Recommendation information determination method based on voice data and related apparatuses

Abstract

This application provides a method and related apparatus for determining recommendation information based on voice data, comprising: acquiring a recommendation request; acquiring reference voice interaction data of the user regarding reference historical recommendation information; determining the actual interaction duration of the user for each reference historical recommendation information based on the reference voice interaction data; determining the reference historical recommendation information whose actual interaction duration is greater than a first preset duration as target historical recommendation information; determining the interaction continuity of the user for each target historical recommendation information based on the target voice interaction data; determining available historical recommendation information from the target historical recommendation information based on the interaction continuity; acquiring the user's basic information; determining target recommendation tags based on the available historical recommendation information, the available voice interaction data, and the basic information; and determining target recommendation information based on the target recommendation tags. This method can improve the accuracy of recommendation information determination.

Description

Technical Field

[0001] This application belongs to the field of general data processing technology in the Internet industry, and specifically relates to a method and related apparatus for determining recommendation information based on voice data. Background Technology

[0002] Currently, most user profiles are generated based solely on the basic information users fill in when logging in. A few solutions also update user profiles based on users' browsing history. However, since users' browsing status cannot be determined while browsing, their browsing history cannot accurately reflect their preferences. Therefore, information recommendations based on user profiles determined in this way are also inaccurate. Summary of the Invention

[0003] This application provides a method and related apparatus for determining recommendation information based on voice data, so as to improve the accuracy of the determined recommendation information and make the recommended content more in line with user needs.

[0004] In a first aspect, embodiments of this application provide a method for determining recommendation information based on voice data, the method comprising:

[0005] Obtain a recommendation request, wherein the recommendation information includes voice interaction content, voice interaction products, or voice interaction services to recommend to the user;

[0006] Obtain the user's reference voice interaction data in response to the reference historical recommendation information, wherein the reference historical recommendation information is the recommendation information in which the user has voice interaction among all the recommendation information recommended to the user within a first preset time period;

[0007] The actual interaction duration of the user for each reference historical recommendation is determined based on the reference voice interaction data.

[0008] The reference historical recommendation information whose actual interaction duration is greater than a first preset duration is determined to be the target historical recommendation information, and the reference voice interaction data corresponding to the target historical recommendation information is the target voice interaction data.

[0009] The continuity of user interaction with each target historical recommendation information is determined based on the target voice interaction data;

[0010] Based on the interaction continuity, available historical recommendation information is determined from the target historical recommendation information, and the target voice interaction data corresponding to the available historical recommendation information is the available voice interaction data.

[0011] Obtain the user's basic information, which includes the user's identity information and interest preference information, and the interest preference information is associated with the type and domain of the recommendation information;

[0012] The target recommendation tag is determined based on the available historical recommendation information, the available voice interaction data, and the basic information;

[0013] Target recommendation information is determined based on the target recommendation tags.

[0014] Secondly, embodiments of this application provide a device for determining recommendation information based on voice data, the device comprising:

[0015] The first acquisition unit is used to acquire a recommendation request, the recommendation request being used to acquire recommendation information, the recommendation information including voice interaction content, voice interaction products, or voice interaction services to be recommended to the user;

[0016] The second acquisition unit is used to acquire the user's reference voice interaction data in response to the reference historical recommendation information, wherein the reference historical recommendation information is the recommendation information in which the user has voice interaction among all the recommendation information recommended to the user within a first preset time period;

[0017] The first determining unit is used to determine the actual interaction duration of the user for each reference historical recommendation based on the reference voice interaction data.

[0018] The second determining unit is used to determine the reference historical recommendation information whose actual interaction duration is greater than the first preset duration as the target historical recommendation information, and the reference voice interaction data corresponding to the target historical recommendation information as the target voice interaction data.

[0019] The third determining unit is used to determine the continuity of the user's interaction with each target historical recommendation information based on the target voice interaction data.

[0020] The fourth determining unit is used to determine available historical recommendation information from the target historical recommendation information based on the interaction continuity, wherein the target voice interaction data corresponding to the available historical recommendation information is available voice interaction data.

[0021] The third acquisition unit is used to acquire the user's basic information, which includes the user's identity information and attention preference information, and the attention preference information is associated with the type and domain to which the recommendation information belongs.

[0022] The fifth determining unit is used to determine the target recommendation tag based on the available historical recommendation information, the available voice interaction data, and the basic information;

[0023] The fourth acquisition unit is used to determine target recommendation information based on the target recommendation tags.

[0024] Thirdly, embodiments of this application provide an electronic device including a processor, a memory, a communication interface, and one or more programs, the one or more programs being stored in the memory and configured to be executed by the processor, the programs including instructions for performing the steps in the first aspect of embodiments of this application.

[0025] Fourthly, embodiments of this application provide a computer storage medium having a computer program / instructions stored thereon, which is executed by a processor to implement the steps of the method described in the first aspect above.

[0026] Fifthly, embodiments of this application provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps described in the first aspect of embodiments of this application. The computer program product may be a software installation package.

[0027] As can be seen, in this embodiment, the electronic device first obtains a recommendation request, the recommendation information including voice interaction content, voice interaction products, or voice interaction services to recommend to the user. Then, it obtains reference voice interaction data of the user regarding historical recommendation information, where the historical recommendation information consists of all recommendations made to the user within a first preset time period, in which the user has engaged in voice interaction. Next, based on the reference voice interaction data, it determines the actual interaction duration of the user for each historical recommendation. Finally, it determines the historical recommendation information whose actual interaction duration exceeds the first preset duration as the target historical recommendation information, and the reference voice interaction data corresponding to the target historical recommendation information is... The process involves several steps: first, obtaining target voice interaction data; then, determining the user's interaction continuity for each target historical recommendation based on the target voice interaction data; then, determining available historical recommendation information from the target historical recommendation information based on the interaction continuity, where the target voice interaction data corresponding to the available historical recommendation information is considered available voice interaction data; next, acquiring the user's basic information, including the user's identity information and attention preference information, where the attention preference information is associated with the type and domain of the recommendation information; then, determining target recommendation tags based on the available historical recommendation information, the available voice interaction data, and the basic information; and finally, determining target recommendation information based on the target recommendation tags. This approach improves the accuracy of the determined recommendation tags, thereby increasing the accuracy of the recommended information and making the recommended voice interaction content, products, or services more aligned with user needs. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a schematic diagram illustrating the composition of a recommendation system provided in an embodiment of this application;

[0030] Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application;

[0031] Figure 3 This is a flowchart illustrating a method for determining recommendation information based on voice data, provided in an embodiment of this application.

