Voice interaction method based on dynamic label library and terminal equipment

Abstract

The invention relates to a voice interaction method based on a dynamic tag library and terminal equipment. The method comprises the following steps: receiving voice interaction information sent by a user; performing intention recognition on the voice interaction information based on a preset dynamic tag library, and determining an intention recognition result matched with the voice interaction information; wherein the preset dynamic label library comprises a user behavior label library and a dynamic content label library; the user behavior tag library comprises behavior tags related to user preferences and behavior habits; the dynamic content label library comprises dynamic content labels related to user preferences; based on the intention recognition result, executing an action sequence corresponding to the intention recognition result; the interaction intention of the user can be more comprehensively and accurately understood, so that accurate interaction is realized, and the individuation degree and accuracy of voice interaction are improved.

Description

Technical Field

[0001] The present application relates to the field of speech recognition technology, and in particular to a speech interaction method and terminal device based on a dynamic tag library. Background Art

[0002] In modern smart devices and applications, voice interaction has become a mainstream form of user interface. With the advancement of artificial intelligence and natural language processing technology, devices can understand and respond to users' voice commands, which greatly improves the user experience.

[0003] However, traditional voice interaction methods often rely on fixed rules, which limits the ability to understand complex and diverse user intentions. Summary of the invention

[0004] Based on this, it is necessary to provide a voice interaction method and terminal device based on a dynamic tag library to address the above technical problems.

[0005] In a first aspect, the present application provides a voice interaction method based on a dynamic tag library, the method comprising: Receive voice interaction information sent by users; Based on a preset dynamic tag library, the voice interaction information is subjected to intent recognition, and an intent recognition result matching the voice interaction information is determined; wherein the preset dynamic tag library includes a user behavior tag library and a dynamic content tag library; the user behavior tag library includes behavior tags related to user preferences and behavior habits; the dynamic content tag library includes dynamic content tags related to user preferences; Based on the intention recognition result, an action sequence corresponding to the intention recognition result is executed.

[0006] In one implementation, the method further includes: Obtain historical voice information corresponding to the target behavior within a preset update period; Determining a plurality of candidate behavior tags corresponding to the target behavior from the historical speech information; Determine the usage frequency corresponding to each of the plurality of candidate behavior tags; Determining an updated behavior tag corresponding to the target behavior from the plurality of candidate behavior tags according to the usage frequencies corresponding to the plurality of candidate behavior tags; The user behavior tag library is dynamically updated according to the target behavior and the updated behavior tag corresponding to the target behavior.

[0007] In one implementation, the method further includes: Obtain user preference information and historical interaction information; The dynamic content tag library is dynamically updated according to a preset dynamic event data set, the user preference information and the historical interaction information.

[0008] In one implementation, dynamically updating the dynamic content tag library according to the preset dynamic event data set, the user preference information, and the historical interaction information includes: Determining tag weight information according to the historical interaction information; According to the tag weight information, determining a plurality of target dynamic event information matching the user preference information from the preset dynamic event data set; The dynamic content tag library is dynamically updated according to the plurality of target dynamic event information.

[0009] In one implementation, the method further includes: Get historical interaction information; Determining, from the historical interaction information, a voice interaction instance that does not meet user expectations; Performing a behavior cause analysis on the voice interaction instance to determine a target behavior cause corresponding to the voice interaction instance; The user behavior tag library is dynamically updated according to the target behavior reason.

[0010] In one implementation, the performing intent recognition on the voice interaction information based on a preset dynamic tag library and determining an intent recognition result matching the voice interaction information includes: Determining a target tag matching the voice interaction information from the preset dynamic tag library; If there is a target tag matching the voice interaction information in the preset dynamic tag library, determining an intent recognition result corresponding to the target tag from the preset dynamic tag library; If there is no target tag matching the voice interaction information in the preset dynamic tag library, performing function word segmentation processing on the voice interaction information to obtain multiple target keywords; According to the multiple target keywords, an intention recognition result corresponding to the voice interaction information is determined.

[0011] In one implementation, the performing of an action sequence corresponding to the intention recognition result based on the intention recognition result includes: If the intention recognition result is a professional field intention, a vertical category model matching the professional field intention is called to provide knowledge answers; If the intention recognition result is not a professional field intention, an action sequence corresponding to the intention recognition result is executed.

[0012] In a second aspect, the present application also provides a voice interaction device based on a dynamic tag library, the device comprising: A receiving module, used to receive voice interaction information sent by the user; An intention recognition module, configured to perform intention recognition on the voice interaction information based on a preset dynamic tag library, and determine an intention recognition result matching the voice interaction information; wherein the preset dynamic tag library includes a user behavior tag library and a dynamic content tag library; the user behavior tag library includes behavior tags related to user preferences and behavior habits; the dynamic content tag library includes dynamic content tags related to user preferences; An execution module is used to execute an action sequence corresponding to the intention recognition result based on the intention recognition result.

