the user's explicit demand information based on the keywords and explicit preference information in the service demand information. Any keyword extraction method can be used to extract keywords from the service demand information, such as the TF-IDF (TermFrequency-Inverse Document Frequency) method, or TextRank (a graph-based sorting algorithm for text), etc., which is not limited in this embodiment. Keywords and explicit preference information in the service demand information can be directly used as the user's explicit demand information, or sentences / paragraphs can be generated based on the keywords and explicit preference information in the service demand information to obtain explicit demand information, which is not limited in this embodiment.

[0046] Step R3: Perform semantic understanding on the service demand information to obtain semantic information of the service demand information, and generate the user's implicit demand information based on the semantic information of the historical evaluation information and the semantic information of the service demand information. The semantic information of the historical evaluation information and the semantic information of the service demand information may each have corresponding weights. Based on the weights corresponding to the semantic information of the historical evaluation information and the semantic information of the service demand information, a weighted calculation may be performed on the semantic information of the historical evaluation information and the semantic information of the service demand information to obtain the user's implicit demand information.

[0047] In this way, multimodal demand understanding can be performed more accurately based on the user's service demand information and historical behavior information, and the user's dual demand information can be obtained.

[0048] In some optional embodiments, the multimodal information includes different unimodal information corresponding to different stages in the historical service process; any unimodal information is the resume text or interview voice in the interview stage, the service video in the service stage, the service effect image in the acceptance stage, or the user evaluation information in the evaluation stage.

[0049] In the aforementioned embodiment, “building a capability profile of a service personnel based on multimodal information of multiple service personnel” can be based on the following steps K1 to K3:

[0050] Step K1: For any service personnel, identify the service personnel's basic user information and skill information based on the text features and / or image features of the service personnel's resume information as explicit skill information. Basic user information may include name, place of origin, and years of experience, though this embodiment does not limit this. Skill information refers to information related to the service personnel's skills, such as domestic service skills and the time period and geographical area in which domestic service is provided.

[0051] Among them, the resume information of the service personnel may include at least one of text and image. Based on this, the resume information can be feature extracted to obtain at least one of text features and image features, and based on at least one of the text features and image features, the user basic information and skill information of the service personnel can be determined from the resume information.

[0052] Step K2: Identify the personality information of the service personnel based on at least one of the audio features of the service personnel's interview voice and the facial expression features of the service video.

[0053] Audio features include spectrogram, pitch, speech rate, energy, and prosody. Spectrograms reflect the frequency components of a sound signal over time. Pitch refers to the high or low pitch of a voice and is related to a person's emotional state and personality. Speech rate can indicate whether a person is impatient or calm. The energy distribution of speech can also provide information about emotion and personality. Prosody, including changes in intonation, helps understand the speaker's intentions and attitude.

[0054] The expression features of service videos can include facial action units (AUs), expressions, and head posture. Facial action units are the basic units that describe facial muscle movement. Expressions can be categorized as happiness, sadness, anger, or surprise. Head posture refers to changes in the angle and position of the head, which helps understand an individual's focus and openness during communication. Personality information can include emotional impulsivity, extroversion, conscientiousness, or emotional stability.

[0055] Step K3: Identify the service personnel's ability information based on the image features of the service personnel's service effect images and / or the text features of the user evaluation information, and use the service personnel's personality information and ability information as implicit trait information.

[0056] For example, the user evaluation information implicitly mentions that the service staff has a strong educational awareness, and the service staff's service effect image also shows the user's English test score after the service staff provided educational services to the user. The implicit trait information obtained based on the above user evaluation information and service effect image may include "strong educational awareness" and "strong English teaching ability."

[0057] In this way, a more accurate capability profile of service personnel can be constructed based on the multimodal information of multiple service personnel.

[0058] In some optional embodiments, the aforementioned step of "matching the user's dual needs with multiple service personnel based on the user's dual needs information and the capability profiles of multiple service personnel to obtain identification information of a target service personnel suitable for providing services to the user" may be implemented based on the following steps:

[0059] Step T1: Calculate a first similarity between the user's explicit demand information and the explicit skill information of multiple service personnel, and determine multiple candidate service personnel in different display orders from the multiple service personnel based on the first similarity.