[0032] Figure 4 This is a schematic diagram of a recommendation page provided in an embodiment of this application;

[0033] Figure 5 This is a functional unit block diagram of a recommendation information determination device based on voice data provided in an embodiment of this application;

[0034] Figure 6 This is a block diagram of the functional units of another recommendation information determination device based on voice data provided in an embodiment of this application. Detailed Implementation

[0035] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0036] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0037] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0038] Currently, most user profiles are generated based solely on the basic information users fill in during login. A few solutions also update user profiles based on browsing history. However, since it's impossible to determine a user's browsing status, browsing history cannot accurately reflect user preferences. For example, although a user may spend a long time watching a video, it could be due to the user leaving midway, resulting in prolonged playback. Therefore, information recommendations based on user profiles determined in this way are also inaccurate.

[0039] To address the aforementioned issues, this application provides a method and related apparatus for determining recommendation information based on voice data. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0040] Please see Figure 1 , Figure 1 This is a schematic diagram illustrating the composition of a recommendation system provided in an embodiment of this application. For example... Figure 1 As shown, the recommendation system 10 includes a voice server 101 and terminal devices 102, which are communicatively connected. It should be noted that the recommendation system 10 may include multiple terminal devices 102 and voice servers 101, and each voice server 101 may refer to a server cluster. Simultaneously, one voice server 101 can establish communication connections with multiple terminal devices 102. The terminal device 102 is used to log in to user accounts and to provide users with a display interface for voice interaction services, voice interaction content, or voice interaction products. The voice server 101 is used to analyze user intent based on user voice information from the terminal devices 102 and to interact with the user via voice. When determining recommendation information, either the voice server 101 or the terminal devices 102 can determine it based on the user's interaction records.

[0041] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. For example... Figure 2 As shown, the electronic device 20 in this solution can be the one described above. Figure 1The voice server 101 can also be a terminal device 102. The electronic device can be a communication-capable device, including various handheld devices, in-vehicle devices, wearable devices, computing devices, or other processing devices connected to a wireless modem, as well as various forms of user equipment (UE), mobile station (MS), terminal device, etc. The electronic device 20 includes a processor 120, a memory 130, a communication interface 140, and one or more programs 131. The one or more programs 131 are stored in the memory 130 and configured to be executed by the processor 120. The one or more programs 131 include instructions for performing any step in the following method embodiments. In specific implementations, the processor 120 is used to perform any step executed by the electronic device in the following method embodiments, and when performing data transmission such as sending, the communication interface 140 can be selectively invoked to complete the corresponding operation.

[0042] Please see Figure 3 , Figure 3 This is a flowchart illustrating a method for determining recommendation information based on voice data, as provided in an embodiment of this application. The method includes the following steps.

[0043] S301, Get recommendation request.

[0044] The recommendation information includes voice interaction content, products, or services recommended to the user. When the electronic device in this solution is a terminal device, the method of obtaining the recommendation request may include obtaining the user's voice information or operation information, and determining the user's recommendation request based on the voice information or operation information. The voice information may be voice information containing keywords, such as voice messages containing keywords like "next" or "change one." The operation information may be manual operations by the user, such as swiping up on the terminal device's interface or pressing a volume button. When the electronic device is a voice server, the voice server obtains the recommendation request from the terminal device, and the terminal device sends an operation request to the voice server after obtaining the user's operation information. Alternatively, the terminal device sends the obtained voice information to the voice server, and the voice server parses it to determine the recommendation request.

[0045] For example Figure 4 As shown, Figure 4 This is a schematic diagram of a recommendation page provided in an embodiment of this application. Figure 4The voice interaction content shown is a voice-interactive video. Users can interact with the electronic device via voice while watching the video. The recommendation request is triggered by the user's swipe-up action. That is, when the user swipes up, the electronic device receives the user's recommendation request and displays other recommended information to the user. For example... Figure 4 After the user performs an upward swipe, the voice-interactive video A displayed on the terminal device screen will change to voice-interactive video B. It should be noted that the content of the recommended information in this solution is related to the recommended content of the application corresponding to the voice interaction function. If the application is a shopping app, the recommended information will indicate a voice-interactive product; if the application is a video playback app, the recommended voice-interactive content can be a voice-interactive video.

[0046] S302, Obtain the user's reference voice interaction data regarding the historical recommendation information.

[0047] The reference historical recommendation information refers to the recommendation information in which the user has engaged in voice interaction among all the recommendation information recommended to the user within a first preset time period. This first preset time period can be determined based on the user's frequency of use of the voice interaction function. For example, if the user uses the application corresponding to the voice interaction function frequently, the first preset time period could be one week; if the user uses the application corresponding to the voice interaction function only once in a long while, the first preset time period could be one month. Determining the recommendation information in which the user has engaged in voice interaction includes: determining whether user voice information exists in the interaction record for each recommendation information; if it exists, the recommendation information is determined to be information in which the user has engaged in voice interaction; if it does not exist, the user has determined not to have engaged in voice interaction with the recommendation information.

[0048] S303, determine the actual interaction duration of the user for each reference historical recommendation based on the reference voice interaction data.

[0049] The determination of the actual interaction duration includes: obtaining the target time for the first acquisition of user voice data in the interaction record corresponding to each reference historical recommendation; determining the time interval between the target time and the initial time when the user browses the reference historical recommendation; if the time interval is greater than a first preset value, the target time is determined as the start time; if the time interval is not greater than the first preset value, the initial time is determined as the start time; and determining the actual duration based on the start time and the time when the user exits the interaction with the reference historical recommendation. The time when the user exits the interaction includes: the time when the voice interaction script corresponding to the recommendation indicates that the interaction has ended, or the time when the user sends a recommendation request again.

