Cloud response system and method based on 5G message

Abstract

The invention relates to the technical field of data response, and discloses a cloud response system and method based on a 5G message. The system comprises a message legitimacy verification module, a semantic analysis module, a cloud response content fusion module, a dynamic return strategy generation module and an optimal cloud response content generation module, for a 5G message sent by a client, a message format is identified and mapped with a message transmission link to verify the legitimacy; after legal messages are uploaded to the cloud end, semantic vectors are obtained through algorithm analysis, response sub-contents are generated and fused into cloud end response contents, a return factor and a dynamic return strategy are returned to the client end, feedback semantics are extracted, and the optimal cloud end response contents are determined through a feedback annular strategy. According to the invention, the accuracy of cloud response of the 5G message can be improved.

Description

Technical Field

[0001] The present invention relates to the field of data response technology, and in particular to a cloud response system and method based on 5G messages. Background Art

[0002] With the continuous development and popularization of 5G technology, 5G messages have gradually been applied in various industries. The high speed, low latency and large capacity characteristics of 5G technology provide strong support for message interaction, making information transmission more efficient and real-time. Traditional message services are in urgent need of upgrading to match the advantages of 5G, which in turn has given rise to cloud-based responses for 5G messages to meet users' expectations for high-quality communications.

[0003] The existing cloud response technology for 5G messages uses deep learning algorithms, such as the Transformer architecture, to perform lexical, syntactic, and semantic analysis on 5G messages, understand user intent, and generate responses. In actual applications, when a response is wrong or unreasonable, it is difficult to accurately identify which analysis link has a problem because it is difficult to know the specific decision-making process within the model, which is not conducive to targeted improvement and optimization, resulting in low accuracy in cloud response to 5G messages. Summary of the invention

[0004] The present invention provides a cloud response system and method based on 5G messages, the main purpose of which is to solve the problem of low accuracy when performing cloud response to 5G messages.

[0005] To achieve the above-mentioned purpose, the present invention provides a cloud response system based on 5G messages, which includes a message legitimacy verification module, a semantic parsing module, a cloud response content fusion module, a dynamic return strategy generation module and an optimal cloud response content generation module, wherein:

[0006] The message legitimacy verification module is used to identify the message format of the 5G message corresponding to the pre-received client, map the message format with the pre-acquired message transmission link, and verify the message legitimacy of the 5G message according to the mapped message transmission link, and is specifically used to: determine the message element compliance factor according to the message content corresponding to the 5G message; determine the message transmission compliance factor according to the mapped message transmission link; verify the element compliance value corresponding to the message element compliance factor, and verify the transmission compliance value corresponding to the message transmission compliance factor; calculate the message legitimacy of the 5G message according to the element compliance value and the transmission compliance value using the following preset time decay algorithm:

[0007]

[0008] Among them, L is the legitimacy of the message, λ is the message attenuation coefficient, t is the message consumption time, xi is the compliance value of the element corresponding to the i-th compliance factor in the message compliance factor, y i is the transmission compliance value corresponding to the i-th compliance factor in the message transmission compliance factor, n is the total number of compliance factors in the message element compliance factor, and m is the total number of compliance factors in the message transmission compliance factor;

[0009] The semantic parsing module is used to upload the 5G message to the cloud according to the legitimacy of the message, and perform semantic parsing on the 5G message in the cloud according to a preset multi-dimensional algorithm to obtain a multi-dimensional semantic vector corresponding to the 5G message;

[0010] The cloud response content fusion module is used to identify the message pool corresponding to the 5G message through the multi-dimensional semantic vector, generate the response sub-content corresponding to the 5G message according to the message pool using a preset message collaboration algorithm, and fuse the response sub-content into the cloud response content corresponding to the 5G message;

[0011] The dynamic return strategy generation module is used to generate a return factor of the 5G message according to the cloud response content, generate a dynamic return strategy corresponding to the 5G message through the return factor, and return the cloud response content to the client according to the dynamic return strategy;

[0012] The optimal cloud response content generation module is used to extract the feedback semantics of the cloud response content transmitted back to the client, generate a feedback loop strategy based on the feedback semantics, and use the feedback loop strategy to generate the optimal cloud response content corresponding to the 5G message.

[0013] Optionally, when identifying the message format of a 5G message corresponding to a pre-received client, the message legitimacy verification module is specifically used to:

[0014] Parsing the message body content of the 5G message corresponding to the client received in advance;

[0015] Determine metadata information of the 5G message according to the message body content;

[0016] Parsing a message header field in the metadata information, and identifying a file type identifier in the message header field;

[0017] Identify the message format of the 5G message according to the file type identifier.

[0018] Optionally, when the semantic parsing module performs semantic parsing on the 5G message in the cloud according to a preset multi-dimensional algorithm to obtain a multi-dimensional semantic vector corresponding to the 5G message, it is specifically used to:

[0019] Calculate the topic probability distribution of the 5G message using the topic model in the multidimensional algorithm;

[0020] Determine the scenario type corresponding to the 5G message according to the topic probability distribution;

[0021] Calculate the intention probability distribution of the 5G message using the intention model in the multi-dimensional algorithm;

[0022] Determine the intent type corresponding to the 5G message according to the intent probability distribution;

[0023] The business type corresponding to the 5G message in the cloud database is extracted through the scene type and the intent type, and the scene type, the intent type and the business type are integrated into a multi-dimensional semantic vector corresponding to the 5G message.

[0024] Optionally, when the cloud response content fusion module generates the response sub-content corresponding to the 5G message according to the message pool using a preset message coordination algorithm, it is specifically used to:

[0025] Extract key information corresponding to the 5G message, determine a first business field of the 5G message according to the key information, and extract a second business field in the message pool;

[0026] Calculating the semantic association between the first business field and the second business field;

[0027] Constructing a business process fit matrix based on the first business area and the second business area;

[0028] The message response coordination degree in the message pool corresponding to the 5G message is calculated according to the semantic relevance and the business process fit matrix using the following message coordination algorithm:

[0029]

[0030] Among them, S k is the message response coordination degree of the kth message party in the message pool, ω1 is the semantic association weight, ω2 is the business process fit weight, A is the business field vector corresponding to the 5G message, and B k is the business domain vector corresponding to the kth message party in the message pool, f uv is the semantic correlation between the business domain u corresponding to the 5G message and the business domain v corresponding to the message party, p uv is the degree of fit in the business process fit matrix between the business domain u corresponding to the 5G message and the business domain v corresponding to the message party;

[0031] Determine the message responder in the message pool according to the message response coordination degree, and generate the response sub-content corresponding to the 5G message according to the message responder.

[0032] Optionally, when the cloud response content fusion module fuses the response sub-content into the cloud response content corresponding to the 5G message, it is specifically used to:

[0033] Identify the content format corresponding to the response sub-content;

[0034] Mapping the content format with a predefined content template to obtain a mapping relationship;

[0035] The response sub-content is embedded into the content template through the mapping relationship to obtain the cloud response content corresponding to the 5G message.

[0036] Optionally, when the dynamic backhaul strategy generation module generates the backhaul factor of the 5G message according to the cloud response content, it is specifically used to:

[0037] Detect the message reply completeness of the cloud response content to the 5G message, where the message reply completeness calculation formula is:

[0038]

[0039] Among them, Q is the completeness of the message reply, N is the number of sentences corresponding to the 5G message, and a h is the vector corresponding to the hth sentence in the 5G message, b d is the vector corresponding to the dth sentence in the cloud response content, and max is the maximum value function;

[0040] When the message reply integrity is equal to a preset integrity threshold, determining the return factor of the 5G message as a single-round return factor;

[0041] When the message reply completeness is not equal to a preset completeness threshold, detecting the message reply logical relevance of the cloud response content to the 5G message;

[0042] When the message reply logic correlation meets the preset correlation range, the return factor of the 5G message is determined as a multi-round return factor;

[0043] When the message reply logic correlation does not meet the preset correlation range, the return factor of the 5G message is determined as an abnormal return factor.

[0044] Optionally, when the dynamic backhaul strategy generation module generates the dynamic backhaul strategy corresponding to the 5G message through the backhaul factor, it is specifically used to:

[0045] Generating a return static strategy for the 5G message according to a single-round return factor in the return factor, and determining a return strategy for the 5G message through the return static strategy;

[0046] Identify the multi-round indicator change trend of the 5G message according to the multi-round return factor in the return factor, adjust the reply direction of the 5G message according to the multi-round indicator change trend, and generate a dynamic return strategy for the 5G message through the reply direction and the multi-round return time deadline;

[0047] According to the abnormal return factor in the return factor, the reply indicator corresponding to the 5G message is identified, the cloud response content is reply-reconstructed according to the reply indicator, and a dynamic return strategy for the 5G message is generated according to the reconstructed cloud response content.