[0060] Step T2: Calculate the second similarity between the user's implicit demand information and the implicit trait information of multiple service personnel, and adjust the display order and / or screen the multiple candidate service personnel with different display orders based on the second similarity to obtain the identification information of the target service personnel. When adjusting the display order of multiple candidate service personnel with different display orders based on the second similarity, the display order of the multiple candidate service personnel can be adjusted in descending order of the second similarity and in descending order of the display order. That is, the more similar the service personnel's implicit trait information is to the user's implicit demand information, the higher the priority displayed. When screening the multiple candidate service personnel with different display orders based on the second similarity, preset screening rule information can be obtained, and the identification information of the target service personnel can be determined from the identification information of the multiple candidate service personnel based on the screening rule information.

[0061] It should be noted that, the display order of multiple candidate service personnel with different display orders can be adjusted only based on the second similarity, without screening the candidate service personnel; it is also possible to only screen the multiple candidate service personnel with different display orders based on the second similarity, without adjusting the display order; it is also possible to adjust the display order and screen the multiple candidate service personnel with different display orders based on the second similarity, and this embodiment does not impose any restrictions.

[0062] Through the above steps T1 and T2, dual demand matching can be more accurately performed between the user and multiple service personnel based on the user's dual demand information and the ability portraits of multiple service personnel to obtain the target service personnel.

[0063] The embodiments of the present application do not limit the specific type and specific training method of the multimodal recommendation model in the aforementioned embodiments. In an exemplary embodiment, the multimodal recommendation model in the aforementioned embodiments may be a model based on contrastive learning, which may be trained based on the following steps:

[0064] Step H1: Obtain a service demand information sample of a sample user, a multimodal information sample of multiple sample service personnel, and a target service personnel sample; the multimodal information sample of any sample service personnel is positive sample information or negative sample information; the similarity between the positive sample information and the service demand information sample is not lower than a similarity threshold, and the similarity between the negative sample information and the service demand information sample is lower than a similarity threshold.

[0065] In contrastive learning, positive and negative sample information are key elements in constructing contrastive loss. They help the model learn an effective representation of the data, so that similar data points (positive samples) are close to each other in the embedding space, while dissimilar data points (negative samples) are as far away as possible.

[0066] Step H2: Input the service demand information sample, the multimodal information samples of multiple sample service personnel, and the target service personnel sample into the comparative learning network.

[0067] Step H3: Under the supervision of the target service personnel sample, with the goal of converging the joint loss function of the contrastive learning network to a specified range, the contrastive learning network is trained using service demand information samples and multimodal information samples of multiple sample service personnel. The joint loss function includes: the main task loss function and the contrastive learning loss function. Training is stopped until the joint loss function converges to the specified range to obtain a multimodal recommendation model.

[0068] Among them, the contrastive learning loss function is used to make the service demand information sample and the positive sample information close, and to make the service demand information sample and the negative sample information far away; the main task loss function is used to calculate the error between the target service personnel and the target service personnel sample based on the contrastive learning network's dual demand matching of the service demand information sample and the multimodal information samples of multiple sample service personnel.

[0069] In this way, a multimodal recommendation model can be trained more efficiently based on comparative learning. The multimodal recommendation model can more accurately screen target service personnel that meet user needs from multiple service personnel.

[0070] In some optional embodiments, the multimodal recommendation model pre-learns dual demand information for different users. After pushing the target service personnel's detailed information to the second application page on the terminal device, the model can also determine other dual demand information other than the user's dual demand information from the dual demand information of different users, i.e., target dual demand information. The target dual demand information may include: target explicit demand information and target implicit demand information.

[0071] Based on the target dual needs information, a demand query statement is generated and sent to the user via the application page to determine whether the target dual needs exist. For example, the target dual needs information may include "specializing in Cantonese cuisine," "more attentive care," and "nurse at a tertiary hospital." Therefore, a demand query statement can be generated: "Do you still need a service person who specializes in Cantonese cuisine, has experience as a tertiary hospital nurse, and is good at caring?"