[0050] S304, determine the reference historical recommendation information whose actual interaction duration is greater than the first preset duration as the target historical recommendation information.

[0051] The reference voice interaction data corresponding to the target historical recommendation information is the target voice interaction data. The first preset duration for each reference historical recommendation information can be the same, in which case the first preset duration represents that the user has engaged in voice interaction for at least a period of time, such as 20 seconds. Alternatively, the first preset duration for each reference historical recommendation information can be different, in which case the first preset duration can be determined based on the script content of the reference historical recommendation information. For example, if the script content of reference historical recommendation information A indicates that the user's second voice input time is earlier, while the script content of reference historical recommendation information B indicates that the user's second voice input time is later, then the first preset duration corresponding to reference historical recommendation information A is shorter than the first preset duration corresponding to reference historical recommendation information B. This ensures that there are no cases where the user only interacts briefly once and then exits, thus avoiding accidental clicks or recommendations that are only initially attractive but later become uninteresting.

[0052] S305, determine the continuity of the user's interaction with each target's historical recommendation information based on the target voice interaction data.

[0053] S306, determine available historical recommendation information from the target historical recommendation information based on the interaction continuity.

[0054] Interaction continuity primarily characterizes whether an interaction is interrupted when a user engages in voice interaction with a historical recommendation. The target voice interaction data corresponding to the available historical recommendation information is the available voice interaction data.

[0055] S307, Obtain the user's basic information.

[0056] The basic information includes the user's identity information and viewing preferences information. The viewing preferences information is associated with the type and domain of the recommended information. The user's identity information may be information filled in when registering an account, and the viewing preferences information may be content related to the type and domain of voice interaction provided by the current application. For example, when the application can provide voice-interactive video services, the type may be a category for voice-interactive videos, such as comedy or science popularization, and the domain may be entertainment or education.

[0057] S308, determine the target recommendation tag based on the available historical recommendation information, the available voice interaction data, and the basic information.

[0058] S309, Determine target recommendation information based on the target recommendation tag.

[0059] As can be seen, in this example, the electronic device first obtains a recommendation request, the recommendation information including voice interaction content, voice interaction products, or voice interaction services to recommend to the user. Then, it obtains reference voice interaction data from the user regarding historical recommendation information, which consists of all recommendations given to the user within a first preset time period in which the user has engaged in voice interaction. Next, based on the reference voice interaction data, it determines the actual interaction duration of the user for each historical recommendation. Finally, it determines the historical recommendation information whose actual interaction duration exceeds the first preset duration as the target historical recommendation information, and the reference voice interaction data corresponding to the target historical recommendation information is the target historical recommendation information. The process involves several steps: First, the user's voice interaction data is tagged. Then, based on this target voice interaction data, the continuity of the user's interaction with each target historical recommendation is determined. Next, available historical recommendation information is determined from the target historical recommendation information based on this interaction continuity. The target voice interaction data corresponding to the available historical recommendation information is considered available voice interaction data. Then, the user's basic information is obtained, including the user's identity information and attention preference information. The attention preference information is associated with the type and domain of the recommendation information. Then, target recommendation tags are determined based on the available historical recommendation information, the available voice interaction data, and the basic information. Finally, target recommendation information is determined based on the target recommendation tags. This approach improves the accuracy of the determined recommendation tags, thereby increasing the accuracy of the recommended information and making the recommended voice interaction content, products, or services more aligned with user needs.

[0060] In one possible instance, determining the user's interaction continuity for each target historical recommendation based on the target voice interaction data includes:

[0061] Perform the following operations on the target voice interaction data corresponding to each target's historical recommendation information:

[0062] Semantic analysis is performed on the target voice interaction data to obtain multiple semantics;

[0063] Determine the actual interval between each pair of semantics;

[0064] Determine whether there is a target interval duration in the actual interval duration that is longer than the second preset duration;

[0065] If it exists, determine whether the difference between the actual interaction duration corresponding to the target voice interaction data and the target interval duration is greater than the first preset duration;

[0066] If so, then the continuity of the user's interaction with the target historical recommendation information corresponding to the target voice interaction data is determined to be continuous;

[0067] If not, determine the number of times the target interval appears in the actual interval duration, and determine the interaction continuity of the target historical recommendation information based on the number of occurrences;

[0068] If it does not exist, then the user's interaction continuity with the target historical recommendation information corresponding to the target voice interaction data is determined to be continuous.

[0069] In determining the continuity of interaction for each target historical recommendation, the primary basis is the time interval between user responses. The second preset duration for each target voice interaction data point can be different. This second preset duration can be determined based on the user response timing within the interaction script of the target historical recommendation, which establishes the average time interval between user responses for that target voice interaction data. Then, combined with the user's historical response records, the difference between the mode of the time interval between each two semantic segments in the user's historical response records and the average time interval is determined. Based on the average time interval, the second preset duration is determined. In other words, the second preset duration is related to the interaction script of the target historical recommendation and the user's usual interaction habits. However, when the interaction script corresponding to the target historical recommendation only includes an interaction outline and does not specify the user's response timing, the second preset duration becomes a pre-set uniform duration, which is the maximum user response interval in general interactive content. Of course, this uniform first preset duration can also be related to the user's age and the difficulty of the recommended content's interaction; that is, the time users need to think about the interactive content may differ, thus affecting the corresponding first preset duration.

[0070] For example, when a user interacts with a target historical recommendation A via voice, the user replies with 5 sentences, and the actual interaction time is 10 minutes. Two of these replies have intervals of up to 5 minutes, while the first preset time is 3 minutes. Therefore, it can be determined that the 5-minute intervals are not the normal duration required for a user's responses during a voice interaction. The user might have left the interaction for other reasons, but the target historical recommendation A page remained on the homepage, creating the illusion of a long interaction time. Since the first preset time is also only 3 minutes, the reference historical recommendation A is only confirmed as the target historical recommendation when the interaction time exceeds the first preset time. This means the user's actual interaction time with the target historical recommendation A might be less than 6 minutes. In this case, the difference between the actual interaction time and the target interval is used to determine the user's true interaction time. This means that the potentially interfering 5-minute interval is removed. Since the difference of 5 minutes is greater than the first preset time, the interaction is considered continuous. Since there may be multiple target interval durations in the target voice data for a target historical recommendation information, even if the difference is less than the first preset duration, it cannot be directly determined that the interaction is discontinuous.