[0048] Optionally, when extracting the feedback semantics of the cloud response content transmitted back to the client, the optimal cloud response content generation module is specifically used to:

[0049] Extracting keywords from the cloud response content transmitted back to the client;

[0050] Performing vector conversion on the keyword to obtain a keyword vector;

[0051] Determine the weight value of the keyword vector using a preset entropy method;

[0052] Constructing a semantic matrix of user feedback according to the keywords and the weight values;

[0053] The feedback semantics of the cloud response content is determined through the semantic matrix.

[0054] Optionally, when generating the feedback loop strategy according to the feedback semantics, the optimal cloud response content generation module is specifically used to:

[0055] identifying a feedback form of the feedback semantics, and determining a feedback level of the feedback semantics according to the feedback form;

[0056] When the feedback level meets the preset feedback level rule of the cloud response content, an open-loop control condition of the feedback semantics is generated;

[0057] When the feedback level does not meet the preset feedback level rule of the cloud response content, generating a closed-loop control condition of the feedback semantics;

[0058] The open-loop control condition and the closed-loop control condition are integrated into a feedback loop strategy.

[0059] In order to solve the above problems, the present invention also provides an operation method of a cloud response system based on 5G messages, the method comprising:

[0060] Identify the message format of the 5G message corresponding to the pre-received client, map the message format with the pre-acquired message transmission link, and verify the message legitimacy of the 5G message according to the mapped message transmission link;

[0061] Uploading the 5G message to the cloud according to the legitimacy of the message, and performing semantic analysis on the 5G message in the cloud according to a preset multi-dimensional algorithm to obtain a multi-dimensional semantic vector corresponding to the 5G message;

[0062] Identify the message pool corresponding to the 5G message through the multi-dimensional semantic vector, generate the response sub-content corresponding to the 5G message according to the message pool using a preset message collaboration algorithm, and merge the response sub-content into the cloud response content corresponding to the 5G message;

[0063] Generate a return factor of the 5G message according to the cloud response content, generate a dynamic return strategy corresponding to the 5G message through the return factor, and return the cloud response content to the client according to the dynamic return strategy;

[0064] Extract the feedback semantics of the cloud response content transmitted back to the client, generate a feedback loop strategy based on the feedback semantics, and use the feedback loop strategy to generate the optimal cloud response content corresponding to the 5G message.

[0065] The embodiment of the present invention uploads the message to the cloud according to the legitimacy of the message, avoiding the waste of cloud storage and computing resources by invalid and illegal messages; by using multi-dimensional semantic vectors to identify the message pool corresponding to the message, similar or related types of messages can be classified and managed; the response sub-content is merged into the cloud response content, and each sub-content is integrated and sorted out, and combined into a complete, standardized and organized reply according to certain rules and templates; by generating the return factor of the 5G message, the quality of the cloud response content can be quantitatively evaluated from multiple angles, and the cloud response content is returned to the client according to the dynamic return strategy, which fully considers the actual situation of the user receiving the reply; extracting the feedback semantics of the cloud response content returned to the client can deeply understand the user's real views on the reply, demand changes, and potential expectations; using the feedback loop strategy to generate the best cloud response content can meet user needs to the greatest extent and provide users with a high-quality 5G message interaction experience. Therefore, the cloud response system and method based on 5G messages proposed by the present invention can solve the problem of low accuracy of users when responding to 5G messages in the cloud. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A functional module diagram of a cloud response system based on 5G messages provided in one embodiment of the present invention;

[0067] Figure 2 A flowchart of an operating method of a cloud response system based on 5G messages provided in accordance with an embodiment of the present invention.

[0068] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings, and "multiple" generally includes at least two.

[0071] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0072] In addition, the step sequence in the following method embodiments is only an example and not a strict limitation.

[0073] In fact, the server-side device deployed by the cloud response system based on 5G messages may be composed of one or more devices. The above-mentioned cloud response system based on 5G messages can be implemented as: business instances, virtual machines, hardware devices. For example, the cloud response system based on 5G messages can be implemented as a business instance deployed on one or more devices in the cloud node. In simple terms, the cloud response system based on 5G messages can be understood as a software deployed on the cloud node, which is used to provide a cloud response system based on 5G messages for each user terminal. Alternatively, the cloud response system based on 5G messages can also be implemented as a virtual machine deployed on one or more devices in the cloud node. The virtual machine is installed with application software for managing each user terminal. Alternatively, the cloud response system based on 5G messages can also be implemented as a server composed of many hardware devices of the same or different types, and one or more hardware devices are set to provide a cloud response system based on 5G messages for each user terminal.

[0074] In terms of implementation, the cloud response system based on 5G messages and the user end are adapted to each other. That is, the cloud response system based on 5G messages is an application installed on the cloud service platform, and the user end is a client that establishes a communication connection with the application; or the cloud response system based on 5G messages is implemented as a website, and the user end is implemented as a web page; or the cloud response system based on 5G messages is implemented as a cloud service platform, and the user end is implemented as a small program in the instant messaging application.

[0075] Reference Figure 1 As shown, it is a functional module diagram of a cloud response system based on 5G messages provided by one embodiment of the present invention.

[0076] The cloud response system 100 based on 5G messages described in the present invention can be set in a cloud server. In terms of implementation, it can be used as one or more service devices, or it can be installed as an application on the cloud (such as a server of a mobile service operator, a server cluster, etc.), or it can be developed as a website. According to the functions implemented, the cloud response system 100 based on 5G messages can include a message legitimacy verification module 101, a semantic parsing module 102, a cloud response content fusion module 103, a dynamic return strategy generation module 104 and an optimal cloud response content generation module 105. The module described in the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0077] In an embodiment of the present invention, in a cloud response system based on 5G messages, each of the above modules can be implemented independently and called with other modules. The call here can be understood as a module that can connect to multiple modules of another type and provide corresponding services to the multiple modules connected to it. For example, the cloud response content fusion module can call the semantic parsing module to obtain the information collected by the semantic parsing module. Based on the above characteristics, in the cloud response system based on 5G messages provided by an embodiment of the present invention, the scope of application of the cloud response system architecture based on 5G messages can be adjusted by adding modules and directly calling them without modifying the program code, so as to achieve cluster-based horizontal expansion, so as to achieve the purpose of quickly and flexibly expanding the cloud response system based on 5G messages. In actual applications, the above modules can be set in the same device or different devices, or they can be set in virtual devices, such as service instances in cloud servers.

[0078] The following is a description of each component and specific workflow of the cloud response system based on 5G messages in conjunction with a specific embodiment:

[0079] The message legitimacy verification module 101 is used to identify the message format of a 5G message corresponding to a pre-received client, map the message format with a pre-acquired message transmission link, and verify the message legitimacy of the 5G message based on the mapped message transmission link.

[0080] In an embodiment of the present invention, after receiving a 5G message from a client, the message body content carries the core information that the user wants to convey, and may include data in various forms such as text, pictures, videos, links, etc., and the message format refers to the specific specifications and structural forms followed by the 5G message in data organization, encoding, and presentation, such as text format, picture format, video format, etc.

[0081] In the embodiment of the present invention, when identifying the message format of the 5G message corresponding to the client received in advance, the message legitimacy verification module 101 is specifically used to:

[0082] Parsing the message body content of the 5G message corresponding to the client received in advance;

[0083] Determine metadata information of the 5G message according to the message body content;

[0084] Parsing a message header field in the metadata information, and identifying a file type identifier in the message header field;

[0085] Identify the message format of the 5G message according to the file type identifier.

[0086] In detail, natural language processing technology is used to perform operations such as word segmentation, part-of-speech tagging, and named entity recognition on the text to structure the text content. For multimedia data (pictures, videos), corresponding image recognition, video parsing and other technologies are used to extract key features or obtain the text information contained therein (such as subtitles in videos, etc.). If the main content of the message is a text description introducing a product, the natural language processing technology can be used to extract the product name, attributes, and descriptive words involved. If it is a product picture, image recognition technology can identify key elements such as the product body, color, and style in the picture, so that the metadata information of the 5G message can be determined based on the main content of the message. Metadata information is data that describes the characteristics and related attributes of the 5G message itself, and understands the 5G message from multiple dimensions, such as the message The source of the message (the identifier of the sending client, etc.), sending time, message size, subject classification, and some business-related attributes (such as whether it is a consulting message, whether it is associated with a specific business process, etc.) are determined by parsing the message body content and combining some additional information attached during the message transmission process (such as some metadata initial identifiers that may be included in the message header, etc.). For example, the approximate subject classification of the message (product consultation, service complaint, or other types) can be inferred from the message body content through keyword analysis, semantic judgment, etc.; at the same time, when the message is transmitted on the network, the sender may attach the sending time, some basic identifiers of the sending client, and other information in the message header or according to a specific protocol. This information is collected and integrated to form the metadata information of the 5G message.