[0072] If a user has dual needs, the system can match the user with multiple service personnel based on their dual needs information and the ability profiles of multiple service personnel to identify at least one candidate. Based on the dual needs information and the multimodal information of the at least one candidate, a reverse recommendation explanation is generated, such as "I recommend Aunt Zhang because she specializes in Cantonese cuisine and has experience as a nurse at a tertiary hospital, which meets your implicit need for 'healthy eating'." This reverse recommendation explanation is then pushed to the second application page.

[0073] In this way, service personnel can be recommended to users more accurately by pushing reverse recommendation explanation statements, which effectively alleviates the idleness of other service personnel and avoids the situation where some service needs of users are ignored.

[0074] It should be noted that the execution entity of each step of the method provided in the above embodiment can be the same device, or the method can be executed by different devices. For example, the execution entity of steps 11 to 14 can be device A; for another example, the execution entity of steps 11 to 12 can be device A, and the execution entity of steps 13 to 14 can be device B; and so on.

[0075] In addition, some of the processes described in the above embodiments and the accompanying drawings include multiple operations that appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The sequence numbers of the operations, such as 12 and 13, are merely used to distinguish between different operations and do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel.

[0076] It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., and do not represent the order of precedence, nor do they limit "first" and "second" to different types.

[0077] Figure 2 is a structural diagram of an electronic device provided by an exemplary embodiment of the present application, and the electronic device is applicable to the information recommendation method provided by the aforementioned embodiment, such as Figure 2 As shown, the electronic device may include: a memory 201 , a processor 202 and a communication component 203 .

[0078] The memory 201 is used to store computer programs and can be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on the electronic device, contact data, phone book data, messages, pictures, videos, etc.

[0079] In some exemplary embodiments, the processor 202 is coupled to the memory 201 and is configured to execute a computer program in the memory 201 to: obtain service demand information described in a natural language submitted by a user on a first application page of the user's terminal device, and obtain historical behavior information generated by the user; obtain multimodal information of multiple service personnel who can provide services to the user, the multimodal information reflecting the static attribute information and dynamic behavior information of the service personnel; input the user's service demand information, historical behavior information and the multimodal information of the multiple service personnel into a multimodal recommendation model; in the multimodal recommendation model, perform multimodal demand recommendation based on the user's service demand information and the historical behavior information. Seeking understanding, obtaining the dual demand information of the user, the dual demand information including explicit demand information and implicit demand information; constructing the capability portrait of the service personnel according to the multimodal information of the multiple service personnel, the capability portrait including the explicit skill information and implicit trait information of the service personnel; performing dual demand matching between the user and the multiple service personnel according to the dual demand information of the user and the capability portraits of the multiple service personnel, so as to obtain the identification information of the target service personnel who is suitable for providing services to the user; according to the identification information of the target service personnel, pushing the detailed information of the target service personnel to the second application page on the terminal device, so as to recommend the target service personnel to the user.

[0080] Optionally, when the processor 202 obtains the service demand information described in natural language submitted by the user on the first application page, it is specifically used to: respond to the user's demand submission operation in any round on the application page to obtain the user's initial demand information; input the initial demand information into the multimodal recommendation model, perform feature extraction on the initial demand information to obtain initial features, and determine whether the initial demand information is fuzzy demand information based on the initial features; if yes, predict the candidate demands of different tendencies of the user based on the initial features, generate guidance information based on the candidate demands of different tendencies, and display the guidance information through the first application page to guide the user to select target demands from the candidate demands of different tendencies; determine whether the target demands determined by the user in multiple rounds meet the service demand conditions, and if yes, generate the service demand information based on the target demands determined by the user in multiple rounds.

[0081] Optionally, the target application to which the first application page belongs provides a personnel screening page presenting different service personnel; the historical behavior information includes: the user's historical browsing information on the personnel screening page and the historical evaluation information made for different service personnel; the processor 202 performs multimodal demand understanding based on the user's service demand information and the historical behavior information, and when obtaining the user's dual demand information, is specifically used to: determine the user's explicit preference information based on the filtering tags with a click frequency higher than a first threshold and / or the page content with a stay time higher than a second threshold in the historical browsing information; generate the user's explicit demand information based on the keywords in the service demand information and the explicit preference information; perform semantic understanding on the service demand information to obtain the semantic information of the service demand information, and generate the user's implicit demand information based on the semantic information of the historical evaluation information and the semantic information of the service demand information.