[0013] In a third aspect, the present application further provides a terminal device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method described in any one of the embodiments of the first aspect are implemented.

[0014] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method described in any one of the embodiments of the first aspect above are implemented.

[0015] In a fifth aspect, the present application further provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the method described in any one of the embodiments of the first aspect above.

[0016] The above-mentioned voice interaction method and terminal device based on dynamic tag library; receive voice interaction information sent by the user; further, based on the preset dynamic tag library, perform intent recognition on the voice interaction information, and determine the intent recognition result matching the voice interaction information; wherein, the preset dynamic tag library includes a user behavior tag library and a dynamic content tag library; the user behavior tag library includes behavior tags related to user preferences and behavior habits; the dynamic content tag library includes dynamic content tags related to user preferences; based on the preset dynamic tag library, not only can the user's true intention be accurately identified, but also the user's personalized preferences, historical behavior habits, the latest dynamic content, etc. can be taken into consideration, thereby providing personalized services that are more in line with user needs; finally, based on the intention recognition result, execute the action sequence corresponding to the intention recognition result; based on this, the user's interaction intention can be understood more comprehensively and accurately to achieve precise interaction, thereby improving the personalization and accuracy of voice interaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 This is an application environment diagram of a voice interaction method based on a dynamic tag library in one embodiment; Figure 2 A flowchart of a voice interaction method based on a dynamic tag library in one embodiment; Figure 3 A schematic diagram of a process for updating a user behavior tag library in one embodiment; Figure 4 A flowchart of an intention identification step in one embodiment; Figure 5 is a structural block diagram of a voice interaction device based on a dynamic tag library in one embodiment; Figure 6 FIG. 4 is a diagram showing the internal structure of a terminal device in an embodiment. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0020] Voice interaction technology is becoming increasingly mature, and with the improvement of the hardware capabilities of smart devices and the development of general large model technology, voice as the terminal entrance has gradually become one of the main ways for people to interact with computers. Users can issue control instructions to various devices through voice. Compared with traditional natural language processing technology, general large models have higher logical reasoning capabilities and can better understand user intentions. However, due to the diversity and complexity of language expression and the increasingly complex functions provided by smart devices, the human-computer interaction experience based on fixed rule voice information is poor. For example, all are given to the corresponding application intact, resulting in the information processed by the application being too long or too scattered, and the user's meaning cannot be well understood. The embodiment of the present application aims to provide a voice interaction method based on a dynamic tag library to solve the problem of over-reliance on fixed rules in the above-mentioned prior art, resulting in poor understanding of complex and diverse user intentions.

[0021] The voice interaction method based on the dynamic tag library provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablets, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car devices, projection devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, etc. The server 104 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0022] In an exemplary embodiment, Figure 2 As shown, Figure 2 The figure is a flow chart of a voice interaction method based on a dynamic tag library in an embodiment; this embodiment uses the method applied to a smart terminal as an example, and the voice interaction method based on a dynamic tag library includes the following steps: Step S201, receiving voice interaction information sent by a user.

[0023] Among them, voice interaction information refers to the natural language data input by the user; for example, "play music" etc.

[0024] Exemplarily, the application of the voice interaction method based on the dynamic tag library to a mixed reality (MR) device is taken as an example for explanation, and the voice interaction information sent by the user is received through the microphone of the MR device.

[0025] Step S202: Based on a preset dynamic tag library, perform intent recognition on the voice interaction information to determine an intent recognition result that matches the voice interaction information.

[0026] Among them, the preset dynamic tag library includes a user behavior tag library and a dynamic content tag library; the user behavior tag library includes behavior tags related to user preferences and behavior habits; among them, behavior tags include instructions related to human-computer interaction actions, such as "listen" and "watch".

[0027] It is understandable that for each behavior tag in the user behavior tag library, there is a workflow corresponding to the behavior tag; for example, when the behavior tag is "Watch TV series A", the workflow corresponding to the behavior tag can be "Open the target playback platform corresponding to TV series A, search for TV series A in the target playback platform, and play TV series A".

[0028] The dynamic content tag library includes dynamic content tags related to user preferences. The dynamic content tags are dynamic content that is continuously updated based on variables such as time, events, and popular trends. It is understandable that the dynamic content tag library will be updated in real time or regularly based on the latest events or popular trends, combined with user preferences, to ensure the timeliness of the information and ensure that the content recommendations are more in line with personal interests.

[0029] Among them, the method for performing intent recognition on voice interaction information may include, but is not limited to, performing tag recognition on the voice interaction information; illustratively, determining whether there is a tag matching the voice interaction information from a preset dynamic tag library.

[0030] Among them, the intent recognition result refers to the understanding of the specific action or request that the user wants to perform; it can be understood that the intent recognition result can be directly converted into a specific control instruction.