[0071] As can be seen, in this example, determining the continuity of interaction based on the semantic intervals when the user answers can avoid situations where the user simply stays on the page of the recommendation information for other reasons, which could lead to misjudgment of the user's preferences. This also reduces the amount of interference from historical recommendation information used to determine recommendation tags and improves the accuracy of determining recommendation tags.

[0072] In one possible instance, determining the interaction continuity of the target historical recommendation information based on the number of interactions includes: obtaining the interaction difficulty of the target historical recommendation information corresponding to the target voice interaction data; obtaining the user's age based on the identity information; determining a reference interval duration based on the interaction difficulty and the user's age; determining the predicted total duration of the target interval duration based on the number of interactions and the reference interval duration; determining the predicted interaction duration based on the actual interaction duration and the predicted total duration; determining whether the predicted interaction duration is greater than the first preset duration; if it is greater, then determining that the user's interaction continuity with the target historical recommendation information corresponding to the target voice interaction data is continuous; if it is not greater, then determining that the user's interaction continuity with the target historical recommendation information corresponding to the target voice interaction data is discontinuous.

[0073] In this case, since there may be multiple target interval durations in the target voice data for a single target historical recommendation, if the difference between the actual interaction duration and the target interval duration is less than the first preset duration, it is possible that the user may engage in in-depth voice interaction even if there are long intervals between multiple voice interactions. Furthermore, since the interaction script for this target historical recommendation may only have an outline, it is impossible to determine the depth of the user's voice interaction based on the content of the interaction. Therefore, the time the user needs to think during each voice interaction can be preliminarily determined based on the interaction difficulty of the target historical recommendation and the user's age. This time can then be used to replace the target interval duration to obtain the predicted actual user interaction duration, which can then be used to determine whether the user's interaction is continuous. For example, if the user's actual interaction duration for target historical recommendation B is 10 minutes, and the corresponding first preset duration is 3 minutes, but two of the user's five responses have an interval exceeding the second preset duration of 40 seconds (5 minutes and 3 minutes respectively), then... If, based on the interaction difficulty of target recommendation information B and the user's age, a reference interval of 20 seconds is determined, then the total predicted duration is 40 seconds, corresponding to a predicted interaction duration of 2 minutes and 40 seconds. This duration is less than the first preset duration, so the interaction is determined to be discontinuous. However, if the determined reference interval is 35 seconds, the corresponding predicted interaction duration is 3 minutes and 10 seconds, which exceeds the first preset duration. In this case, the interaction of the target historical recommendation information B is continuous and can be used to determine recommendation tags. The interaction difficulty of the target historical recommendation information can be determined based on the type and domain to which the target historical recommendation information belongs.

[0074] As can be seen in this example, when multiple target intervals exist, the predicted interaction duration is determined by comprehensively considering the interaction difficulty and the user's age, thereby judging the continuity of the user's interaction duration. This approach fully considers the actual user interaction in various possible scenarios, reduces the amount of interference from historical recommendation information used to determine recommendation tags, and improves the accuracy of recommendation tag determination.

[0075] In one possible instance, determining the target recommendation tag based on the available historical recommendation information, the available voice interaction data, and the basic information includes: determining the basic recommendation tag based on the domain to which the available historical recommendation information belongs, the type to which the available historical recommendation information belongs, and the basic information; determining the semantic type of multiple available semantics included in each available voice interaction data, wherein the semantic type includes a regular semantic type, an active semantic type, and an irrelevant semantic type, wherein the regular semantic type is used to indicate the semantics whose first relevance to the available recommendation information corresponding to the available voice information is higher than a first preset value, and the active semantic type is used to indicate the semantics whose first relevance is lower than the first preset value and whose first relevance to the available voice information is higher than a first preset value. The available recommendation information includes semantics whose second relevance to the domain of voice interaction content, voice interaction products, or voice interaction services is higher than a second preset value. The irrelevant semantic type is used to indicate semantics whose first relevance is lower than the first preset value and whose second relevance is lower than the second preset value. A first number of available semantics is determined for each semantic type corresponding to each available voice interaction data. Available historical recommendation information corresponding to the available voice interaction data is marked according to the first number. A second number of all available semantics included in each semantic type is determined. The user's speaking style is determined according to the second number. The credibility of the marking is determined according to the speaking style. The basic recommendation tag is modified according to the credibility and the marking to obtain a target recommendation tag.

[0076] Specifically, the ability to represent user preferences can be determined based on the first quantity of various semantic types included in available voice interaction data. For example, if available voice interaction data A includes 5 sentences of regular semantic types, 3 sentences of active semantic types, and 1 sentence of irrelevant semantic types, then this available voice interaction data can be determined as normal voice interaction and can be used to represent user preferences. Therefore, this available voice interaction data is marked accordingly, and it is considered that the type and domain to which available voice interaction data A belongs can represent user preferences. However, if available voice interaction data B includes 2 sentences of regular semantic types, 0 sentences of active semantic types, and 7 sentences of irrelevant semantic types, then there may be instances of users speaking randomly during the interaction. Therefore, this available voice interaction data is also marked accordingly, and it is considered that the type and domain to which available voice interaction data B belongs cannot be used to represent user preferences. Of course, when determining the target recommendation label, it is also necessary to correct the previous labeling based on the user's speaking style. For example, if among multiple available voice interaction data sets, the voice interaction data in the regular semantic type includes 50 sentences, the voice interaction data in the active semantic type includes 10 sentences, and the voice interaction data in the irrelevant semantic type includes 100 sentences, then the user's voice interaction style is considered to be rambling. Therefore, the reliability of the previously determined label is not high, meaning that the target voice interaction data B can also be used to represent user preferences. In this case, the basic recommendation label needs to be modified to obtain the target recommendation label.