[0087] Specifically, the message header field is part of the metadata information, located at the beginning of the message, which is equivalent to an information index and contains many key identifiers used to describe the basic characteristics and format of the message. That is, according to the established protocol or specification, the message header field is parsed through data extraction, field matching and other technical means. If a message header specification based on XML format definition is adopted, the XML parser can be used to extract the specific field values ​​according to the definition of elements such as tags and attributes, and find out the field contents related to the file type that need attention, so as to identify the file type identifier in the message header field. The file type identifier is an element of the message header field, which indicates the overall data organization form of the 5G message, such as .doc for Word document and .jpg for JPEG image file. Then, the message format of the 5G message is identified according to the file type. For example, if the file type identifier is .doc, the message format of the 5G message is in text form; if the file type identifier is .jpg, the message format of the 5G message is in image form.

[0088] Furthermore, in order to ensure that the message can achieve the best performance during the transmission process and ensure that it reaches the receiving end from the sending end in an efficient, accurate and stable manner, it is necessary to accurately map the message format and the message transmission link, and make each message format uniquely correspond to a message transmission link, map the message format with the pre-acquired message transmission link, and each message format has a unique corresponding message transmission link. The mapping relationship between each established message format and the unique corresponding message transmission link is stored. A database method can be used, for example, to create a mapping table containing a "message format" field and a "message transmission link identifier" field, such as {"message format": "JSON", "transmission link identifier": "mobile application interactive link"}. By specifying a unique corresponding message transmission link for each message format, the resources of the link can be fully utilized according to the characteristics of the message format, such as achieving the best match in terms of bandwidth adaptation and delay reduction, thereby improving the overall efficiency of message transmission, reducing unnecessary resource waste, and enabling messages to reach the receiving end faster.

[0089] Furthermore, the channel has a clear definition of the legitimacy of the message. In order to ensure that 5G messages strictly follow the rules required by the corresponding link, such as the format of the message header, the standard filling of metadata, the limitation of data types, etc., so that the entire communication system can operate in an established order, avoid the mixing of messages that do not comply with the rules and disrupt the normal operation of the system, and ensure the standardization and stability of the communication process, it is necessary to verify the legitimacy of the 5G messages.

[0090] In an embodiment of the present invention, the message legitimacy refers to a comprehensive measurement indicator, which is used to comprehensively evaluate the legitimacy of 5G messages in terms of content elements and transmission process. It reflects the degree to which the message is sent, transmitted and received in accordance with established rules and requirements in the entire communication system. The higher the legitimacy of the message, the more compliant the message is with the specification, and it can flow more smoothly and reliably in the system and be correctly processed; conversely, low legitimacy means that there are many places in the message that do not meet the requirements.

[0091] In the embodiment of the present invention, when the message legitimacy verification module 101 verifies the message legitimacy of the 5G message according to the mapped message transmission link, it is specifically used to:

[0092] Determine a message element compliance factor according to the message content corresponding to the 5G message;

[0093] Determine a message transmission compliance factor according to the mapped message transmission link;

[0094] Verifying the element compliance value corresponding to the message element compliance factor, and verifying the transmission compliance value corresponding to the message transmission compliance factor;

[0095] The message legitimacy of the 5G message is calculated according to the element compliance value and the transmission compliance value using the following preset time decay algorithm:

[0096]

[0097] Among them, L is the legitimacy of the message, λ is the message attenuation coefficient, t is the message consumption time, x i is the compliance value of the element corresponding to the i-th compliance factor in the message compliance factor, y i is the transmission compliance value corresponding to the i-th compliance factor in the message transmission compliance factor, n is the total number of compliance factors in the message element compliance factor, and m is the total number of compliance factors in the message transmission compliance factor.

[0098] In detail, the message element compliance factor refers to the message format compliance factor and the message content compliance factor, and the message transmission compliance factor refers to the identity authentication factor and the link security factor. The message format compliance factor is based on the pre-defined 5G message standard format to determine whether the message format is compliant. The compliance value is 1 and the non-compliance value is 0. The message content compliance factor uses natural language processing technology combined with the compliance vocabulary and rule library to determine whether the message content has any violations (such as illegality, violation of platform regulations, etc.). The value of no violation is 1, and the value of violation is proportional to the severity of the violation. The identity authentication factor is determined by verifying the identity of the sender (such as the match between the mobile phone number, the unique device identifier, etc. and the pre-stored legal user information). The value range is 0, 1. If it is a complete match, it is 1, and if it is not matched, it is 0. The link security factor is determined based on the assessment of the security of the message transmission link (such as the link encryption status, whether there are signs of attack, etc.), and the value range is 0, 1. If the security situation is good, it is 1, and if there are security risks, the value is reduced proportionally.

[0099] Specifically, by incorporating the timeliness factor into the comprehensive evaluation of the legitimacy of the message, it is possible to more closely match the actual application scenario to determine whether the message is compliant and valid. Let the initial legitimacy of the message (assuming that the legitimacy is not considered when the message is just sent) be L0 (its calculation can be obtained by weighting multiple factors when the timeliness is not considered in the previous example), the attenuation coefficient of the message be λ (determined according to factors such as business importance and message type, with a small attenuation coefficient for important messages and a large attenuation coefficient for less important messages), and the elapsed time be t (from sending to the current verification moment). The timeliness factor is represented by an exponential decay function. Then, the comprehensive calculation results of the identity authentication factor, message format compliance factor, content compliance factor, and link security factor are multiplied by the timeliness factor to reflect the impact of timeliness on the overall legitimacy, thereby obtaining the legitimacy of the 5G message.

[0100] Furthermore, as a platform for centralized storage and processing of a large amount of data, the cloud depends on high-quality data for subsequent operations such as analysis and response generation. By screening 5G messages according to the message legality, only messages with a qualified legality are uploaded to the cloud, which can effectively prevent illegal, incorrect, or non-standard format messages from being mixed in. Screening according to the legality ensures that the uploaded messages are accurate and reliable in content, helps to maintain the overall accuracy of the cloud data, and guarantees the normal operation of various services based on the cloud data.

[0101] The semantic parsing module 102 is used to upload the 5G message to the cloud according to the message legality, and perform semantic parsing on the 5G messages in the cloud according to a preset multi-dimensional algorithm to obtain a multi-dimensional semantic vector corresponding to the 5G message.

[0102] In the embodiment of the present invention, uploading the 5G message to the cloud according to the message legality means that when the legality L = 2n, it means that the message is very legal, and the 5G message can be directly uploaded to the cloud through the transmission interface of the cloud data; when the legality L = 0, it means that the message is not legal at all, and there is no need to upload this 5G message to the cloud; when 0 < L < 2n, it means that the message is partially illegal or the degree of unimportance of the message decay over time, which is determined according to factors such as business importance and message type. The decay coefficient of important messages is small, and the decay coefficient of less important messages is large. In the case of partial illegality, the message may have certain problems in some element compliance factors or transmission compliance factors. And the situation where the message decays to less important over time is often because the business involved in the message has strong timeliness. As time goes by, even if the message was originally relatively compliant, but due to the change of the business scenario (for example, the time-limited discount activity has expired, and the importance of related consultation messages has decreased), its legality also decreases accordingly. If the problem is not too serious and can meet the basic requirements after appropriate processing, it can still be considered for uploading to the cloud for subsequent processing, but it needs to be marked or given special attention for subsequent further verification and improvement of relevant information.

[0103] Furthermore, in order to accurately respond to 5G messages, it is necessary to parse the 5G messages to parse out rich and accurate semantic information, so that more accurate response reply content that fits the user's intention and needs can be generated.

[0104] In an embodiment of the present invention, the multi-dimensional semantic vector is a structured representation of the semantic information of the 5G message, which combines semantic information of multiple dimensions such as scenario type, intent type and business type in the form of a vector. For example, for a consumer electronics product consultation message, its multi-dimensional semantic vector may be expressed as ["consumer electronics product sales scenario", "information inquiry", "consumer electronics product information query service"]. The different dimensions of this vector represent different levels of semantic information, and together constitute a comprehensive description of the semantics of the 5G message, providing a comprehensive and structured semantic foundation for subsequent message processing, business operations, intelligent reply generation, etc., so that the system can process at different dimensions based on the multi-dimensional semantic vector to meet different business needs.