[0031] Exemplarily, based on a preset dynamic tag library, intent recognition is performed on voice interaction information. When there is a tag matching the voice interaction information in the preset dynamic tag library, the intent recognition result corresponding to the voice interaction information is determined based on the tag matching the voice interaction information.

[0032] Step S203: based on the intention recognition result, execute an action sequence corresponding to the intention recognition result.

[0033] An action sequence refers to a series of sub-actions executed in a specific order. Each sub-action is a step in the entire action sequence, and each sub-action is related to each other.

[0034] For example, when the intent recognition result is "play music A", the action sequence corresponding to the intent recognition result may be "open the player", "search for music A", "load music A", and "play music A". It can be understood that "open the player", "search for music A", "load music A", and "play music A" are all sub-actions in the action sequence.

[0035] In this embodiment, voice interaction information sent by the user is received; further, based on a preset dynamic tag library, intent recognition is performed on the voice interaction information to determine an intent recognition result that matches the voice interaction information; based on the preset dynamic tag library, not only can the user's true intention be accurately identified, but the user's personalized preferences, historical behavioral habits, latest dynamic content, etc. can also be taken into consideration, thereby providing personalized services that better meet user needs; finally, based on the intent recognition result, an action sequence corresponding to the intent recognition result is executed; based on this, the user's interaction intention can be understood more comprehensively and accurately to achieve precise interaction, thereby improving the personalization and accuracy of voice interaction.

[0036] In one embodiment, Figure 3 As shown, Figure 3 The figure is a flowchart of the user behavior tag library update step in one embodiment; the voice interaction method based on the dynamic tag library also includes the following steps: Step S301, obtaining historical voice information within a preset update period.

[0037] Among them, the preset update cycle needs to be set according to the actual update requirements of the user behavior tag library, and is not specifically limited here; for example, the preset update cycle can be one month or three months.

[0038] The historical voice information refers to the historical natural voice input by the user. For example, the historical voice information includes "watch a movie", "watch a movie", "watch a movie", "listen to music", "turn on music", etc.

[0039] Step S302: determining a plurality of candidate behavior tags corresponding to the target behavior from the historical speech information.

[0040] The target behavior refers to the actual operation corresponding to the specific intention expressed by the user through voice input during the interaction with the smart terminal. The candidate behavior label refers to the instruction related to the target behavior.

[0041] Exemplarily, assuming that the target behavior is "playing a movie", the historical voice information includes "watch a movie", "watch a movie", "watch a movie", "listen to music", "turn on music", etc., multiple candidate behavior labels corresponding to the target behavior are determined from the historical voice information, and the multiple candidate behavior labels obtained are "watch a movie", "watch a movie", and "watch a movie".

[0042] It is understandable that based on multiple candidate behavior tags, the user's behavior habits can be analyzed, and then the user behavior tag library can be dynamically updated according to the user's behavior habits, which can effectively improve the personalization of voice interaction.

[0043] Step S303: determining the usage frequencies corresponding to each of the plurality of candidate behavior tags.

[0044] Exemplarily, for each candidate behavior tag, the number of times it appears in the historical voice information is counted, and the usage frequency corresponding to each candidate behavior tag is determined according to the number of times each candidate behavior tag appears in the historical voice information.

[0045] Step S304 : determining an updated behavior tag corresponding to the target behavior from the plurality of candidate behavior tags according to the usage frequencies corresponding to the plurality of candidate behavior tags.

[0046] In an exemplary embodiment, according to the usage frequencies corresponding to each of the multiple candidate behavior tags, a method for determining an updated behavior tag corresponding to a target behavior from multiple candidate behavior tags may be: according to the usage frequencies corresponding to each of the multiple candidate behavior tags, the candidate behavior tag with the highest usage frequency is used as the updated behavior tag corresponding to the target behavior.

[0047] In another exemplary embodiment, according to the usage frequencies corresponding to each of the multiple candidate behavior tags, the method for determining the updated behavior tag corresponding to the target behavior from multiple candidate behavior tags may also be: according to the usage frequencies corresponding to each of the multiple candidate behavior tags, the candidate behavior tag with the highest usage frequency is used as the first behavior tag; according to the preset dynamic event data set, a second behavior tag matching the first behavior tag is determined; and the second behavior tag is used as the updated behavior tag corresponding to the target behavior. Among them, the first behavior tag refers to the candidate behavior tag with the highest usage frequency among the multiple candidate behavior tags; the second behavior tag refers to the behavior tag associated with the first behavior tag to be updated to the user behavior tag library. The preset dynamic event data set refers to a data set containing real-time updated information from multiple sources; the preset dynamic event data set can be, but is not limited to, the latest dynamic data obtained through multiple channels such as search engines.

