Data transmission method and system for cloud computers
Through the advanced text parsing network and semantic difference analysis framework, key information in session data is parsed and encrypted, and the problems of low parsing efficiency and high encryption complexity of traditional methods are solved, achieving efficient and secure data transmission.
Patent Information
- Application Number
- CN202410577672.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-10
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-05-10
AI Technical Summary
Traditional text data processing methods cannot effectively parse and extract key information in session data, and the complexity of existing encryption technologies affects data transmission efficiency, making it difficult to ensure the security and privacy of data transmission.
Adopting an advanced text parsing network and semantic difference analysis framework, we obtain topic discriminant tags and semantic information by analyzing business session text, determine global semantic differences, and make intelligent encryption decisions to ensure the security of data transmission.
It improves the efficiency and accuracy of information processing, enhances the security and privacy of data transmission, and improves the level of information security.
Smart Images

Figure CN118427859B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a data transmission method and system for a cloud computer. Background Art
[0002] With the rapid development of cloud computing information technology, business conversation text data has become more frequent and important in enterprises and organizations. However, traditional text data processing methods often fail to effectively parse and extract key information from these conversation data, resulting in low efficiency and inaccuracy in information processing. At the same time, since these conversation data may contain sensitive information, how to ensure its security during transmission has also become an urgent problem to be solved.
[0003] In response to the above problems, although some technologies have attempted to protect the security and privacy of conversation data through text parsing and encryption, they often have many limitations. For example, traditional text parsing methods may not be able to accurately identify the topic and key information of the conversation text, and existing encryption technologies may be too complex, affecting the efficiency of data transmission. Summary of the invention
[0004] In order to improve the above problems, the present application provides a data transmission method and system for a cloud computer.
[0005] The present application provides a data transmission method for a cloud computer, which is applied to a cloud computing processing system. The method includes:
[0006] Acquire first transmission text data to be encrypted, wherein the first transmission text data includes at least two business session texts;
[0007] Parsing each business conversation text in the first transmission text data through a first text parsing network to obtain a topic discrimination label of each business conversation text and a conversation text semantics of each business conversation text, wherein the topic discrimination label is used to reflect the topic type to which the business conversation text belongs;
[0008] Based on the topic discrimination label of each business conversation text, obtaining user identity information and X conversation event information from the first transmission text data, wherein the user identity information and the conversation event information both belong to the business conversation text, and X is an integer not less than 1;
[0009] Determine X first semantic differences and X second semantic differences based on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information, wherein the X first semantic differences include a first semantic difference between each conversation event information and the user identity information, and the X second semantic differences include a second semantic difference corresponding to each conversation event information;
[0010] Determining a global semantic difference corresponding to each session event information based on the X first semantic differences and the X second semantic differences, wherein the global semantic difference is used to reflect the distinctiveness of the session event information;
[0011] Acquire the maximum global semantic difference from the global semantic differences corresponding to each session event information, and use the session event information corresponding to the maximum global semantic difference as the target session event information;
[0012] A transmission encryption decision is made on the target session event information through a transmission encryption network to obtain a transmission encryption decision tag of the target session event information, wherein the transmission encryption decision tag is used to determine the data to be encrypted in the target session event information.
[0013] Preferably, the obtaining of the first transmission text data to be encrypted comprises:
[0014] Acquire second transmission text data to be encrypted, wherein the second transmission text data includes at least two business conversation texts;
[0015] Parsing each business conversation text in the second transmission text data through a second text parsing network to obtain a sensitivity parsing result of each business conversation text, wherein the sensitivity parsing result is used to reflect the information sensitivity of the business conversation text;
[0016] Based on the sensitivity analysis result of each business conversation text, the business conversation text whose sensitivity analysis result in the second transmission text data belongs to the sensitive information category is used as the business conversation text in the first transmission text data.