[0105] In the embodiment of the present invention, when the semantic parsing module 102 performs semantic parsing on the 5G message in the cloud according to a preset multi-dimensional algorithm to obtain a multi-dimensional semantic vector corresponding to the 5G message, it is specifically used to:

[0106] Calculate the topic probability distribution of the 5G message using the topic model in the multidimensional algorithm;

[0107] Determine the scenario type corresponding to the 5G message according to the topic probability distribution;

[0108] Calculate the intention probability distribution of the 5G message using the intention model in the multi-dimensional algorithm;

[0109] Determine the intent type corresponding to the 5G message according to the intent probability distribution;

[0110] The business type corresponding to the 5G message in the cloud database is extracted through the scene type and the intent type, and the scene type, the intent type and the business type are integrated into a multi-dimensional semantic vector corresponding to the 5G message.

[0111] In detail, a topic model is a statistical model used to discover potential topics in a document collection, such as Latent Dirichlet Allocation (LDA). When processing 5G messages, the message is regarded as a bag of words, that is, each word in the message is regarded as an independent element, regardless of its order. By analyzing the words in the message, the topic model can query the potential topics behind these words and calculate the probability distribution of each topic appearing in the message. For example, for a 5G message about buying a new mobile phone with a powerful camera function and a good screen display, the topic model may identify potential topics such as electronic products and mobile phone functions, and calculate the probability distribution of electricity. The probability of the sub-product topic in the message is 0.6, the probability of the mobile phone function is 0.4, etc., and the topic corresponding to the highest probability of the topic probability distribution is selected as the scenario type corresponding to the 5G message. For example, if the electronic product topic has the highest probability, then this scenario type is a consumer electronic product sales scenario. For example, when the probability of the topic of the message being related to consumer products is high, it is classified as a consumer scenario according to the rules; when the topic is highly relevant to financial products and services, it is classified as a financial service scenario, that is, the relatively microscopic topic information is promoted to a scenario category level that is more suitable for business operations and system processing, so as to better manage and process different types of messages.

[0112] Specifically, the intent model aims to identify the potential intentions of users in 5G messages. The model analyzes the semantic features of messages and uses machine learning or deep learning technology to classify user intentions and calculate the probability distribution of each intention. For example, for a message like "How long is the battery life of this phone?", the intent model may calculate that the probability of information inquiry intention is 0.8, the probability of complaint intention is 0.1, and the probability of suggestion intention is 0.1. Because the message is mainly asking for information about the battery life of the phone, the intention of asking for information is the most obvious. The intent types include consultation, request, feedback, complaint, purchase, etc. The intent type of the 5G message is determined by selecting the intent category corresponding to the maximum probability in the intent probability distribution. For example, if the probability of information inquiry is the highest in the intent probability distribution of a message, then the intent type of the message is information inquiry. The intent probability distribution can clearly show the main and secondary purposes of users when sending 5G messages, avoid misjudgment of user intentions, and improve the pertinence and effectiveness of services.

[0113] Furthermore, according to the determined scenario type and intent type, the matching business type is searched from the cloud database. The cloud database stores business information under different combinations of scenario types and intent types. For example, in the consumer electronics sales scenario, when the user's intention is to inquire about information, the corresponding business type may be the consumer electronics information query business; in the medical service scenario, for the complaint feedback intention, it may correspond to the medical service complaint processing business. The business type is to further refine the scenario and intent to the specific business operation level, and more directly relate to the actual business process and processing logic, so as to combine the semantic information of multiple dimensions such as scenario type, intent type and business type in the form of vectors to obtain a multi-dimensional semantic vector, which converts the 5G message from the original text form into a multi-dimensional semantic vector with rich semantic information, and provides important semantic support for the efficient processing, precise service and intelligent decision-making of 5G messages in different business scenarios.

[0114] Furthermore, for the multi-dimensional semantic vector, it is necessary to query the corresponding response semantics in the cloud. Therefore, it is necessary to query data in the cloud in order to respond to the multi-dimensional semantic vector and accurately respond to the 5G message.

[0115] The cloud response content fusion module 103 is used to identify the message pool corresponding to the 5G message through the multi-dimensional semantic vector, generate the response sub-content corresponding to the 5G message according to the message pool using a preset message collaboration algorithm, and fuse the response sub-content into the cloud response content corresponding to the 5G message.

[0116] In the embodiment of the present invention, the message pool is a logical concept for classifying, managing and storing messages. It is a collection formed by gathering many messages together according to certain rules and characteristics. The message pool corresponding to the 5G message is identified by the multi-dimensional semantic vector, that is, the various types of databases corresponding to the cloud database are identified. The cloud database is a centralized storage system for storing a large amount of business data. It contains multiple different types of databases. These databases are classified according to the nature and purpose of the data. Common ones include user databases (storing user's basic information, personal information, preferences and other data), product databases (covering various attributes, specifications, inventory and other information of products), transaction record databases (recording user's transaction behavior, transaction time, amount and other transaction-related data), etc. Different types of databases serve different business needs and are the data basis for the normal operation of the entire business system.

[0117] For example, for messages inquiring about product inventory, the current quantity information must be obtained from the product inventory database; for messages inquiring about users' historical transactions, the corresponding data must be found in the transaction record database. When the intention of the 5G message is to inquire about the inventory status of a certain product, through multi-dimensional semantic vector analysis, it is known that it involves product-related business, then it can be clearly determined that the product database needs to be accessed to obtain the corresponding data; if it is to inquire about the user's historical purchase records, based on the user-related business needs reflected by the semantic vector, it can be determined that data must be found in the transaction record database, and the database related to message processing can be accurately located to avoid blind searches in irrelevant data, thereby improving the efficiency and accuracy of data queries, and integrating and sorting the data obtained from the query to prepare for generating response content.

[0118] Furthermore, for the message response content of the 5G message, it is necessary to extract the relevant response content corresponding to the 5G message in the message pool in the cloud, so it is necessary to coordinate multiple messages in the message pool to obtain the complete response content of the 5G message.

[0119] In an embodiment of the present invention, the response sub-content refers to the response data corresponding to each database in the message pool, which is a partial reply content corresponding to the 5G message generated by a determined message replying party based on its own business information, past reply experience, and relevant business rules.

[0120] In the embodiment of the present invention, when the cloud response content fusion module 103 generates the response sub-content corresponding to the 5G message according to the message pool using a preset message coordination algorithm, it is specifically used to:

[0121] Extract key information corresponding to the 5G message, determine a first business field of the 5G message according to the key information, and extract a second business field in the message pool;

[0122] Calculating the semantic association between the first business field and the second business field;

[0123] Constructing a business process fit matrix based on the first business area and the second business area;

[0124] The message response coordination degree in the message pool corresponding to the 5G message is calculated according to the semantic relevance and the business process fit matrix using the following message coordination algorithm:

[0125]

[0126] Among them, S k is the message response coordination degree of the kth message party in the message pool, ω1 is the semantic association weight, ω2 is the business process fit weight, A is the business field vector corresponding to the 5G message, and Bk is the business domain vector corresponding to the kth message party in the message pool, f uv is the semantic correlation between the business domain u corresponding to the 5G message and the business domain v corresponding to the message party, p uv is the degree of fit in the business process fit matrix between the business domain u corresponding to the 5G message and the business domain v corresponding to the message party;

[0127] Determine the message responder in the message pool according to the message response coordination degree, and generate the response sub-content corresponding to the 5G message according to the message responder.

[0128] In detail, for the received 5G messages, it is necessary to use natural language processing technology (such as word segmentation, named entity recognition, keyword extraction, etc.) to extract key information from them. The key information can reflect the core theme and main focus of the message. For example, a 5G message is "I want to know where the repair service outlets for this mobile phone are." Through extraction, key information such as mobile phones and repair service outlets can be obtained. Then, the first business field to which the 5G message belongs is determined based on the key information. In this case, it is the after-sales repair service field of mobile phones, which clarifies the approximate business scope involved in the message. For the messages in the message pool, it is also necessary to analyze the business-related content covered by each of them, and then extract the corresponding second business field. The message pool brings together many messages of the same or related types, and the business scope involved in each message is different. For example, the second business field includes mobile phone sales, mobile phone accessories, and after-sales repair fields.