[0048] Exemplarily, assuming the target behavior is "play comedy variety show A", based on historical voice information, it is determined that the most frequently used first behavior label is "open the variety show of comedy group B"; wherein comedy group B is the group of comedians in comedy variety show A; further, based on a preset dynamic event data set, a second behavior label that matches the first behavior label is automatically determined, for example, combined abbreviation terms, hot search terms, fan-made words, etc. related to comedy group B; the second behavior label is used as the updated behavior label corresponding to the target behavior; based on this, when the user subsequently uses combined abbreviation terms, fan-made words and other information related to comedy group B as voice interaction information, the voice interaction information can be directly responded to based on the updated user behavior label library, thereby realizing a targeted update of the user behavior label library and further improving the personalization of voice interaction.

[0049] Step S305: dynamically update the user behavior tag library according to the target behavior and the updated behavior tag corresponding to the target behavior.

[0050] In an exemplary embodiment, it is assumed that when the MR device is used for the first time, the user behavior tag library includes behavior tags such as "open", "play", and "listen". After the MR device is used for a period of time, historical voice information is obtained according to a preset update cycle. By analyzing the historical voice information, it can be determined that the user often uses "start", "listen", and "I want to watch" as common words for the corresponding intentions, and the frequency of use is very high. Then, the user behavior tag library is updated according to "start", "listen", and "I want to watch" to improve the personalization of the user behavior tag library.

[0051] It should be noted that the user behavior tag library is stored locally on the smart terminal to ensure the privacy of user behavior and speed up the response speed of user operations.

[0052] In this embodiment, based on the historical voice information within the preset update cycle, the user preferences and behavioral habits can be understood more accurately, and then the user behavior tags are regularly updated according to the user preferences and behavioral habits. This not only improves the personalization of the user behavior tag library, but also can capture the changes in user behavior in a timely manner, thereby improving the timeliness of the user behavior tag library.

[0053] In one embodiment, the voice interaction method based on the dynamic tag library further includes the following steps: Step 1: Obtain user preference information and historical interaction information.

[0054] The historical interaction information refers to all interaction records between the user and the smart terminal in a certain period of time in the past; the historical interaction information may include, but is not limited to, the historical usage frequency of each application, the historical interaction content type, etc.

[0055] Among them, user preference information refers to the habits, preferences and tendencies of users when using smart terminals; for example, user preference information may include, but is not limited to, content preferences, behavior patterns, etc.; content preferences may include user preferred media forms, such as music, movies, books, news, etc.; content preferences may also include user preferred content styles or genres, such as rock music, science fiction movies, historical novels, etc. Behavior patterns may include the frequency of users using certain functions or services, such as listening to the music of musician A every day.

[0056] Step 2: Dynamically update the dynamic content tag library based on the preset dynamic event data set, user preference information, and historical interaction information.

[0057] The preset dynamic event data set refers to a data set containing real-time updated information from multiple sources, which aims to capture and reflect current social hot spots, public interests and media trends; the preset dynamic event data set can be, but is not limited to, the latest dynamic data obtained through multiple channels such as search engines. The multiple channels can include, but are not limited to, news, Baidu search index, Weibo hot search list, Douyin hot search list, WeChat hot search list, Toutiao hot search list, etc.

[0058] In an exemplary embodiment, step 2, dynamically updating the dynamic content tag library according to the preset dynamic event data set, user preference information and historical interaction information, includes the following steps: Step 2.1: Determine label weight information based on historical interaction information.

[0059] Among them, the tag weight information is used to indicate the importance of the corresponding tag. For example, based on the historical interaction information, the usage frequency of each application in the MR device is determined; the usage frequency of each application is sorted by weight ratio; for example, based on the usage frequency of video, music, short video, and literature applications, the weight ratio is determined to be 4:3:2:1, and then the tag weight information corresponding to each type of application can be determined to be 0.4, 0.3, 0.2, and 0.1 respectively according to the weight ratio.

[0060] Optionally, in an exemplary embodiment, further weight ratio sorting can be performed based on the user's preferences and interests in each type of application; for example, for video applications, weight ratio sorting can be performed based on the frequency of users browsing a certain type of video to obtain the corresponding weight ratio; further, label weight information related to the video type can be obtained based on the weight ratio; for example, assuming that the corresponding weight ratio of TV series, variety shows, and movies is 6:3:1, it can be determined that the label weight information of the TV series is 0.6, the label weight information of the variety show is 0.3, and the label weight information of the movie is 0.1.

[0061] Step 2.2, based on the tag weight information, determine multiple target dynamic event information matching the user preference information from the preset dynamic event data set.

[0062] Step 2.3, dynamically update the dynamic content tag library according to multiple target dynamic event information.