[0017] Preferably, the method further comprises:
[0018] Obtaining a first transmission text data example set, wherein the first transmission text data example set includes at least one first transmission text data example, and each first transmission text data example corresponds to a sensitivity authentication result;
[0019] For each first transmission text data example in the first transmission text data example set, obtaining a sensitivity heat map corresponding to each first transmission text data example according to the second text parsing network;
[0020] Determining a first training error according to the sensitivity heat map corresponding to each first transmission text data example and the sensitivity authentication result;
[0021] The first training error is used to optimize the neuron bias weights of the second text parsing network.
[0022] Preferably, the parsing of each business conversation text in the first transmission text data by a first text parsing network to obtain a topic discrimination label of each business conversation text and a conversation text semantics of each business conversation text includes:
[0023] For each business conversation text in the first transmission text data, acquiring conversation text semantics of each business conversation text through a text feature embedding branch included in the first text parsing network;
[0024] For the conversation text semantics of each business conversation text, obtaining the downsampled semantics of each business conversation text through the semantic downsampling branch included in the first text parsing network;
[0025] For the downsampled semantics of each business conversation text, obtaining a topic heat map of each business conversation text through a discriminant branch included in the first text parsing network;
[0026] Based on the topic heat map of each business conversation text, a topic discrimination label of each business conversation text is determined.
[0027] Preferably, determining the topic identification label of each business conversation text based on the topic heat map of each business conversation text includes:
[0028] Comparing the thermal weights included in the subject thermal map of each business conversation text to obtain the maximum thermal weight in the subject thermal map of each business conversation text;
[0029] For each business conversation text, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the first thermal unit distribution, the subject discrimination label of the business conversation text is determined as the first subject category;
[0030] For each of the business conversation texts, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the second thermal unit distribution, the subject discrimination label of the business conversation text is determined as the second subject category, wherein the business conversation text belonging to the second subject category is the user identity information;
[0031] For each business conversation text, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the third thermal unit distribution, the topic discrimination label of the business conversation text is determined as the third topic category, wherein the business conversation text belonging to the third topic category is the conversation event information.
[0032] Preferably, determining the topic identification label of each business conversation text based on the topic heat map of each business conversation text includes:
[0033] Comparing the thermal weights included in the subject thermal map of each business conversation text to obtain the maximum thermal weight in the subject thermal map of each business conversation text;
[0034] For each business conversation text, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the first thermal unit distribution, the subject discrimination label of the business conversation text is determined as the first subject category;
[0035] For each of the business conversation texts, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the second thermal unit distribution, the subject discrimination label of the business conversation text is determined as the second subject category, wherein the business conversation text belonging to the second subject category is the user identity information;
[0036] For each of the business conversation texts, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the third thermal unit distribution, the subject discrimination label of the business conversation text is determined as the third subject category, wherein the business conversation text belonging to the third subject category is the conversation event information;
[0037] For each business conversation text, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the fourth thermal unit distribution, the topic discrimination label of the business conversation text is determined as the fourth topic category, wherein the business conversation text belonging to the fourth topic category is question and answer text information.
[0038] Preferably, the method further comprises:
[0039] Obtaining a second set of transmitted text data examples, wherein the second set of transmitted text data examples includes at least one second transmitted text data example, each second transmitted text data example corresponding to a topic discriminative training annotation;
[0040] For each second transmission text data example in the second transmission text data example set, obtaining a topic heat map corresponding to each second transmission text data example according to the first text parsing network;
[0041] Determining a second training error based on the topic heat map corresponding to each second transmission text data example and the topic discrimination training annotation;
[0042] The second training error is used to optimize the neuron bias weights of the first text parsing network.
[0043] Preferably, the determining of X first semantic differences and X second semantic differences based on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information comprises: performing a first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a first semantic difference between each conversation event information and the user identity information; performing a second semantic difference operation on the conversation text semantics of each conversation event information and the conversation text semantics corresponding to other conversation event information to obtain a second semantic difference corresponding to each conversation event information, wherein the other conversation event information is the remaining conversation event information among the X conversation event information except the conversation event information;
[0044] The determining, based on the X first semantic differences and the X second semantic differences, a global semantic difference corresponding to each session event information includes: for each session event information, integrating the first semantic difference and the second semantic difference corresponding to the session event information to obtain a global semantic difference corresponding to the session event information.