[0082] Optionally, the multimodal information includes different unimodal information corresponding to different stages in the historical service process; any unimodal information is the resume text or interview voice in the interview stage, the service video in the service stage, the service effect image in the acceptance stage, or the user evaluation information in the evaluation stage; the processor 202 constructs a capability portrait of the service personnel based on the multimodal information of the multiple service personnel. When the capability portrait includes explicit skill information and implicit trait information, it is specifically used to: for any service personnel, identify the user basic information and skill information of the service personnel as explicit skill information based on the text features and / or image features of the resume information of the service personnel; identify the personality information of the service personnel based on the audio features of the interview voice and / or the expression features of the service video; identify the capability information of the service personnel based on the image features of the service effect image of the service personnel and / or the text features of the user evaluation information; and use the personality information and capability information of the service personnel as implicit trait information.

[0083] Optionally, when the processor 202 performs dual demand matching between the user and the multiple service personnel based on the dual demand information of the user and the capability portraits of the multiple service personnel to obtain identification information of a target service personnel suitable for providing services to the user, it is specifically used to: calculate a first similarity between the explicit demand information of the user and the explicit skill information of the multiple service personnel, and determine multiple candidate service personnel with different display orders from the multiple service personnel based on the first similarity; calculate a second similarity between the implicit demand information of the user and the implicit trait information of the multiple service personnel, and adjust the display order and / or screen the multiple candidate service personnel with different display orders based on the second similarity to obtain the identification information of the target service personnel.

[0084] Optionally, the processor 202 is further configured to: obtain a sample of service demand information of a sample user, a plurality of multimodal information samples of sample service personnel, and a target service personnel sample; the multimodal information sample of any sample service personnel is positive sample information or negative sample information; the similarity between the positive sample information and the service demand information sample is not less than a similarity threshold, and the similarity between the negative sample information and the service demand information sample is less than the similarity threshold; input the service demand information sample, the multimodal information samples of the plurality of sample service personnel, and the target service personnel sample into a contrastive learning network, and under the supervision of the target service personnel sample, with the goal of converging the joint loss function of the contrastive learning network to a specified range, using The contrastive learning network is trained with the service demand information samples and the multimodal information samples of the multiple sample service personnel. The joint loss function includes: a main task loss function and a contrastive learning loss function. The training is stopped until the joint loss function converges to a specified range to obtain the multimodal recommendation model. The contrastive learning loss function is used to make the service demand information sample and the positive sample information close to each other, and to make the service demand information sample and the negative sample information far away. The main task loss function is used to calculate the error between the target service personnel matched by the contrastive learning network based on the dual demands of the service demand information sample and the multimodal information samples of the multiple sample service personnel and the target service personnel sample.

[0085] Optionally, the multimodal recommendation model pre-learns the dual demand information of different users; after the processor 202 pushes the detailed information of the target service personnel to the second application page on the terminal device, it is also used to: determine the target dual demand information other than the dual demand information of the user from the dual demand information of the different users; the target dual demand information includes: target explicit demand information and target implicit demand information; based on the target dual demand information, generate a demand query statement, and send it to the user through the first application page to determine whether the target dual demand information exists; if so, perform dual demand matching between the user and the multiple service personnel based on the user's target dual demand information and the capability portraits of the multiple service personnel to determine at least one candidate service personnel from the multiple service personnel; based on the target dual demand information and the multimodal information of the at least one candidate service personnel, generate a reverse recommendation explanation statement, and push the reverse recommendation explanation statement to the second application page of the terminal device.

[0086] Further, if Figure 2 As shown, the electronic device also includes: a display 204, a power supply component 205, an audio component 206 and other components. Figure 2Only some components are shown schematically, which does not mean that the electronic device only includes Figure 2 Components shown.

[0087] An embodiment of the present application further provides a computer-readable storage medium, which, when the computer program is executed by a processor, enables the processor to implement the steps in the information recommendation method.

[0088] An embodiment of the present application further provides a computer program product, comprising a computer program / instruction, which, when executed by a processor, executes the steps in the information recommendation method.