[0063] Exemplarily, based on historical interaction information, determine the tag weight information corresponding to different tags, and then determine multiple target dynamic event information that matches the user preference information from the preset dynamic event data set based on the tag weight information; for example, when the tag weight information corresponding to the current user's use of video applications is 0.5, and the tag weight information corresponding to the use of music applications is 0.3, if 300 words need to be updated, the preset dynamic event data set will be applied for the target dynamic event information with the highest comprehensive ranking of 150 video applications and the target dynamic event information with the highest comprehensive ranking of 90 music applications. Then, based on the above multiple target dynamic event information, the dynamic content tag library is dynamically updated to ensure that the dynamic content tag library can timely update the information, popular trends or hot topics of interest to users, improve the timeliness of dynamic content tags, and lay the foundation for improving the personalization of voice interaction.

[0064] It should be noted that the dynamic content tag library is stored in the cloud server, which facilitates the rapid update of news hotspots and other content, and can effectively improve the response speed and timeliness of the dynamic content tag library.

[0065] In this embodiment, based on historical interaction information, the user's preferences can be analyzed to accurately determine the tag weight information; then, based on the tag weight information, multiple target dynamic event information matching the user's preference information is determined from the preset dynamic event data set, and the dynamic content tag library is dynamically updated based on the multiple target dynamic event information, which can ensure that the dynamic content tag library quickly responds to changes in the latest trends and hot topics of interest to users, laying the foundation for improving the personalization of voice interaction.

[0066] In one embodiment, the voice interaction method based on the dynamic tag library further includes the following steps: Step 1: Obtain historical interaction information.

[0067] Step 2: Determine the voice interaction instances that do not meet the user's expectations from the historical interaction information.

[0068] In an exemplary embodiment, a method for determining voice interaction instances that do not meet user expectations from historical interaction information may be: based on the historical interaction information, determine whether each voice interaction successfully solves the user's problem or request; if there is voice information that does not meet preset conditions, determine the voice information as a voice interaction instance that does not meet user expectations; wherein the preset conditions need to be set according to actual needs and are not specifically limited here; for example, the preset conditions may be that they can always effectively solve user needs, etc.

[0069] For example, the default music player currently set by the user is player A, but the user always says "close player A" or "open player B" immediately after saying "play the song of musician A". If the user performs too many such operations, "play the song of musician A" can be determined as a voice interaction instance that does not meet the user's expectations.

[0070] Step 3: Analyze the behavior reasons of the voice interaction instance and determine the target behavior reasons corresponding to the voice interaction instance.

[0071] In an exemplary embodiment, the method for performing behavioral cause analysis on a voice interaction instance may be: performing behavioral cause analysis on the voice interaction instance based on existing clues of the voice interaction instance and a local preset knowledge base; if the behavioral cause corresponding to the voice interaction instance cannot be analyzed based on the local preset knowledge base, uploading the voice interaction instance to a cloud server; performing behavioral cause analysis on the voice interaction instance through the cloud server to determine the target behavioral cause corresponding to the voice interaction instance.

[0072] Step 4: Dynamically update the user behavior tag library based on the target behavior reasons.

[0073] For example, take "playing the song of musician A" as an example of voice interaction that does not meet user expectations. By analyzing the behavior reasons of "playing the song of musician A", the target behavior reason obtained is "currently only player B has the copyright of the song of musician A"; further, musician A and player B can be associated, and the user behavior tag library can be dynamically updated. Based on this, when the user sends the voice interaction information of "playing the song of musician A", a series of action sequences such as opening player B, searching for musician A, and playing the song of musician A can be directly executed according to the user behavior tag library.

[0074] In this embodiment, by determining the voice interaction instances that do not meet the user's expectations from the historical interaction information and performing behavioral cause analysis on the voice interaction instances, the target behavior causes corresponding to the voice interaction instances can be accurately determined, and based on this, potential behavior patterns can be mined, further improving the reliability of voice interaction; based on the target behavior causes, the user behavior tag library is dynamically updated, further improving the personalization and accuracy of voice interaction.

[0075] In one embodiment, Figure 4 As shown, Figure 4 The flowchart of the intention recognition step in one embodiment is as follows; based on a preset dynamic tag library, the intention recognition is performed on the voice interaction information, and the intention recognition result matching the voice interaction information is determined, including the following steps: Step S401: Determine a target tag that matches the voice interaction information from a preset dynamic tag library.

[0076] Step S402: determine whether there is a target tag matching the voice interaction information in the preset dynamic tag library.

[0077] If it exists, execute step S403; if it does not exist, execute step S404.

[0078] Step S403: If there is a target tag matching the voice interaction information in the preset dynamic tag library, an intent recognition result corresponding to the target tag is determined from the preset dynamic tag library.