[0129] Specifically, the semantic association between the first business field and the second business field is calculated. The semantic association is an indicator that measures the closeness or correlation between two business fields at the semantic level. For example, there may be a certain semantic association between the mobile phone after-sales repair service field and the mobile phone accessories sales field, because repairs may require corresponding accessories; while the semantic association with the home appliance sales field is relatively weak, wherein the semantic association can be calculated by word vector models (such as Word2Vec, BERT, etc.) and semantic similarity calculation methods, and the key words in the business field are converted into vector representations. The degree of semantic association is measured by calculating indicators such as cosine similarity and Euclidean distance between vectors; the business process fit matrix is ​​a matrix structure used to represent the degree of fit between the business field of 5G messages and the business field corresponding to each message in the message pool in terms of business process. The fit between business fields u and v in the actual business process is analyzed, and a business process fit matrix P is constructed, whose element p uv Determine in the following way: If business domains u and v are directly related in the business process (for example, the ordering link in the 5G message business is closely linked to the logistics delivery link in the partner business), and the degree of connection is high, set p uv =1; if there is a certain correlation but the degree of correlation is average, let puv =0.5; if there is basically no correlation, set p uv =0.

[0130] Furthermore, the coordination degree S is calculated by combining the semantic relevance and business process fit. k , using the weighted summation method, assuming that the weight of semantic relevance is ω1, the weight of business process fit is ω2, and ω1+ω2=1, The correlation between business areas is measured from the perspective of semantic understanding. Considering the actual fit of the business process, the coordination degree is obtained after weighting the two, which can more comprehensively reflect the degree of coordination between the 5G message business field and the partner business field. The coordination degree value range is 0, 1, generally ω1 = 0.5, ω2 = 0.5, and the two are considered to be equally important. The A / B test method can be used to set up multiple experimental groups with different weights in the actual business environment to handle the 5G message reply task. For example, for messages in which users consult about logistics and delivery information, the semantic relevance weight and the business process fit weight are set to different values, such as ω1 = 0.3, ω2 = 0.7 for group 1; ω1 = 0.5, ω2 = 0.5 for group 2; ω1 = 0.7, ω2 = 0.3 for group 3, etc., and then observe the reply effect indicators under different groups, and compare and analyze which group of weight combinations can enable users to obtain effective information faster, be more satisfied with the reply, and have fewer follow-up consultations. After a period of testing and data collection, the final weight value is determined based on the group with the best performance.

[0131] Furthermore, by comparing the message response coordination degree of each message party in the message pool, the top three message parties in terms of coordination are selected as message reply parties, which have the best match with the 5G message in terms of semantics and business processes. Their reply is more likely to generate content that is logical and meets user needs. The determined message reply party generates partial reply content corresponding to the 5G message based on its own business information, past reply experience and related business rules. For example, the message reply party may be a knowledge base containing a lot of mobile phone after-sales repair service knowledge or a collection of historical message records that have dealt with similar problems before. Specific information about the location of mobile phone repair service outlets will be extracted from it and organized into a text description. For example, the mobile phone repair service outlet in your area is located at No. XX, XX Street, XX District, and you can go there for consultation during working hours. This is the response sub-content generated for the 5G message. There may be other response sub-contents integrated in the future to form a complete reply content sent to the user.

[0132] In an embodiment of the present invention, the cloud response content refers to the complete reply content finally generated in a cloud-based 5G message processing system for a 5G message sent by a user after a series of analysis, data query, content generation and fusion operations. It is a combination of multiple response sub-contents and is integrated according to certain specifications and logic, and is intended to accurately, comprehensively and systematically respond to users' inquiries, requests or feedback.

[0133] In the embodiment of the present invention, when the cloud response content fusion module 103 fuses the response sub-content into the cloud response content corresponding to the 5G message, it is specifically used to:

[0134] Identify the content format corresponding to the response sub-content;

[0135] Mapping the content format with a predefined content template to obtain a mapping relationship;

[0136] The response sub-content is embedded into the content template through the mapping relationship to obtain the cloud response content corresponding to the 5G message.

[0137] In detail, the content formats corresponding to the response sub-content include text, images, videos, links, etc., and the predefined content template refers to the corresponding positions of text, images, videos, and link data. The corresponding positions of text, links, images, and videos are configured in the content template in order from top to bottom.

[0138] Exemplarily, if the corresponding position of the text data in the content template is at position A, the corresponding position of the link is at position B, the corresponding position of the image is at position C, and the corresponding position of the video is at position D, and if the content format corresponding to the response sub-content is text type or image type, then the response sub-content is embedded in positions A and C, and positions B and D are configured as blank areas, thereby obtaining the cloud response content corresponding to the 5G message.

[0139] Furthermore, the cloud response content is a reply to the 5G message sent by the user. Whether it accurately answers the user's questions and meets the user's needs is crucial. In order to measure the accuracy of the message reply, it is necessary to measure from multiple dimensions to ensure that high-quality cloud response content can be generated stably and efficiently for various users and business scenarios.

[0140] The dynamic return strategy generation module 104 is used to generate a return factor of the 5G message according to the cloud response content, generate a dynamic return strategy corresponding to the 5G message through the return factor, and return the cloud response content to the client according to the dynamic return strategy.

[0141] In an embodiment of the present invention, the feedback factor is a set of indicators that comprehensively evaluates the cloud's response to 5G messages. It reflects the quality, characteristics and adaptation of the response to user messages from multiple angles through different types (single-round feedback factor, multi-round feedback factor, abnormal feedback factor) and their corresponding measurement dimensions (such as response completeness, logical relevance, etc.).

[0142] In the embodiment of the present invention, when the dynamic backhaul strategy generation module 104 generates the backhaul factor of the 5G message according to the cloud response content, it is specifically used to:

[0143] Detect the message reply completeness of the cloud response content to the 5G message, where the message reply completeness calculation formula is:

[0144]

[0145] Among them, Q is the completeness of the message reply, N is the number of sentences corresponding to the 5G message, and a h is the vector corresponding to the hth sentence in the 5G message, b d is the vector corresponding to the dth sentence in the cloud response content, and max is the maximum value function;

[0146] When the message reply integrity is equal to a preset integrity threshold, determining the return factor of the 5G message as a single-round return factor;

[0147] When the message reply completeness is not equal to a preset completeness threshold, detecting the message reply logical relevance of the cloud response content to the 5G message;

[0148] When the message reply logic correlation meets the preset correlation range, the return factor of the 5G message is determined as a multi-round return factor;

[0149] When the message reply logic correlation does not meet the preset correlation range, the return factor of the 5G message is determined as an abnormal return factor.

[0150] In detail, the message reply completeness is an indicator to measure the comprehensiveness of the cloud response content to the 5G message sent by the user. It reflects whether the reply content covers the key information and key points involved in the 5G message. If the reply completeness is high, it means that the user's question is likely to be answered more comprehensively, which can meet the user's need to obtain information; conversely, if the completeness is low, there may be insufficient replies and omissions of important content, affecting the user experience and satisfaction with the reply. The cloud response content comprehensively considers the coverage of the information conveyed by each sentence in the 5G message, and then the overall reply completeness is obtained. If the message reply is When the completeness Q is equal to the preset completeness threshold, where the completeness threshold is 1, it means that the cloud response content meets the established standards in terms of completeness, and the feedback factor is determined as a single-round feedback factor. The round-trip feedback factor usually means that in this round of 5G message interaction, the cloud's reply to the user's message has met the expected completeness requirements to a certain extent. From the perspective of reply completeness, this reply can be regarded as a relatively independent and relatively complete response to the user's message, and no additional rounds of interaction are required to supplement information or further explain it, reflecting a more ideal single-round reply state, that is, one round of reply can basically meet the user's current information needs.

[0151] Specifically, when the completeness of the message reply does not meet the threshold requirement, it means that the reply may be missing information. At this time, further detection of logical relevance is to determine from a logical perspective whether the reply content is logically reasonable and whether it has an effective connection with the user's question even if it is incomplete, so as to more comprehensively evaluate the quality and characteristics of the reply. For example, although the reply content does not cover all the points of concern to the user, the part of the answer is logically coherent with the user's question and can reasonably respond to the user's question. In this way, the logical relevance will be reflected to a certain extent; on the contrary, if the reply content is pieced together and has no logical connection with the user's question, even if there is some content, it will be difficult for the user to understand and accept it. The detection of logical relevance can be achieved through a variety of natural language processing technologies and logical analysis methods, such as analyzing the causal relationship, progressive relationship, parallel relationship and other logical connections between sentences in the reply content. Whether the connection is reasonable and whether it conforms to normal language expression and business logic, the similarity between the question and the reply is calculated, and whether the logical structure of the reply content matches the question logic of the 5G message is judged based on the similarity. When it is detected that the logical correlation of the message reply meets the preset correlation range, where the correlation range is set according to the actual business situation, past interaction data, etc., such as being set between [0.6, 0.8], it means that the reply content is logically acceptable, but it may require multiple rounds of communication to supplement the complete information, that is, its return factor is determined as a multi-round return factor. The multi-round return factor indicates that although the cloud's reply to the 5G message may not meet the standard in terms of the completeness of a single round of reply, from the perspective of logical correlation, the reply content has a reasonable logical connection with the user message. It may be due to the large amount of information and complex questions that multiple rounds of interaction are needed to gradually improve the reply content and further answer the user's questions.