[0077] As can be seen, in this example, when determining recommendation tags based on available historical recommendation information, the user's speaking style is also determined based on the user's available voice interaction data to determine the credibility of the previous tags and improve the accuracy of the recommendation tag determination.

[0078] In one possible instance, marking the available historical recommendation information corresponding to the available voice interaction data according to the first quantity includes: performing the following steps for each available historical recommendation information: determining the reference domain to which the available historical recommendation information belongs; determining whether the basic recommendation tags include reference domain tags corresponding to the reference domain; if yes, then marking the available historical recommendation information as a first mark if the first quantity of available semantics included in the irrelevant semantic type corresponding to the available historical recommendation information is greater than a third preset value; if no, then marking the available historical recommendation information as a second mark if the first quantity of available semantics included in the active semantic type corresponding to the available historical recommendation information is greater than a fourth preset value.

[0079] When tagging available historical recommendation information, it's necessary to first determine if the domain to which the available historical recommendation information belongs already belongs to a domain identified in the basic recommendation tags. If so, if the number of irrelevant semantics included in the available historical recommendation information exceeds a third preset value, a first tag is applied, indicating a decrease in the user's preference for that reference domain. If the basic recommendation tags do not include reference domain tags for that reference domain, the number of active semantic types in the available historical recommendation information is determined. If it exceeds a fourth preset value, a second tag is applied, indicating an increase in preference for that reference domain. For example, if a user already prefers interactive content in domain A during voice interaction, then the basic recommendation tags include tags for domain A. If, during this interaction, the user mostly speaks irrelevant statements regarding domain A, it can be assumed that the user's preference for domain A has decreased. However, if the user speaks very few irrelevant statements in the interactive content for domain A, it can be assumed that the user still prefers interactive content for domain A. Since the tags for domain A are already within the basic recommendation tags, there is no need to tag this part of the content again, reducing the amount of subsequent data analysis and increasing the speed of determining the target recommendation tags. If the tag for domain A was not originally in the basic recommendation tags, meaning the user did not originally prefer content in domain A, but now the user has said a lot about domain A in relation to the available historical recommendation information, then it is determined that the user is interested in domain A, and therefore needs to be tagged.

[0080] As can be seen in this example, labeling according to different situations allows for targeted analysis of only a portion of the data to understand the user's current preferences, thereby improving the accuracy of determining the target recommendation tags.

[0081] In one possible instance, modifying the base recommendation tag based on the credibility and the tag to obtain the target recommendation tag includes: determining a third number of available historical recommendation information containing the first tag and belonging to the reference domain; correcting the third number based on the credibility; if the corrected third number is greater than a fifth preset value, deleting the reference domain tag from the base recommendation tag to obtain the target recommendation tag; determining a fourth number of available historical recommendation information containing the second tag and not belonging to the reference domain; correcting the fourth number based on the credibility; if the corrected fourth number is greater than a sixth preset value, adding the reference domain tag to the base recommendation tag to obtain the target recommendation tag.

[0082] The presence of the first marker indicates a decrease in user preference. After credibility correction, if the number of first markers is still greater than the number of third markers, it means that the user has shown a decreasing trend in preference for content in that reference domain across multiple voice interactions, and the reference domain tag will be removed from the basic recommendation tags. However, if the number of second markers is greater than the sixth preset value, it is considered that the user is actively seeking information corresponding to the reference domain, and the user's willingness to interact via voice is strong. In this case, the reference domain tag, which was not originally in the basic recommendation tags, can be added.

[0083] Modifying the third or fourth quantity based on credibility includes: the credibility level is used to characterize the target quantity proportion range for each semantic type corresponding to each user's speaking style. Then, it is determined that among the available historical recommendation information belonging to the reference domain and marked with the first label, those whose first quantity proportion range does not match the target quantity proportion range are labeled as the first label, thus obtaining the corrected third or fourth quantity. For example, if a user's speaking style is to speak irrelevantly, then the target quantity proportion range indicated by the credibility level is 40%-60% for irrelevant semantic types. If the first quantity proportion of the active semantic type of available historical recommendation information A, which was originally marked with the first label and belongs to the reference domain, is 50% of the first quantity of all available voice interaction data corresponding to available historical recommendation information A, then the label of available historical recommendation information A is not the first label. Therefore, when calculating the quantity, the quantity of available historical recommendation information A is not counted, and only available historical recommendation information with an irrelevant semantic type first quantity proportion higher than the target proportion range is included in the third quantity. Similarly, when correcting the fourth quantity, only available historical recommendation information with an active semantic type first quantity proportion higher than the target proportion range is included in the fourth quantity.

[0084] As can be seen, in this example, modifying the domain labels in the basic recommendation labels according to the number of the first and second labels to obtain the target recommendation labels can improve the accuracy of the target recommendation label determination.

[0085] In one possible instance, before determining the basic recommendation tag based on the domain to which the available historical recommendation information belongs, the type to which the available historical recommendation information belongs, and the basic information, the method further includes: determining the frequency of acquiring the recommendation request within a second preset time period; determining the modification time interval between the last modification time of the basic information and the current time; determining the credibility of the basic information based on the frequency and the modification time interval; the step of determining the basic recommendation tag based on the domain to which the available historical recommendation information belongs, the type to which the available historical recommendation information belongs, and the basic information includes: if the credibility of the basic information is greater than a seventh preset value, then determining the basic recommendation tag based on the domain to which the available historical recommendation information belongs, the type to which the available historical recommendation information belongs, and the basic information; if the credibility of the basic information is not greater than the seventh preset value, then determining the basic recommendation tag based on the domain to which the available historical recommendation information belongs and the type to which the available historical recommendation information belongs.

[0086] Specifically, if a user frequently modifies their basic information within a short period, the credibility of that information is considered to be no greater than the seventh preset value. In other words, if a user frequently modifies their information, it cannot be determined whether the modified content is genuine. Therefore, when determining basic recommendation tags, if the credibility of the basic information is low, basic recommendation tags will not be generated based on that information, to avoid misleading the generation of basic recommendation tags.