[0079] For example, when the voice interaction information sent by the user is "Watch TV series A", the target tag matching "Watch TV series A" is determined to be "TV series A" from the preset dynamic tag library; further, the playback platform information associated with "TV series A" can be determined to be "Video application A" from the preset dynamic tag library. According to "TV series A" and "Video application A", the corresponding intent recognition result can be determined to be "Open video application A to play TV series A". Furthermore, based on the intent recognition result, the action sequence corresponding to the intent recognition result can be executed to realize voice interaction.

[0080] Step S404: If there is no target tag matching the voice interaction information in the preset dynamic tag library, the voice interaction information is segmented into function words to obtain a plurality of target keywords.

[0081] Among them, the function word segmentation process refers to eliminating words that do not carry actual semantic information and retaining keywords that carry actual semantic information.

[0082] Step S405: Determine the intention recognition result corresponding to the voice interaction information based on the multiple target keywords.

[0083] Exemplarily, when there is no target tag matching the voice interaction information in the preset dynamic tag library, the voice interaction information is processed by function word segmentation to remove the function words in the voice interaction information and retain multiple target keywords; for example, the user says "Help me open video application B and play TV series B to watch", if there is no target tag matching the voice interaction information in the preset dynamic tag library, then "Help me open video application B and play TV series B to watch" is processed by function word segmentation to obtain multiple target keywords, namely "open", "video application B", "TV series B", and "watch"; further, according to multiple target keywords, the intention is pieced together to determine the intention recognition result corresponding to the voice interaction information, and the intention recognition result is "open application: video application B", "search video: TV series B"; further, the intention recognition result is converted into a format to obtain an operation instruction corresponding to the intention recognition result; voice interaction is realized based on the operation instruction. At the same time, the above-mentioned behavior is saved and recorded so as to be used as a check for omissions in the subsequent update of the preset dynamic tag library.

[0084] In this embodiment, when there is a target tag matching the voice interaction information in the preset dynamic tag library, the corresponding intent recognition result is directly obtained from the preset dynamic tag library, ensuring the speed and accuracy of the response; when there is no target tag matching the voice interaction information in the preset dynamic tag library, the voice interaction information is processed by function word segmentation, multiple target keywords are extracted, and the intent recognition result is determined based on the multiple target keywords, which not only improves the ability to understand new or rare interaction patterns, but also enhances the flexibility and adaptability of voice interaction, ensuring that accurate services can be provided even in the face of unforeseen user input.

[0085] In an exemplary embodiment, when there is no target tag matching the voice interaction information in the preset dynamic tag library, and the intention recognition result corresponding to the voice interaction information cannot be determined based on multiple target keywords after the function word segmentation process, this embodiment provides a fallback strategy to ensure the reliability of the voice interaction. The fallback strategy includes multiple rounds of questions and answers; it can be understood that multiple rounds of questions and answers can guide users to complete their intentions, thereby ensuring the reliability and accuracy of the voice interaction.

[0086] Exemplarily, taking the voice interaction information sent by the user as "Open video application A, I want to watch C" as an example, when there is no target tag matching the voice interaction information in the preset dynamic tag library, and the intention recognition result corresponding to the voice interaction information cannot be determined based on multiple target keywords after the function word segmentation processing, the big model is used to guide the user to multiple rounds of questions and answers; for example, "Is this a TV series? Do you want to open the video software to play it?", the user answers "yes"; further, the big model is used to ask a follow-up question, "I am about to open video application A, is this drama in video application A?", the user answers "yes", and then all the voice questions are completed; according to the results of multiple rounds of questions and answers, corresponding actions can be performed.

[0087] In this embodiment, by setting a fallback strategy, it is possible to ensure that the user's intention is effectively completed, further ensuring the reliability and accuracy of the voice interaction.

[0088] In one embodiment, based on the intention recognition result, executing an action sequence corresponding to the intention recognition result includes the following steps: Step 1: If the intent recognition result is a professional domain intent, call the vertical category model that matches the professional domain intent to provide knowledge answers.

[0089] Among them, the vertical model mainly provides professional knowledge questions and answers based on the user's intention recognition results. The vertical model can include, but is not limited to, a local small vertical model and a cloud-based large vertical model. For example, when the smart device is connected to the Internet, the cloud-based large vertical model is called for answering; when the smart device is not connected to the Internet, the local small vertical model is called for answering.

[0090] Among them, vertical models can include, but are not limited to, vertical models in the health care field, vertical models in the legal field, vertical models in the financial field, and vertical models in the writing field. For example, when a user directly asks "how to judge a traffic accident", it can be determined based on the intention recognition result that this voice interaction information belongs to a legal issue, and the vertical model in the legal field will be called up; the vertical model in the legal field will give professional interpretations of laws and regulations, similar case information, and evidence suggestions.

[0091] Step 2: If the intent recognition result is not a professional domain intention, execute the action sequence corresponding to the intent recognition result.

[0092] For example, if the intent recognition result is not a professional field intent but other intents, the action sequence corresponding to the intent recognition result can be directly executed to ensure a quick response to user needs.