[0152] Furthermore, when the logical correlation of the message reply is not within the preset correlation range, it means that the reply content has serious logical flaws, cannot effectively connect and respond to the user's message, and does not conform to the normal interaction logic. The return factor is determined as an abnormal return factor. The abnormal return factor means that the cloud response content not only has problems with the completeness of the reply, but also does not meet the requirements in terms of logical correlation. That is, there is a lack of reasonable logical connection between the reply content and the user's 5G message, and serious situations such as irrelevant answers and logical confusion may occur, indicating that there is a major abnormality in this reply.

[0153] Furthermore, the process and judgment method of generating 5G message feedback factors based on cloud response content can classify and evaluate the reply situations more carefully and accurately. Therefore, different feedback strategies need to be generated based on different feedback factors to make the message response more accurate.

[0154] In an embodiment of the present invention, the dynamic backhaul strategy is a set of strategies that flexibly and dynamically adjusts the reply content, reply direction, reply time arrangement, etc. according to the actual situation in the 5G message reply process and based on the reply quality, characteristics and other information reflected by the backhaul factor. It is not static, but will be continuously optimized and adjusted with the changes in each round of reply situations (such as the trend of changes in multiple rounds of indicators, whether abnormalities occur, etc.).

[0155] In the embodiment of the present invention, when the dynamic backhaul strategy generating module 104 generates the dynamic backhaul strategy corresponding to the 5G message through the backhaul factor, it is specifically used to:

[0156] Generating a return static strategy for the 5G message according to a single-round return factor in the return factor, and determining a return strategy for the 5G message through the return static strategy;

[0157] Identify the multi-round indicator change trend of the 5G message according to the multi-round return factor in the return factor, adjust the reply direction of the 5G message according to the multi-round indicator change trend, and generate a dynamic return strategy for the 5G message through the reply direction and the multi-round return time deadline;

[0158] According to the abnormal return factor in the return factor, the reply indicator corresponding to the 5G message is identified, the cloud response content is reply-reconstructed according to the reply indicator, and a dynamic return strategy for the 5G message is generated according to the reconstructed cloud response content.

[0159] In detail, the single-round feedback factor reflects the response completeness reaching the standard. Since the single-round response can already meet the user's needs well, the feedback static strategy aims to consolidate this response effect, further improve the response from some details, and at the same time maintain the current response status, waiting for the user's next message interaction or ending the current interaction process. After the feedback static strategy is generated, it is directly determined as the feedback strategy of the 5G message, because in the scenario corresponding to the single-round feedback factor, the feedback static strategy can already adapt well to the current response situation. Subsequent processing according to this strategy can ensure the quality of the response, so that the user receives satisfactory response content and realizes a relatively complete and effective message interaction. The trend of multi-round indicators refers to the changes in key indicators (such as response completeness, logical relevance, user emotional tendency, etc.) as the number of interaction rounds increases during the multi-round 5G message interaction process. The identification of multi-round indicator change trends can be achieved by recording the feedback factor-related indicator values ​​corresponding to each round of responses, and analyzing the increase, decrease, and fluctuation of these values ​​between different rounds. Adjusting the reply direction based on the identified trend of multi-round indicator changes means adjusting the emphasis, degree of detail, and logical guidance of the subsequent reply content based on the information reflected by the indicator changes. If it is found that the completeness of the reply is gradually improving, but the logical relevance is declining, it may mean that although the reply content is becoming more and more comprehensive, it is logically not clear enough. In this case, the reply direction needs to focus on strengthening the logical sorting so that the subsequent replies can maintain a reasonable and close logical connection with the user's questions while continuing to supplement information. If the user's emotional tendency has always been less than ideal, the reply direction should consider more about how to enhance emotional resonance in language expression to improve the user's emotional state, ensure that the reply can better meet the user's expectations and needs, and promote multi-round interactions in a positive direction. The reply direction determines the general direction of the subsequent reply content, and the multi-round feedback deadline is a time limit factor. Considering that multi-round interactions cannot continue indefinitely, it is necessary to complete effective answers to user questions within a certain time range to ensure interaction efficiency and user experience.

[0160] Specifically, when the feedback factor is an abnormal feedback factor, it indicates that there are serious problems with the cloud response content, such as the extremely low logical correlation between the reply content and the user's 5G message, the serious lack of reply completeness, etc. Based on the identified reply indicators, the cloud response content is reconstructed to modify and improve the problems in a targeted manner. If a large number of key information are missing, it is necessary to re-sort the user's questions and find and supplement the missing key information from relevant data sources (such as knowledge bases, databases, etc.); if the number of logical error points is high, the logical structure of the reply content must be re-sorted, the sentence order must be adjusted, and reasonable logical conjunctions must be added to make it logically coherent and clear. After completing the reply reconstruction, a dynamic feedback strategy is generated based on the reconstructed content. For example, the dynamic feedback strategy may stipulate that some of the reconstructed key content is sent to the user first, and the user's feedback is observed. Then, based on the feedback, it is decided whether to continue to supplement other content or further explain.

[0161] Furthermore, the cloud response content is transmitted back to the client according to the dynamic feedback strategy, that is, the cloud response content is updated and reconstructed according to the dynamic feedback strategy, and the updated cloud response content is transmitted back to the client, so that the client can accurately receive and parse the returned content, ensuring that the user can receive the response to his or her inquiry or request in a timely, accurate and satisfactory manner, improving the user experience, and ensuring the smoothness and efficiency of the entire 5G message interaction process.

[0162] The optimal cloud response content generation module 105 is used to extract the feedback semantics of the cloud response content transmitted back to the client, generate a feedback loop strategy based on the feedback semantics, and use the feedback loop strategy to generate the optimal cloud response content corresponding to the 5G message.

[0163] In the embodiment of the present invention, the feedback semantics refers to the semantic information such as the understanding, attitude, potential needs and further focus of the reply content reflected from the user's perspective after the client receives the cloud response content. It is not only the literal meaning of the reply content itself, but also includes the user's feelings and thoughts on the reply based on his own expectations, cognition and other factors.

[0164] In the embodiment of the present invention, when extracting the feedback semantics of the cloud response content transmitted back to the client, the optimal cloud response content generation module 105 is specifically used to:

[0165] Extracting keywords from the cloud response content transmitted back to the client;

[0166] Performing vector conversion on the keyword to obtain a keyword vector;

[0167] Determine the weight value of the keyword vector using a preset entropy method;

[0168] Constructing a semantic matrix of user feedback according to the keywords and the weight values;

[0169] The feedback semantics of the cloud response content is determined through the semantic matrix.

[0170] In detail, a toolkit with a keyword extraction function can be used to extract keywords from the cloud response content sent back to the client, wherein the words that appear most frequently in the user feedback are used as keywords. The toolkit includes but is not limited to a TextRank algorithm tool, a simplified Chinese text processing tool, and performs vector conversion on the keywords through a preset vector conversion model to obtain keyword vectors. The vector conversion model includes but is not limited to a word2vec model and a Bert model.

[0171] Specifically, the entropy method refers to a data method used to determine the degree of discreteness of a certain indicator. The greater the degree of discreteness, the greater the impact of the indicator on the comprehensive evaluation. That is, it is necessary to first calculate the feature vector of each keyword, and calculate the weight value of the keyword based on the feature vector, wherein the keyword and the weight value are in a one-to-one correspondence. According to the keywords and the weight values, the semantic matrix of the user feedback can be constructed. For example, the weight value corresponding to the keyword mobile phone is 0.1, the weight value corresponding to the keyword high-performance processor is 0.9, and the weight corresponding to the keyword large-capacity battery is 0.5. According to the semantic matrix, it can be known that the target therapy object is not satisfied with the reply response corresponding to the current mobile phone and large-capacity battery. Therefore, it is necessary to adjust this and determine the feedback semantics corresponding to the cloud response content to adjust the reply response corresponding to the mobile phone and large-capacity battery.