[0089] In this embodiment, in a multimodal recommendation model, a multimodal demand understanding can be performed based on the user's service demand information and historical behavior information to obtain the user's dual demand information. Based on the multimodal information of multiple service personnel, a capability profile of the service personnel can be constructed. The capability profile includes explicit skill information and implicit trait information. Based on the dual demand information and capability profile, a dual demand matching is performed between the user and the service personnel to obtain the identification information of the target service personnel. Based on the identification information of the target service personnel, the detailed information of the target service personnel is pushed to the user terminal device to recommend the target service personnel to the user. In this way, the user's service demand can be understood more accurately and the capability profile of the service personnel can be constructed. Based on the dual demand matching, the compatibility between the user and the service personnel can be improved, and service personnel can be recommended to the user more accurately.

[0090] The above-mentioned memory can be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random-access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0091] The above-mentioned communication component is configured to facilitate wired or wireless communication between the device where the communication component is located and other devices. The device where the communication component is located can access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G / LTE, 5G and other mobile communication networks, or a combination thereof. In an exemplary embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra wide band (UWB) technology, Bluetooth (BT) technology and other technologies.

[0092] The above-mentioned display includes a screen, which may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor can not only sense the boundary of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

[0093] The power supply assembly provides power to various components of the device in which the power supply assembly is located. The power supply assembly may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device in which the power supply assembly is located.

[0094] The above-mentioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC), and when the device where the audio component is located is in an operating mode, such as call mode, recording mode, and voice recognition mode, the microphone is configured to receive external audio signals. The received audio signal can be further stored in a memory or sent via a communication component. In some embodiments, the audio component also includes a speaker for outputting audio signals.

[0095] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to magnetic disk storage, compact disc read-only memory (CD-ROM), optical storage, etc.) that contain computer-usable program code.

[0096] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0097] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0099] In a typical configuration, a computing device includes one or more processors (Central Processing Unit, CPU), input / output interfaces, network interfaces, and memory.

[0100] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0101] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be used to store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0102] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0103] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. An information recommendation method, characterized in that: include: Obtaining service demand information described in natural language submitted by a user on a first application page of the user's terminal device, and obtaining historical behavior information generated by the user; Acquire multimodal information of a plurality of service personnel who can provide services to the user, wherein the multimodal information reflects static attribute information and dynamic behavior information of the service personnel; Inputting the user's service demand information, historical behavior information, and multimodal information of the multiple service personnel into a multimodal recommendation model; In the multimodal recommendation model, a multimodal demand understanding is performed based on the user's service demand information and the historical behavior information to obtain the user's dual demand information, wherein the dual demand information includes explicit demand information and implicit demand information; a capability profile of the service personnel is constructed based on the multimodal information of the multiple service personnel, wherein the capability profile includes the service personnel's explicit skill information and implicit trait information; based on the user's dual demand information and the capability profiles of the multiple service personnel, a dual demand matching is performed between the user and the multiple service personnel to obtain identification information of a target service personnel suitable for providing services to the user; According to the identification information of the target service personnel, detailed information of the target service personnel is pushed to the second application page on the terminal device to recommend the target service personnel to the user.

2. The method according to claim 1, characterized in that Obtain service demand information described in natural language submitted by the user on the first application page of the terminal device, including: Responding to any round of demand submission operation by the user on the first application page, obtaining initial demand information of the user; Inputting the initial demand information into the multimodal recommendation model, performing feature extraction on the initial demand information to obtain initial features, and determining whether the initial demand information is fuzzy demand information based on the initial features; if so, predicting candidate demands of different tendencies of the user based on the initial features, generating guidance information based on the candidate demands of different tendencies, and displaying the guidance information on the first application page to guide the user to select a target demand from the candidate demands of different tendencies; It is determined whether the target requirements determined by the user in multiple rounds meet the service requirement conditions. If yes, the service requirement information is generated according to the target requirements determined by the user in multiple rounds.

3. The method according to claim 1, characterized in that The target application to which the first application page belongs provides a personnel screening page presenting different service personnel; the historical behavior information includes: the user's historical browsing information on the personnel screening page and historical evaluation information made for different service personnel; Performing multimodal demand understanding based on the user's service demand information and the historical behavior information to obtain the user's dual demand information includes: Determining the user's explicit preference information based on the filter tags with a click frequency higher than a first threshold and / or the page content with a dwell time higher than a second threshold in the historical browsing information; generating explicit demand information of the user according to the keywords in the service demand information and the explicit preference information; The service demand information is semantically understood to obtain semantic information of the service demand information, and implicit demand information of the user is generated according to the semantic information of the historical evaluation information and the semantic information of the service demand information.