[0093] In this embodiment, when the intent recognition result is a professional field intent, the vertical model matching the professional field intent is called to provide knowledge answers, ensuring that users can obtain highly accurate, authoritative and detailed answers when consulting complex issues in specific fields. At the same time, the appropriate vertical model can be automatically selected according to the networking status of the smart device, which improves the flexibility of voice interaction. When the intent recognition result is a non-professional field intent, the action sequence corresponding to the intent recognition result is directly executed, which simplifies the processing flow, reduces unnecessary computing overhead, and makes the response faster and more efficient.

[0094] In a specific embodiment, a voice interaction big model is embedded in the smart terminal. The voice interaction big model has the authority to access hardware resources and software resources. When installing an application, the voice interaction big model can understand information such as its software functions and the content services provided.

[0095] The above-mentioned voice interaction method based on the dynamic tag library receives voice interaction information sent by the user; further, based on the preset dynamic tag library, the voice interaction information is subjected to intent recognition, and the intent recognition result matching the voice interaction information is determined; based on the preset dynamic tag library, not only can the user's true intention be accurately identified, but also the user's personalized preferences, historical behavior habits, the latest dynamic content, etc. can be taken into consideration, thereby providing personalized services that better meet the user's needs; finally, based on the intent recognition result, an action sequence corresponding to the intent recognition result is executed; based on this, the user's interaction intention can be understood more comprehensively and accurately to achieve precise interaction, thereby improving the personalization and accuracy of voice interaction.

[0096] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0097] Based on the same inventive concept, the embodiment of the present application also provides a voice interaction device based on a dynamic tag library for implementing the voice interaction method based on a dynamic tag library involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the voice interaction device based on a dynamic tag library provided below can refer to the limitations of the voice interaction method based on a dynamic tag library above, and will not be repeated here.

[0098] In an exemplary embodiment, Figure 5 As shown, Figure 5 5 is a structural block diagram of a voice interaction device based on a dynamic tag library in an embodiment; the voice interaction device based on the dynamic tag library includes: a receiving module 501, an intention recognition module 502 and an execution module 503; The receiving module 501 is used to receive the voice interaction information sent by the user; The intention recognition module 502 is used to perform intention recognition on the voice interaction information based on a preset dynamic tag library, and determine the intention recognition result matching the voice interaction information; wherein the preset dynamic tag library includes a user behavior tag library and a dynamic content tag library; the user behavior tag library includes behavior tags related to user preferences and behavior habits; the dynamic content tag library includes dynamic content tags related to user preferences; The execution module 503 is used to execute an action sequence corresponding to the intention recognition result based on the intention recognition result.

[0099] The above-mentioned voice interaction device based on the dynamic tag library receives voice interaction information sent by the user; further, based on the preset dynamic tag library, the voice interaction information is subjected to intent recognition, and the intent recognition result matching the voice interaction information is determined; wherein, the preset dynamic tag library includes a user behavior tag library and a dynamic content tag library; the user behavior tag library includes behavior tags related to user preferences and behavior habits; the dynamic content tag library includes dynamic content tags related to user preferences; based on the preset dynamic tag library, not only the user's true intention can be accurately identified, but also the user's personalized preferences, historical behavior habits, the latest dynamic content, etc. can be taken into consideration, thereby providing personalized services that better meet user needs; finally, based on the intention recognition result, an action sequence corresponding to the intention recognition result is executed; based on this, the user's interaction intention can be understood more comprehensively and accurately to achieve precise interaction, thereby improving the personalization and accuracy of voice interaction.

[0100] In one embodiment, it also includes a user behavior tag library update module; the user behavior tag library update module is used to Obtain historical voice information corresponding to the target behavior within a preset update period; Determine multiple candidate behavior labels corresponding to the target behavior from the historical speech information; Determine the usage frequency corresponding to each of the multiple candidate behavior tags; Determine, from the plurality of candidate behavior tags, an updated behavior tag corresponding to the target behavior according to the usage frequencies corresponding to the plurality of candidate behavior tags; The user behavior tag library is dynamically updated according to the target behavior and the updated behavior tag corresponding to the target behavior.

[0101] In one embodiment, it also includes a dynamic content tag library update module; the dynamic content tag library update module is used to Obtain user preference information and historical interaction information; The dynamic content tag library is dynamically updated based on the preset dynamic event data set, user preference information, and historical interaction information.

[0102] In one embodiment, the dynamic content tag library update module is also used to Determine label weight information based on historical interaction information; According to the tag weight information, determining multiple target dynamic event information matching the user preference information from the preset dynamic event data set; The dynamic content tag library is dynamically updated according to multiple target dynamic event information.