[0172] Furthermore, users convey their changing needs, expectations, and concerns through the feedback semantics of cloud response content, which can dynamically capture the changes in such needs and adjust subsequent reply content, recommendation strategies, etc., continuously optimize the interaction process with users, and better meet the different needs of users at different stages. It is necessary to adjust the cloud response content based on feedback semantics.

[0173] In an embodiment of the present invention, the feedback loop strategy is an overall strategy that integrates open-loop control conditions and closed-loop control conditions. It aims to comprehensively and flexibly respond to user feedback semantics of different feedback levels by combining open expansion optimization (open loop) with local fine adjustment (closed loop), thereby achieving continuous optimization of reply content and service quality in 5G message interaction.

[0174] In the embodiment of the present invention, when generating the feedback loop strategy according to the feedback semantics, the optimal cloud response content generating module 105 is specifically used to:

[0175] identifying a feedback form of the feedback semantics, and determining a feedback level of the feedback semantics according to the feedback form;

[0176] When the feedback level meets the preset feedback level rule of the cloud response content, an open-loop control condition of the feedback semantics is generated;

[0177] When the feedback level does not meet the preset feedback level rule of the cloud response content, generating a closed-loop control condition of the feedback semantics;

[0178] The open-loop control condition and the closed-loop control condition are integrated into a feedback loop strategy.

[0179] In detail, feedback form refers to the way or type of feedback information expressed by users through feedback semantics, including scoring feedback, opinion and suggestion feedback, and emotional expression feedback. The feedback level of feedback semantics is determined according to the feedback form. The feedback level is a quantitative expression used to measure the importance and urgency of feedback and the degree of influence on subsequent replies (usually expressed in numbers, grade classification, etc., such as high, medium, and low). For scoring feedback, if the score is greater than 90, the feedback level is low; if the score is less than 90 and greater than 60, the feedback level is medium; if the score is less than 60, the feedback level is high; for opinion and suggestion feedback, when the suggestions involve improvements in key aspects such as the accuracy and completeness of the reply content, such as pointing out that the reply omits important information, the feedback level can be set to high; while for suggestions on minor aspects such as the reply language style, the feedback level may be medium or low; in emotional expression feedback, when negative emotions appear and reflect a large misunderstanding or dissatisfaction with the reply content, the feedback level is usually high and needs to be handled in a timely manner to improve the user experience; the feedback level of positive emotions is relatively low.

[0180] Specifically, when the feedback level reaches the corresponding standards set by these rules (for example, setting the feedback level to be low triggers the open-loop control condition), an open-loop control condition is generated, that is, there is no need to adjust the cloud response content. When the feedback level does not reach the corresponding standards set by these rules, on the basis of the existing reply, targeted fine-tuning is performed according to the feedback semantics, so that the reply can better respond to user feedback while maintaining the original overall logic and framework, forming a relatively closed-loop improvement process. The circular cycle improvement mechanism can not only drastically improve the service strategy from a macro level when facing important feedback, but also make meticulous local improvements when dealing with relatively minor feedback, and continuously make reciprocating adjustments based on user feedback, so that the entire 5G message interaction service is always developing in the direction of more in line with user needs and higher quality.

[0181] Furthermore, the feedback loop strategy is used to generate the best cloud response content corresponding to the 5G message, that is, when the feedback loop strategy is an open-loop control condition, the cloud response content at this time is determined as the best cloud response content, and when the feedback loop strategy is a closed-loop control condition, the cloud response content needs to be adjusted according to the feedback strategy until the final feedback loop strategy is an open-loop control condition, and the adjusted cloud response content is determined as the best cloud response content. The feedback loop strategy is then used to generate the best cloud response content corresponding to the 5G message, which can give full play to the advantages of the strategy, comprehensively take into account user feedback at different levels, create high-quality reply content that meets user needs, and improve the overall effect and user experience of 5G message interaction.

[0182] The embodiment of the present invention uploads the message to the cloud according to the legitimacy of the message, avoiding the waste of cloud storage and computing resources by invalid and illegal messages; by using multi-dimensional semantic vectors to identify the message pool corresponding to the message, similar or related types of messages can be classified and managed; the response sub-content is merged into the cloud response content, and each sub-content is integrated and sorted out, and combined into a complete, standardized and organized reply according to certain rules and templates; by generating the return factor of the 5G message, the quality of the cloud response content can be quantitatively evaluated from multiple angles, and the cloud response content is returned to the client according to the dynamic return strategy, which fully considers the actual situation of the user receiving the reply; extracting the feedback semantics of the cloud response content returned to the client can deeply understand the user's real views on the reply, demand changes, and potential expectations; using the feedback loop strategy to generate the best cloud response content can meet user needs to the greatest extent and provide users with a high-quality 5G message interaction experience. Therefore, the cloud response system and method based on 5G messages proposed by the present invention can solve the problem of low accuracy of users when responding to 5G messages in the cloud.

[0183] Reference Figure 2 As shown, it is a flow chart of the operation method of the cloud response system based on 5G message provided by one embodiment of the present invention. In this embodiment, the operation method of the cloud response system based on 5G message includes:

[0184] S1. Identify the message format of the 5G message corresponding to the client received in advance, map the message format with the pre-acquired message transmission link, and verify the message legitimacy of the 5G message according to the mapped message transmission link;

[0185] S2. Upload the 5G message to the cloud according to the legitimacy of the message, and perform semantic analysis on the 5G message in the cloud according to a preset multi-dimensional algorithm to obtain a multi-dimensional semantic vector corresponding to the 5G message;

[0186] S3. Identify the message pool corresponding to the 5G message through the multi-dimensional semantic vector, generate the response sub-content corresponding to the 5G message according to the message pool using a preset message collaboration algorithm, and merge the response sub-content into the cloud response content corresponding to the 5G message;

[0187] S4, generating a return factor of the 5G message according to the cloud response content, generating a dynamic return strategy corresponding to the 5G message through the return factor, and returning the cloud response content to the client according to the dynamic return strategy;

[0188] S5. Extract the feedback semantics of the cloud response content transmitted back to the client, generate a feedback loop strategy based on the feedback semantics, and use the feedback loop strategy to generate the optimal cloud response content corresponding to the 5G message.

[0189] The embodiment of the present invention uploads the message to the cloud according to the legitimacy of the message, avoiding the waste of cloud storage and computing resources by invalid and illegal messages; by using multi-dimensional semantic vectors to identify the message pool corresponding to the message, similar or related types of messages can be classified and managed; the response sub-content is merged into the cloud response content, and each sub-content is integrated and sorted out, and combined into a complete, standardized and organized reply according to certain rules and templates; by generating the return factor of the 5G message, the quality of the cloud response content can be quantitatively evaluated from multiple angles, and the cloud response content is returned to the client according to the dynamic return strategy, which fully considers the actual situation of the user receiving the reply; extracting the feedback semantics of the cloud response content returned to the client can deeply understand the user's real views on the reply, demand changes, and potential expectations; using the feedback loop strategy to generate the best cloud response content can meet user needs to the greatest extent and provide users with a high-quality 5G message interaction experience. Therefore, the cloud response system and method based on 5G messages proposed by the present invention can solve the problem of low accuracy of users when responding to 5G messages in the cloud.

[0190] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0191] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0192] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0193] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0194] Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is therefore intended that all changes that fall within the meaning and range of equivalent elements of the claims are embraced in the present invention.

[0195] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or systems stated in a system claim can also be implemented by one unit or system through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

[0196] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A cloud response system based on 5G messages, characterized in that: The system includes a message legitimacy verification module, a semantic analysis module, a cloud response content fusion module, a dynamic return strategy generation module and an optimal cloud response content generation module, wherein: The message legitimacy verification module is used to identify the message format of the 5G message corresponding to the pre-received client, map the message format with the pre-acquired message transmission link, and verify the message legitimacy of the 5G message according to the mapped message transmission link, and is specifically used to: determine the message element compliance factor according to the message content corresponding to the 5G message; determine the message transmission compliance factor according to the mapped message transmission link; verify the element compliance value corresponding to the message element compliance factor, and verify the transmission compliance value corresponding to the message transmission compliance factor; calculate the message legitimacy of the 5G message according to the element compliance value and the transmission compliance value using the following preset time decay algorithm: Where L is the legitimacy of the message, λ is the message attenuation coefficient, t is the message consumption time, x i is the compliance value of the element corresponding to the i-th compliance factor in the message compliance factor, y i is the transmission compliance value corresponding to the i-th compliance factor in the message transmission compliance factor, n is the total number of compliance factors in the message element compliance factor, and m is the total number of compliance factors in the message transmission compliance factor; The semantic parsing module is used to upload the 5G message to the cloud according to the legitimacy of the message, and perform semantic parsing on the 5G message in the cloud according to a preset multi-dimensional algorithm to obtain a multi-dimensional semantic vector corresponding to the 5G message; The cloud response content fusion module is used to identify the message pool corresponding to the 5G message through the multi-dimensional semantic vector, generate the response sub-content corresponding to the 5G message according to the message pool using a preset message collaboration algorithm, and fuse the response sub-content into the cloud response content corresponding to the 5G message; The dynamic return strategy generation module is used to generate a return factor of the 5G message according to the cloud response content, generate a dynamic return strategy corresponding to the 5G message through the return factor, and return the cloud response content to the client according to the dynamic return strategy; The optimal cloud response content generation module is used to extract the feedback semantics of the cloud response content transmitted back to the client, generate a feedback loop strategy based on the feedback semantics, and use the feedback loop strategy to generate the optimal cloud response content corresponding to the 5G message.