[0087] In the specific implementation, when determining the basic recommendation tags, if the credibility of the basic information is higher than the seventh preset value, the user's current input voice information can be obtained first. Based on the current input voice information, it can be determined whether the current user is the target user corresponding to the basic information. The basic information includes the target user's voiceprint information. If it is found that the current user is not the target user based on the voiceprint information comparison, the basic recommendation tags are not generated based on the basic information. Before obtaining the reference historical recommendation information, the voiceprint information of the voice data is first determined from the historical voice interaction data, and the historical voice interaction data with the voiceprint information of the current user is determined. Then, the reference historical recommendation information is determined from the historical recommendation information corresponding to the historical voice interaction data with the voiceprint information of the current user.

[0088] As can be seen in this example, the ability to determine whether to generate recommendation tags based on the user's basic information can be improved by considering the time interval between the user's modifications to the basic information.

[0089] For examples consistent with the above embodiments, please refer to... Figure 5 , Figure 5This is a functional unit block diagram of a voice data-based recommendation information determination device 40 provided in an embodiment of this application. The voice data-based recommendation information determination device 40 includes: a first acquisition unit 401, configured to acquire a recommendation request, the recommendation request being used to acquire recommendation information, the recommendation information including voice interaction content, voice interaction products, or voice interaction services to be recommended to a user; a second acquisition unit 402, configured to acquire reference voice interaction data of the user for reference historical recommendation information, the reference historical recommendation information being recommendation information for which the user has engaged in voice interaction among all recommendation information recommended to the user within a first preset time period; a first determination unit 403, configured to determine the actual interaction duration of the user for each reference historical recommendation information based on the reference voice interaction data; and a second determination unit 404, configured to determine the reference historical recommendation information whose actual interaction duration is greater than the first preset duration as target historical recommendation information, the reference historical recommendation information corresponding to which the target historical recommendation information is determined. The system includes: a third determining unit 405, used to determine the user's interaction continuity for each target historical recommendation based on the target voice interaction data; a fourth determining unit 406, used to determine available historical recommendation information from the target historical recommendation information based on the interaction continuity, wherein the target voice interaction data corresponding to the available historical recommendation information is available voice interaction data; a third obtaining unit 407, used to obtain the user's basic information, including the user's identity information and attention preference information, wherein the attention preference information is associated with the type and domain to which the recommendation information belongs; a fifth determining unit 408, used to determine target recommendation tags based on the available historical recommendation information, the available voice interaction data, and the basic information; and a fourth obtaining unit 409, used to determine target recommendation information based on the target recommendation tags.

[0090] In one possible instance, regarding the determination of the user's interaction continuity for each target historical recommendation based on the target voice interaction data, the third determining unit 405 is specifically configured to: perform the following operations on the target voice interaction data corresponding to each target historical recommendation: perform semantic analysis on the target voice interaction data to obtain multiple semantics; determine the actual interval duration between each pair of semantics; determine whether there is a target interval duration in the actual interval duration that is longer than a second preset duration; if so, determine whether the difference between the actual interaction duration corresponding to the target voice interaction data and the target interval duration is greater than the first preset duration; if so, determine that the user's interaction continuity for the target historical recommendation data corresponding to the target voice interaction data is continuous; if not, determine the number of times the target interval duration appears in the actual interval duration, and determine the interaction continuity of the target historical recommendation information based on the number of times; if not, determine that the user's interaction continuity for the target historical recommendation information corresponding to the target voice interaction data is continuous.

[0091] In one possible instance, regarding the determination of the interaction continuity of the target historical recommendation information based on the number of interactions, the third determining unit 405 is specifically configured to: obtain the interaction difficulty of the target historical recommendation information corresponding to the target voice interaction data; obtain the user's age based on the identity information; determine a reference interval duration based on the interaction difficulty and the user's age; determine the predicted total duration of the target interval duration based on the number of interactions and the reference interval duration; determine the predicted interaction duration based on the actual interaction duration and the predicted total duration; determine whether the predicted interaction duration is greater than the first preset duration; if it is greater, determine that the user's interaction continuity with the target historical recommendation information corresponding to the target voice interaction data is continuous; if it is not greater, determine that the user's interaction continuity with the target historical recommendation information corresponding to the target voice interaction data is discontinuous.

[0092] In one possible instance, regarding the determination of target recommendation tags based on the available historical recommendation information, the available voice interaction data, and the basic information, the fifth determining unit 408 is specifically configured to: determine basic recommendation tags based on the domain to which the available historical recommendation information belongs, the type to which the available historical recommendation information belongs, and the basic information; determine the semantic type of multiple available semantics included in each available voice interaction data, the semantic type including a regular semantic type, an active semantic type, and an irrelevant semantic type, wherein the regular semantic type is used to indicate that the first relevance of the available recommendation information corresponding to the available voice information to the voice interaction content, voice interaction product, or voice interaction service is higher than a first preset value, and the active semantic type is used to indicate that the first relevance is lower than the first preset value and is related to the available... The available recommendation information corresponding to the voice information includes semantics whose second relevance to the domain of the voice interaction content, voice interaction product, or voice interaction service is higher than a second preset value. The irrelevant semantic type is used to indicate semantics whose first relevance is lower than the first preset value and whose second relevance is lower than the second preset value. A first number of available semantics is determined for each semantic type corresponding to each available voice interaction data. Available historical recommendation information corresponding to the available voice interaction data is marked according to the first number. A second number of all available semantics included in each semantic type is determined. The user's speaking style is determined according to the second number. The credibility of the marking is determined according to the speaking style. The basic recommendation tag is changed according to the credibility and the marking to obtain the target recommendation tag.