[0103] In one embodiment, the user behavior tag library update module is also used to Get historical interaction information; Determine voice interaction instances that do not meet user expectations from historical interaction information; Analyze the behavior reasons of the voice interaction instance and determine the target behavior reasons corresponding to the voice interaction instance; Dynamically update the user behavior tag library based on the target behavior reasons.

[0104] In one embodiment, the intent recognition module 502 is also used to Determine a target tag that matches the voice interaction information from a preset dynamic tag library; If there is a target tag matching the voice interaction information in the preset dynamic tag library, then determining the intent recognition result corresponding to the target tag from the preset dynamic tag library; If there is no target tag matching the voice interaction information in the preset dynamic tag library, the voice interaction information is segmented into function words to obtain multiple target keywords; According to multiple target keywords, an intent recognition result corresponding to the voice interaction information is determined.

[0105] In one embodiment, the execution module 503 is used to If the intent recognition result is a professional domain intent, the vertical category model matching the professional domain intent is called to provide knowledge answers; If the intent recognition result is not a professional domain intention, the action sequence corresponding to the intent recognition result is executed.

[0106] Each module in the above-mentioned voice interaction device based on the dynamic tag library can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the terminal device in the form of hardware, or can be stored in the memory in the terminal device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0107] In an exemplary embodiment, a terminal device is provided. The terminal device may be a server, and its internal structure diagram may be as follows: Figure 6 As shown. The terminal device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the terminal device is used to provide computing and control capabilities. The memory of the terminal device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the terminal device is used to store voice interaction related data based on a dynamic tag library. The input / output interface of the terminal device is used to exchange information between the processor and an external device. The communication interface of the terminal device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a voice interaction method based on a dynamic tag library is implemented.

[0108] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the terminal device to which the scheme of the present application is applied. The specific terminal device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0109] In one embodiment, a terminal device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0110] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0111] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0112] 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 relevant regulations.

[0113] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0114] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0115] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the present application. It should be noted that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A voice interaction method based on a dynamic tag library, characterized in that: The method comprises: Receive voice interaction information sent by users; Based on a preset dynamic tag library, the voice interaction information is subjected to intent recognition, and an intent recognition result matching the voice interaction information is determined; wherein the preset dynamic tag library includes a user behavior tag library and a dynamic content tag library; the user behavior tag library includes behavior tags related to user preferences and behavior habits; the dynamic content tag library includes dynamic content tags related to user preferences; Based on the intention recognition result, an action sequence corresponding to the intention recognition result is executed.

2. The method according to claim 1, characterized in that The method further comprises: Obtain historical voice information corresponding to the target behavior within a preset update period; Determining a plurality of candidate behavior tags corresponding to the target behavior from the historical speech information; Determine the usage frequency corresponding to each of the plurality of candidate behavior tags; Determining an updated behavior tag corresponding to the target behavior from the plurality of candidate behavior tags according to the usage frequencies corresponding to the plurality of candidate behavior tags; The user behavior tag library is dynamically updated according to the target behavior and the updated behavior tag corresponding to the target behavior.

3. The method according to claim 1, characterized in that The method further comprises: Obtain user preference information and historical interaction information; The dynamic content tag library is dynamically updated according to a preset dynamic event data set, the user preference information and the historical interaction information.

4. The method according to claim 3, characterized in that: The dynamically updating the dynamic content tag library according to the preset dynamic event data set, the user preference information and the historical interaction information includes: Determining tag weight information according to the historical interaction information; According to the tag weight information, determining a plurality of target dynamic event information matching the user preference information from the preset dynamic event data set; The dynamic content tag library is dynamically updated according to the plurality of target dynamic event information.

5. The method according to claim 1, characterized in that The method further comprises: Get historical interaction information; Determining, from the historical interaction information, a voice interaction instance that does not meet user expectations; Performing a behavior cause analysis on the voice interaction instance to determine a target behavior cause corresponding to the voice interaction instance; The user behavior tag library is dynamically updated according to the target behavior reason.

6. The method according to claim 1, characterized in that The performing intent recognition on the voice interaction information based on a preset dynamic tag library and determining an intent recognition result matching the voice interaction information includes: Determining a target tag matching the voice interaction information from the preset dynamic tag library; If there is a target tag matching the voice interaction information in the preset dynamic tag library, determining an intent recognition result corresponding to the target tag from the preset dynamic tag library; If there is no target tag matching the voice interaction information in the preset dynamic tag library, performing function word segmentation processing on the voice interaction information to obtain multiple target keywords; According to the multiple target keywords, an intention recognition result corresponding to the voice interaction information is determined.

7. The method according to claim 1, characterized in that The executing, based on the intention recognition result, an action sequence corresponding to the intention recognition result comprises: If the intention recognition result is a professional field intention, a vertical category model matching the professional field intention is called to provide knowledge answers; If the intention recognition result is not a professional field intention, an action sequence corresponding to the intention recognition result is executed.

8. A terminal device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

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

Images