[0045] Preferably, performing a first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a first semantic difference between each conversation event information and the user identity information includes:
[0046] For each session event information, performing semantic transmission attribute difference analysis on each semantic transmission attribute in the session text semantics corresponding to the session event information and the semantic transmission attribute of the corresponding distribution area in the session text semantics corresponding to the user identity information, to obtain Y semantic transmission attribute differences between the session event information and the user identity information, where Y is an integer not less than 1;
[0047] For each session event information, session information analysis is performed on the Y semantic transmission attribute differences between the session event information and the user identity information to obtain a first semantic difference between the session event information and the user identity information.
[0048] Preferably, performing a second semantic difference operation on the conversation text semantics of each conversation event information and the conversation text semantics corresponding to other conversation event information to obtain the second semantic difference corresponding to each conversation event information includes:
[0049] For each of the session event information, a semantic transmission attribute difference analysis is performed on each semantic transmission attribute in the session text semantics corresponding to the session event information and the semantic transmission attributes of the corresponding distribution area in the session text semantics corresponding to each of the other session event information, to obtain Y semantic transmission attribute differences between each of the session event information and each of the other session event information, wherein Y is an integer not less than 1;
[0050] For each session event information, performing session information analysis on Y semantic transmission attribute differences between the session event information and other session event information to obtain a first semantic reverse similarity between the session event information and other session event information;
[0051] For each session event information, the first semantic reverse similarities between the session event information and each of the other session event information are averaged to obtain a second semantic difference corresponding to the session event information.
[0052] In some independent embodiments, performing a first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a first semantic difference between each conversation event information and the user identity information includes:
[0053] Performing a semantic reverse similarity operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a conversation text semantic reverse similarity between each conversation event information and the user identity information;
[0054] Performing a semantic reverse similarity operation on the encrypted statistical vector of the user identity information and the encrypted statistical vector of each session event information to obtain a reverse similarity of the encrypted statistical vector between each session event information and the user identity information;
[0055] For each session event information, the conversation text semantic reverse similarity between the session event information and the user identity information and the encrypted statistical vector reverse similarity between the session event information and the user identity information are integrated to obtain a first semantic difference between the session event information and the user identity information.
[0056] In some independent embodiments, performing a semantic reverse similarity operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain the conversation text semantic reverse similarity between each conversation event information and the user identity information includes:
[0057] For each session event information, performing semantic transmission attribute difference analysis on each semantic transmission attribute in the session text semantics corresponding to the session event information and the semantic transmission attribute of the corresponding distribution area in the session text semantics corresponding to the user identity information, to obtain Y semantic transmission attribute differences between the session event information and the user identity information, where Y is an integer not less than 1;
[0058] For each session event information, performing session information analysis on Y semantic transmission attribute differences between the session event information and the user identity information to obtain a reverse similarity of the session text semantics between the session event information and the user identity information;
[0059] The performing a semantic reverse similarity operation on the encrypted statistical vector of the user identity information and the encrypted statistical vector of each session event information to obtain the reverse similarity of the encrypted statistical vector between each session event information and the user identity information includes:
[0060] For each session event information, performing semantic transmission attribute difference analysis on each semantic transmission attribute in the encrypted statistical vector corresponding to the session event information and the semantic transmission attribute of the corresponding distribution area in the encrypted statistical vector corresponding to the user identity information, to obtain Z semantic transmission attribute differences between the session event information and the user identity information, where Z is an integer not less than 1;
[0061] For each session event information, an encrypted statistical vector difference operation is performed on Z semantic transmission attribute differences between the session event information and the user identity information to obtain an encrypted statistical vector inverse similarity between the session event information and the user identity information.