[0093] In one possible instance, regarding the marking of available historical recommendation information corresponding to the available voice interaction data according to the first quantity, the fifth determining unit 408 is specifically configured to: perform the following steps for each available historical recommendation information: determine the reference domain to which the available historical recommendation information belongs; determine whether the basic recommendation tags include reference domain tags corresponding to the reference domain; if yes, then if the first quantity of available semantics included in the irrelevant semantic type corresponding to the available historical recommendation information is greater than a third preset value, mark the available historical recommendation information as a first mark; if no, then if the first quantity of available semantics included in the active semantic type corresponding to the available historical recommendation information is greater than a fourth preset value, mark the available historical recommendation information as a second mark.

[0094] In one possible instance, regarding the modification of the basic recommendation tag based on the credibility and the tag to obtain the target recommendation tag, the fifth determining unit 408 is specifically configured to: determine a third number of available historical recommendation information that contains the first tag and belongs to the reference domain; correct the third number based on the credibility; if the corrected third number is greater than a fifth preset value, delete the reference domain tag from the basic recommendation tag to obtain the target recommendation tag; determine a fourth number of available historical recommendation information that contains the second tag and does not belong to the reference domain; correct the fourth number based on the credibility; if the corrected fourth number is greater than a sixth preset value, add the reference domain tag to the basic recommendation tag to obtain the target recommendation tag.

[0095] In one possible instance, before determining the basic recommendation tag based on the domain of the available historical recommendation information, the type of the available historical recommendation information, and the basic information, the voice data-based recommendation information determining device 40 is further configured to: determine the frequency of acquiring the recommendation request within a second preset time period; determine the modification time interval between the last modification time of the basic information and the current time; determine the credibility of the basic information based on the frequency and the modification time interval; and, specifically, if the credibility of the basic information is greater than a seventh preset value, determine the basic recommendation tag based on the domain of the available historical recommendation information, the type of the available historical recommendation information, and the basic information; and if the credibility of the basic information is not greater than the seventh preset value, determine the basic recommendation tag based on the domain of the available historical recommendation information and the type of the available historical recommendation information.

[0096] It is understood that since the method embodiments and the device embodiments are different presentations of the same technical concept, the content of the method embodiment section in this application should be adapted to the device embodiment section in a synchronous manner, and will not be repeated here.

[0097] In the case of using integrated units, please refer to Figure 6 , Figure 6 This is a functional unit block diagram of another recommendation information determination device based on voice data provided in an embodiment of this application. Figure 6The voice data-based recommendation information determination device 500 includes a processing module 512 and a communication module 511. The processing module 512 controls and manages the actions of the voice data-based recommendation information determination device, for example, executing the steps of the first acquisition unit 401, the second acquisition unit 402, the first determination unit 403, the second determination unit 404, the third determination unit 405, the fourth determination unit 406, the third acquisition unit 407, the fifth determination unit 408, and the fourth acquisition unit 409, and / or executing other processes of the technology described herein. The communication module 511 is used for interaction between the voice data-based recommendation information determination device and other devices. Figure 6 As shown, the recommendation information determination device based on voice data may further include a storage module 513, which is used to store the program code and data of the recommendation information determination device based on voice data.

[0098] The processing module 512 can be a processor or controller, such as a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an ASIC, an FPGA, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc. The communication module 511 can be a transceiver, RF circuitry, or a communication interface, etc. The storage module 513 can be a memory.

[0099] All relevant content in each scenario involved in the above method embodiments can be referenced from the functional descriptions of the corresponding functional modules, and will not be repeated here. The above-mentioned recommendation information determination device 500 based on voice data can perform the above... Figure 3 The method for determining recommendation information based on voice data is shown.

[0100] The above mainly describes the solutions of the embodiments of this application from the perspective of the method execution process. It is understood that, in order to achieve the above functions, the electronic device includes the corresponding hardware structure and software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments provided herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0101] This application embodiment can divide the electronic device into functional units according to the above method example. For example, each function can be divided into a separate functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this application embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0102] This application also provides a chip, wherein the chip includes a processor for calling and running a computer program from a memory, causing a device on which the chip is installed to perform some or all of the steps described in the above method embodiments of the electronic device.

Claims

1. A method for determining recommendation information based on voice data, characterized in that, The method includes: Obtain a recommendation request, wherein the recommendation information includes voice interaction content, voice interaction products, or voice interaction services to recommend to the user; Obtain the user's reference voice interaction data in response to the reference historical recommendation information, wherein the reference historical recommendation information is the recommendation information in which the user has voice interaction among all the recommendation information recommended to the user within a first preset time period; The actual interaction duration of the user for each reference historical recommendation is determined based on the reference voice interaction data. The reference historical recommendation information whose actual interaction duration is greater than a first preset duration is determined to be the target historical recommendation information, and the reference voice interaction data corresponding to the target historical recommendation information is the target voice interaction data. The continuity of user interaction with each target historical recommendation information is determined based on the target voice interaction data; Based on the interaction continuity, available historical recommendation information is determined from the target historical recommendation information, and the target voice interaction data corresponding to the available historical recommendation information is the available voice interaction data. Obtain the user's basic information, which includes the user's identity information and interest preference information, and the interest preference information is associated with the type and domain of the recommendation information; The target recommendation tag is determined based on the available historical recommendation information, the available voice interaction data, and the basic information; Target recommendation information is determined based on the target recommendation tags.

2. The method according to claim 1, characterized in that, Determining the user's interaction continuity for each target historical recommendation based on the target voice interaction data includes: Perform the following operations on the target voice interaction data corresponding to each target's historical recommendation information: Semantic analysis is performed on the target voice interaction data to obtain multiple semantics; Determine the actual interval between each pair of semantics; Determine whether there is a target interval duration in the actual interval duration that is longer than the second preset duration; If it exists, determine whether the difference between the actual interaction duration corresponding to the target voice interaction data and the target interval duration is greater than the first preset duration; If so, then the continuity of the user's interaction with the target historical recommendation information corresponding to the target voice interaction data is determined to be continuous; If not, determine the number of times the target interval appears in the actual interval duration, and determine the interaction continuity of the target historical recommendation information based on the number of occurrences; If it does not exist, then the user's interaction continuity with the target historical recommendation information corresponding to the target voice interaction data is determined to be continuous.