4. The method according to claim 1, wherein The multimodal information includes different unimodal information corresponding to different stages in the historical service process; any unimodal information is a resume text or interview voice in the interview stage, a service video in the service stage, a service effect image in the acceptance stage, or user evaluation information in the evaluation stage; Constructing a capability profile of the service personnel based on the multimodal information of the plurality of service personnel includes: For any service personnel, identifying the basic user information and skill information of the service personnel based on the text features and / or image features of the service personnel's resume information as explicit skill information; Identify the personality information of the service personnel based on the audio features of the service personnel's interview voice and / or the facial expression features of the service video; identify the ability information of the service personnel based on the image features of the service personnel's service effect image and / or the text features of the user evaluation information; and use the personality information and ability information of the service personnel as implicit trait information.

5. The method according to any one of claims 1 to 4, characterized in that According to the dual demand information of the user and the capability profiles of the multiple service personnel, dual demand matching is performed between the user and the multiple service personnel to obtain identification information of a target service personnel suitable for providing services to the user, including: calculating a first similarity between the explicit demand information of the user and the explicit skill information of the plurality of service personnel, and determining a plurality of candidate service personnel in different display orders from the plurality of service personnel based on the first similarity; A second similarity between the implicit demand information of the user and the implicit characteristic information of the multiple service personnel is calculated, and the display order of the multiple candidate service personnel in different display orders is adjusted and / or the candidate service personnel are screened according to the second similarity to obtain the identification information of the target service personnel.

6. The method according to any one of claims 1 to 4, characterized in that Also includes: Obtaining service demand information samples of sample users, multimodal information samples of multiple sample service personnel, and target service personnel samples; The multimodal information sample of any sample service personnel is positive sample information or negative sample information; the similarity between the positive sample information and the service demand information sample is not less than a similarity threshold, and the similarity between the negative sample information and the service demand information sample is less than the similarity threshold; Inputting the service demand information sample, the multimodal information samples of the multiple sample service personnel, and the target service personnel sample into a contrastive learning network, and under the supervision of the target service personnel sample, training the contrastive learning network using the service demand information sample and the multimodal information samples of the multiple sample service personnel with the goal of converging a joint loss function of the contrastive learning network to a specified range, wherein the joint loss function includes: a main task loss function and a contrastive learning loss function, and stopping training until the joint loss function converges to the specified range to obtain the multimodal recommendation model; Among them, the contrastive learning loss function is used to make the service demand information sample and the positive sample information close to each other, and to make the service demand information sample and the negative sample information far away; the main task loss function is used to calculate the error between the target service personnel and the target service personnel sample matched by the contrastive learning network based on the dual needs of the service demand information sample and the multimodal information samples of the multiple sample service personnel.

7. The method according to any one of claims 1 to 4, characterized in that The multimodal recommendation model learns the dual demand information of different users in advance; After pushing the detailed information of the target service personnel to the second application page on the terminal device, the method further includes: Determining target dual demand information other than the dual demand information of the user from the dual demand information of different users; the target dual demand information includes: target explicit demand information and target implicit demand information; generating a demand query statement according to the target dual demand information, and sending the statement to the user through the first application page to determine whether the target dual demand information exists; If yes, performing dual demand matching between the user and the multiple service personnel based on the user's target dual demand information and the capability profiles of the multiple service personnel, so as to determine at least one candidate service personnel from the multiple service personnel; A reverse recommendation explanation statement is generated according to the target dual demand information and the multimodal information of the at least one candidate service personnel, and the reverse recommendation explanation statement is pushed to the second application page of the terminal device.

8. An electronic device, characterized in that: include: A memory and a processor; wherein the memory is used to: store one or more computer instructions; and the processor is used to execute the one or more computer instructions to: execute the steps in the method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that When the computer program is executed by a processor, the processor is enabled to implement the steps of the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that The method comprises a computer program / instruction, which, when executed by a processor, enables the processor to implement the steps of the method according to any one of claims 1 to 7.

Images