[0062] An embodiment of the present application provides a cloud computing processing system, comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor performs the above method.
[0063] An embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. The computer program implements the above method when running.
[0064] In the embodiment of the present application, by introducing an advanced text parsing network and semantic difference analysis framework, the method can more accurately parse and extract key information from session data, and ensure the security of data transmission through intelligent encryption decisions. This is of great significance for improving information security, protecting user privacy, and promoting efficient data transmission. BRIEF DESCRIPTION OF THE DRAWINGS
[0065] Figure 1 A flow chart of a data transmission method for a cloud computer provided in an embodiment of the present application.
[0066] Figure 2 A schematic diagram of the structure of a cloud computing processing system 200 provided in an embodiment of the present application. DETAILED DESCRIPTION
[0067] In order to better understand the above technical scheme, the technical scheme of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical scheme of the present application, rather than limitations on the technical scheme of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.
[0068] Figure 1 A data transmission method of a cloud computer is shown, which is applied to a cloud computing processing system. The method includes the following steps 110 to 170.
[0069] Step 110: Obtain first transmission text data to be encrypted, wherein the first transmission text data includes at least two business conversation texts.
[0070] Step 120: parse each business conversation text in the first transmitted text data through a first text parsing network to obtain a topic discrimination label of each business conversation text and a conversation text semantics of each business conversation text, wherein the topic discrimination label is used to reflect the topic type to which the business conversation text belongs.
[0071] Step 130: Based on the topic identification tag of each business conversation text, obtain user identity information and X conversation event information from the first transmission text data, wherein the user identity information and the conversation event information both belong to the business conversation text, and X is an integer not less than 1.
[0072] Step 140: Determine X first semantic differences and X second semantic differences based on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information, wherein the X first semantic differences include a first semantic difference between each conversation event information and the user identity information, and the X second semantic differences include a second semantic difference corresponding to each conversation event information.
[0073] Step 150: Determine a global semantic difference corresponding to each session event information based on the X first semantic differences and the X second semantic differences, wherein the global semantic difference is used to reflect the distinctiveness of the session event information.
[0074] Step 160: Obtain the maximum global semantic difference from the global semantic differences corresponding to each session event information, and use the session event information corresponding to the maximum global semantic difference as the target session event information.
[0075] Step 170: Perform a transmission encryption decision on the target session event information through a transmission encryption network to obtain a transmission encryption decision tag for the target session event information, wherein the transmission encryption decision tag is used to determine the data to be encrypted in the target session event information.
[0076] In an exemplary application scenario, a cloud computing processing system is responsible for processing and encrypting important session event information, and first obtains a set of first transmission text data to be encrypted. The first transmission text data to be encrypted includes multiple business session texts, which can be records of project discussions, customer exchanges, or internal meetings.
[0077] The cloud computing processing system then uses the first text parsing network to perform in-depth analysis on each business conversation text. Through natural language processing and machine learning algorithms, the cloud computing processing system can identify the topic of each text and label them with topic discrimination tags. At the same time, the cloud computing processing system also extracts the semantic content of each conversation text.
[0078] Based on the topic identification tags obtained above, the cloud computing processing system can identify the user identity information and multiple session event information in the first transmission text data. The user identity information and session event information are crucial for understanding the business context and the roles of the participants.
[0079] The cloud computing processing system further analyzes the differences between the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information. This difference analysis helps the system understand the relevance and uniqueness between different conversation events.
[0080] After analyzing all semantic differences, the cloud computing processing system calculates a global semantic difference (value) for each session event information. The global semantic difference reflects the uniqueness and importance of the session event in the entire business session.
[0081] The cloud computing processing system finds the maximum value from all calculated global semantic difference values, and the session event information corresponding to the maximum value is considered to be the target session event information, that is, the most unique and important part in the entire business session.