3. The method according to claim 2, characterized in that, Determining the interaction continuity of the target historical recommendation information based on the number of interactions includes: The interaction difficulty of obtaining the target historical recommendation information corresponding to the target voice interaction data; The user's age is obtained based on the identity information; The reference interval duration is determined based on the difficulty of the interaction and the user's age; The total predicted duration of the target interval is determined based on the number of times and the reference interval duration. The predicted interaction duration is determined based on the sum of the actual interaction duration and the predicted duration. Determine whether the predicted interaction duration is greater than the first preset duration; If it is greater than, then the user's interaction continuity with the target historical recommendation information corresponding to the target voice interaction data is determined to be continuous; If it is not greater than, then the user's interaction continuity with the target historical recommendation information corresponding to the target voice interaction data is determined to be discontinuous.

4. The method according to claim 3, characterized in that, The step of determining the target recommendation tag based on the available historical recommendation information, the available voice interaction data, and the basic information includes: The basic recommendation tags are determined based on the domain to which the available historical recommendation information belongs, the type to which the available historical recommendation information belongs, and the basic information. The semantic types of multiple available semantics included in each available voice interaction data are determined. The semantic types include regular semantic types, active semantic types, and irrelevant semantic types. The regular semantic types are used to indicate semantics whose first relevance to the available recommendation information corresponding to the available voice information is higher than a first preset value. The active semantic types are used to indicate semantics whose first relevance is lower than the first preset value and whose second relevance to the domain of the available recommendation information corresponding to the available voice information is higher than a second preset value. The irrelevant semantic types are used to indicate semantics whose first relevance is lower than the first preset value and whose second relevance is lower than the second preset value. Determine the first number of available semantics included in each semantic type corresponding to each available voice interaction data; The available historical recommendation information corresponding to the available voice interaction data is marked according to the first quantity; Determine a second number of all available semantics included in each semantic type; The user's speaking style is determined based on the second quantity; The credibility of the tag is determined based on the speaking style; The base recommendation label is modified based on the credibility and the tag to obtain the target recommendation label.

5. The method according to claim 4, characterized in that, The step of marking the available historical recommendation information corresponding to the available voice interaction data according to the first quantity includes: For each available historical recommendation, perform the following steps: Determine the reference domain to which the available historical recommendation information belongs; Determine whether the basic recommendation tags include reference domain tags corresponding to the reference domain; If so, if the first number of available semantics included in the irrelevant semantic type corresponding to the available historical recommendation information is greater than the third preset value, the available historical recommendation information is marked with a first mark, which is used to characterize the user's reduced liking for the reference domain; If not, then if the first number of available semantics included in the active semantic type corresponding to the available historical recommendation information is greater than the fourth preset value, the available historical recommendation information is marked with a second mark, which is used to indicate that the user's preference for the reference domain is enhanced.

6. The method according to claim 5, characterized in that, The step of modifying the base recommendation tag according to the credibility and the tag to obtain the target recommendation tag includes: Determine a third number of available historical recommendation information that contains the first tag and belongs to the reference domain; The third quantity is adjusted based on the stated credibility. If the corrected third quantity is greater than the fifth preset value, the reference domain label is deleted from the basic recommendation label to obtain the target recommendation label; Determine a fourth number of available historical recommendation information that contains the second tag but does not belong to the reference domain; The fourth quantity is adjusted based on the stated credibility. If the corrected fourth quantity is greater than the sixth preset value, the reference domain label is added to the basic recommendation label to obtain the target recommendation label.

7. The method according to claim 6, characterized in that, Before determining the basic recommendation tags based on the domain to which the available historical recommendation information belongs, the type to which the available historical recommendation information belongs, and the basic information, the method further includes: Determine the frequency of obtaining the recommendation request within the second preset time period; Determine the time interval between the last modification time of the basic information and the current time; The credibility of the basic information is determined based on the frequency and the modification time interval; The step of determining the basic recommendation tags based on the domain to which the available historical recommendation information belongs, the type to which the available historical recommendation information belongs, and the basic information includes: If the credibility of the basic information is greater than the seventh preset value, then the basic recommendation tag is determined based on the domain to which the available historical recommendation information belongs, the type to which the available historical recommendation information belongs, and the basic information. If the credibility of the basic information is not greater than the seventh preset value, then the basic recommendation tag is determined according to the domain to which the available historical recommendation information belongs and the type to which the available historical recommendation information belongs.

8. A device for determining recommendation information based on voice data, characterized in that, The device includes: The first acquisition unit is used to acquire a recommendation request, the recommendation request being used to acquire recommendation information, the recommendation information including voice interaction content, voice interaction products, or voice interaction services to be recommended to the user; The second acquisition unit is used to acquire the user's reference voice interaction data in response to the reference historical recommendation information, wherein the reference historical recommendation information is the recommendation information in which the user has voice interaction among all the recommendation information recommended to the user within a first preset time period; The first determining unit is used to determine the actual interaction duration of the user for each reference historical recommendation based on the reference voice interaction data. The second determining unit is used to determine the reference historical recommendation information whose actual interaction duration is greater than the first preset duration as the target historical recommendation information, and the reference voice interaction data corresponding to the target historical recommendation information as the target voice interaction data. The third determining unit is used to determine the continuity of the user's interaction with each target historical recommendation information based on the target voice interaction data. The fourth determining unit is used to determine available historical recommendation information from the target historical recommendation information based on the interaction continuity, wherein the target voice interaction data corresponding to the available historical recommendation information is available voice interaction data. The third acquisition unit is used to acquire the user's basic information, which includes the user's identity information and attention preference information, and the attention preference information is associated with the type and domain to which the recommendation information belongs. The fifth determining unit is used to determine the target recommendation tag based on the available historical recommendation information, the available voice interaction data, and the basic information; The fourth acquisition unit is used to determine target recommendation information based on the target recommendation tags.

9. An electronic device, characterized in that, The method includes a processor, a memory, and one or more programs, said programs being stored in the memory and configured to be executed by the processor, said programs including instructions for performing the steps of the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, The computer program / instructions are executed by the processor to implement the steps of the method according to any one of claims 1-7.

Images