2. The cloud response system based on 5G message as claimed in claim 1, characterized in that: The message legitimacy verification module is specifically used to: Parsing the message body content of the 5G message corresponding to the client received in advance; Determine metadata information of the 5G message according to the message body content; Parsing a message header field in the metadata information, and identifying a file type identifier in the message header field; Identify the message format of the 5G message according to the file type identifier.

3. The cloud response system based on 5G message as claimed in claim 1, characterized in that: The semantic parsing module performs semantic parsing on the 5G message in the cloud according to a preset multi-dimensional algorithm to obtain a multi-dimensional semantic vector corresponding to the 5G message, specifically for: Calculate the topic probability distribution of the 5G message using the topic model in the multidimensional algorithm; Determine the scenario type corresponding to the 5G message according to the topic probability distribution; Calculate the intention probability distribution of the 5G message using the intention model in the multi-dimensional algorithm; Determine the intent type corresponding to the 5G message according to the intent probability distribution; The business type corresponding to the 5G message in the cloud database is extracted through the scene type and the intent type, and the scene type, the intent type and the business type are integrated into a multi-dimensional semantic vector corresponding to the 5G message.

4. The cloud response system based on 5G message as claimed in claim 1, characterized in that: When the cloud response content fusion module generates the response sub-content corresponding to the 5G message according to the message pool using a preset message coordination algorithm, the cloud response content fusion module is specifically used to: Extract key information corresponding to the 5G message, determine a first business field of the 5G message according to the key information, and extract a second business field in the message pool; Calculating the semantic association between the first business field and the second business field; Constructing a business process fit matrix based on the first business area and the second business area; The message response coordination degree in the message pool corresponding to the 5G message is calculated according to the semantic relevance and the business process fit matrix using the following message coordination algorithm: Among them, S k is the message response coordination degree of the kth message party in the message pool, ω1 is the semantic association weight, ω2 is the business process fit weight, A is the business field vector corresponding to the 5G message, and B k is the business domain vector corresponding to the kth message party in the message pool, f uv is the semantic correlation between the business domain u corresponding to the 5G message and the business domain v corresponding to the message party, p uv is the degree of fit in the business process fit matrix between the business domain u corresponding to the 5G message and the business domain v corresponding to the message party; Determine the message responder in the message pool according to the message response coordination degree, and generate the response sub-content corresponding to the 5G message according to the message responder.

5. The cloud response system based on 5G message as claimed in claim 1, characterized in that: When the cloud response content fusion module fuses the response sub-content into the cloud response content corresponding to the 5G message, it is specifically used to: Identify the content format corresponding to the response sub-content; Mapping the content format with a predefined content template to obtain a mapping relationship; The response sub-content is embedded into the content template through the mapping relationship to obtain the cloud response content corresponding to the 5G message.

6. The cloud response system based on 5G message as claimed in claim 1, characterized in that: When the dynamic backhaul strategy generation module generates the backhaul factor of the 5G message according to the cloud response content, it is specifically used to: Detect the message reply completeness of the cloud response content to the 5G message, where the message reply completeness calculation formula is: Where Q is the message reply completeness, N is the number of sentences corresponding to the 5G message, and a h is the vector corresponding to the hth sentence in the 5G message, b d is the vector corresponding to the dth sentence in the cloud response content, and max is the maximum value function; When the message reply integrity is equal to a preset integrity threshold, determining the return factor of the 5G message as a single-round return factor; When the message reply completeness is not equal to a preset completeness threshold, detecting the message reply logical relevance of the cloud response content to the 5G message; When the message reply logic correlation meets the preset correlation range, the return factor of the 5G message is determined as a multi-round return factor; When the message reply logic correlation does not meet the preset correlation range, the return factor of the 5G message is determined as an abnormal return factor.

7. The cloud response system based on 5G message as claimed in claim 1, characterized in that: When the dynamic backhaul strategy generation module generates the dynamic backhaul strategy corresponding to the 5G message through the backhaul factor, the dynamic backhaul strategy generation module is specifically used to: Generating a return static strategy for the 5G message according to a single-round return factor in the return factor, and determining a return strategy for the 5G message through the return static strategy; Identify the multi-round indicator change trend of the 5G message according to the multi-round return factor in the return factor, adjust the reply direction of the 5G message according to the multi-round indicator change trend, and generate a dynamic return strategy for the 5G message through the reply direction and the multi-round return time deadline; According to the abnormal return factor in the return factor, the reply indicator corresponding to the 5G message is identified, the cloud response content is reply-reconstructed according to the reply indicator, and a dynamic return strategy for the 5G message is generated according to the reconstructed cloud response content.

8. The cloud response system based on 5G message as claimed in claim 1, characterized in that: When extracting the feedback semantics of the cloud response content transmitted back to the client, the optimal cloud response content generation module is specifically used to: Extracting keywords from the cloud response content transmitted back to the client; Performing vector conversion on the keyword to obtain a keyword vector; Determine the weight value of the keyword vector using a preset entropy method; Constructing a semantic matrix of user feedback according to the keywords and the weight values; The feedback semantics of the cloud response content is determined through the semantic matrix.

9. The cloud response system based on 5G message as claimed in claim 1, characterized in that: When the optimal cloud response content generation module generates the feedback loop strategy according to the feedback semantics, it is specifically used to: identifying a feedback form of the feedback semantics, and determining a feedback level of the feedback semantics according to the feedback form; When the feedback level meets the preset feedback level rule of the cloud response content, an open-loop control condition of the feedback semantics is generated; When the feedback level does not meet the preset feedback level rule of the cloud response content, generating a closed-loop control condition of the feedback semantics; The open-loop control condition and the closed-loop control condition are integrated into a feedback loop strategy.

10. A method for operating a cloud response system based on 5G messages, characterized in that: Used to implement a cloud response system based on 5G messages as described in any one of claims 1 to 9, the method comprising: Identify the message format of the 5G message corresponding to the pre-received client, map the message format with the pre-acquired message transmission link, and verify the message legitimacy of the 5G message according to the mapped message transmission link, specifically for: determining the message element compliance factor according to the message content corresponding to the 5G message; determining the message transmission compliance factor according to the mapped message transmission link; verifying the element compliance value corresponding to the message element compliance factor, and verifying the transmission compliance value corresponding to the message transmission compliance factor; using the following preset time decay algorithm to calculate the message legitimacy of the 5G message according to the element compliance value and the transmission compliance value: Among them, L is the legitimacy of the message, λ is the message attenuation coefficient, t is the message consumption time, x i is the compliance value of the element corresponding to the i-th compliance factor in the message compliance factor, y i is the transmission compliance value corresponding to the i-th compliance factor in the message transmission compliance factor, n is the total number of compliance factors in the message element compliance factor, and m is the total number of compliance factors in the message transmission compliance factor; Uploading the 5G message to the cloud according to the legitimacy of the message, and performing semantic analysis on the 5G message in the cloud according to a preset multi-dimensional algorithm to obtain a multi-dimensional semantic vector corresponding to the 5G message; Identify the message pool corresponding to the 5G message through the multi-dimensional semantic vector, generate the response sub-content corresponding to the 5G message according to the message pool using a preset message collaboration algorithm, and merge the response sub-content into the cloud response content corresponding to the 5G message; Generate a return factor of the 5G message according to the cloud response content, generate a dynamic return strategy corresponding to the 5G message through the return factor, and return the cloud response content to the client according to the dynamic return strategy; Extract the feedback semantics of the cloud response content transmitted back to the client, generate a feedback loop strategy based on the feedback semantics, and use the feedback loop strategy to generate the optimal cloud response content corresponding to the 5G message.

Images