[0082] Finally, the cloud computing processing system makes an encryption decision on the target session event information through the transmission encryption network. This process involves complex encryption algorithms and protocols to ensure the security and integrity of data during transmission. The result of the encryption decision is a transmission encryption decision tag, which indicates which data needs to be encrypted to ensure security when transmitted in the cloud computer environment.
[0083] Based on the above application scenario, in step 110 and step 120, the first transmission text data to be encrypted: This refers to a set of text data that needs to be processed by the cloud computing processing system, which contains important text content that may involve sensitive information. These data need to be encrypted before transmission to ensure the security and privacy of the data. The first transmission text data can be a collection of multiple sessions, communications or files, which are generated during the business process and need special encryption protection because they contain sensitive or key information.
[0084] Business conversation text is text data generated in the business environment, usually including meeting records, conversation records, transaction records, etc. These data record the content and details of business exchanges and are an important basis for business analysis and decision-making. In the cloud computing environment, business conversation text can involve multiple participants and contain various business information and strategy discussions, so its security and privacy are crucial.
[0085] The first text parsing network is a specially designed network model for analyzing and understanding text data. It may include natural language processing (NLP) algorithms, deep learning models, and other advanced technologies to extract key information from text, such as topics, sentiment, entities, etc. In the cloud computing processing system, the first text parsing network is responsible for converting raw text data into a format that can be understood and operated by computers, providing a basis for subsequent data processing and analysis.
[0086] The topic discrimination label is a label assigned to each business conversation text after the first text parsing network analysis, which is used to identify the topic or category to which the text belongs. The topic discrimination label helps the cloud computing processing system to classify and organize a large amount of text data, thereby improving the efficiency and accuracy of data processing. For example, a conversation about product development may be labeled as the "product development" topic, while a conversation about marketing may be labeled as the "marketing" topic.
[0087] Conversation text semantics (features) refers to the intrinsic meaning or contextual meaning of business conversation text, which reflects the core information and intention expressed by the text. The cloud computing processing system extracts the semantic features of the conversation text through the first text parsing network to understand the text content more deeply. These semantic features may include keywords, phrases, sentence structures, and the overall sentiment and context of the text.
[0088] Topic categories refer to the types of different topics or issues involved in business conversation texts. In a cloud computing environment, business exchanges can cover multiple topics, such as project management, sales strategies, market trends, etc. Each topic category has its own unique characteristics and focus. The cloud computing processing system needs to be able to identify and distinguish these different topic categories in order to provide accurate data support for subsequent decision-making and analysis.
[0089] In detail, the cloud computing processing system first obtains the first transmission text data to be encrypted from various data sources. These data may come from the enterprise's internal database, cloud storage or other business systems. These data contain at least two business conversation texts, which are generated during business communication and may contain important business information, strategy discussions or sensitive data. Obtaining these data is the basis for subsequent encryption and processing. After obtaining the first transmission text data, the cloud computing processing system uses the first text parsing network to conduct in-depth analysis and processing of these data. This network model may contain complex natural language processing algorithms and deep learning technologies, which can accurately extract key information and features from the text. Through the analysis of the first text parsing network, the system can generate a topic discrimination label for each business conversation text, which accurately reflects the topic type to which the text belongs, such as "project management", "sales strategy", etc. At the same time, the system will also extract the conversation text semantics of each business conversation text. These semantic information is an in-depth understanding and summary of the text content, which provides an important basis for subsequent encryption decisions.
[0090] In steps 130-140, user identity information refers to information that can uniquely identify individuals participating in a session. This may include user names, employee numbers, positions, or other information that can confirm the identities of session participants. In cloud computing processing systems, it is critical to accurately identify and extract user identity information because this helps the system understand the participants in the session and their roles in the session, and thus better analyze the content and intent of the session. Session event information refers to information related to key events or activities that occur during a business session. These events may include decision making, the presentation of important ideas, the occurrence of transactions, etc. Session event information is an important part of the session text. They reflect the core content and purpose of the session and have an important impact on subsequent data analysis and business decisions.
[0091] In addition, the first semantic difference refers to the semantic difference between the session event information and the user identity information. In other words, it measures the degree of difference in semantic expression between a specific session event and a specific user identity. This difference may reflect the different understanding and expression of the same event by users with different identities in the session. The second semantic difference refers to the semantic change or uniqueness contained in each session event information itself. This difference may be reflected in the semantic comparison between different events in the same session, or the semantic change of the same event in different sessions. By analyzing and comparing the second semantic difference, the cloud computing processing system can have a deeper understanding of the intrinsic meaning and contextual relationship of the session event.
[0092] In detail, the system has generated a topic discrimination tag for each business conversation text in step 130 through the previous steps. Now, the system uses these tags to deeply explore the first transmission text data, with the purpose of obtaining user identity information and X conversation event information. The user identity information of the embodiment of the present application is not just a simple user name or ID, it also includes key attributes such as the user's role and position in the conversation. The conversation event information is an important event or decision point extracted from the conversation text, which may be closely related to the business process, project progress or market dynamics. In order to ensure the accuracy and completeness of the information, the system will carefully analyze the topic discrimination tags of each business conversation text, and identify and extract relevant user identity information and conversation event information based on these tags. This step is crucial for subsequent data encryption and decision analysis, because it helps the system clarify which information is sensitive and needs to be protected, and which information is the key basis for business decisions. In this step 140, the cloud computing processing system further processes and analyzes the user identity information and conversation event information extracted from step 130. The system not only pays attention to the content of this information, but also pays attention to their semantic meaning and contextual relationship. Through advanced natural language processing and deep learning techniques, the system is able to determine X first semantic differences and X second semantic differences. The first semantic difference reveals the association and difference between the session event information and the identity of a specific user. This difference may arise from the professional background, communication style, or business role of different users. By analyzing these differences, the system can more accurately understand the unique contribution and influence of each user in the session. The second semantic difference focuses on the uniqueness and variability of the session event information itself. This difference may reflect different expressions and interpretations of the same event at different times and in different contexts. By comparing and analyzing these differences, the system can capture subtle changes and important details in the session, thereby providing richer data support for subsequent encryption decisions and business analysis.
[0093] In step 150-step 160, global semantic difference refers to the overall difference in semantic content between a specific conversation event and all other conversation events in the entire conversation event set. This difference is obtained by comparing and analyzing the semantic features of each conversation event. Global semantic difference can reflect the uniqueness and distinctiveness of a conversation event in the entire conversation, and helps to identify and understand the inherent connections and differences between different conversation events. Target conversation event information refers to the conversation event selected as the focus of attention or analysis through a specific algorithm or standard (such as maximizing global semantic difference) among multiple conversation events. This information usually contains important, unique or representative content, and plays a key role in understanding the theme, intention or direction of the entire conversation.
[0094] In detail, the cloud computing processing system starts to process the first semantic difference (difference compared with the previous event) and the second semantic difference (difference with the standard or average session event) of each session event calculated previously. The system generates a global semantic difference value for each session event information by integrating these difference information. This global semantic difference is a comprehensive indicator that reflects the uniqueness of the session event in the entire session and the degree of difference from other events. The calculation of the global semantic difference may involve complex algorithms, including but not limited to weighted summation, normalization processing, etc., to ensure the comparability between different session events. After completing the calculation of the global semantic difference, the cloud computing processing system starts to find the maximum value among these differences. The session event information corresponding to the maximum value is considered to be the target session event information. This selection process is based on one, that is, the session event with the largest global semantic difference is the most representative and important in the entire session. Once the target session event information is determined, the system may further analyze, store or use it for other purposes, such as user behavior analysis, content recommendation, etc. In this way, the system can efficiently identify and process key information in a large amount of session data.
[0095] In step 170, making a transmission encryption decision means that the cloud computing processing system will first make a decision process before processing the data to be transmitted (such as the target session event information), to determine which data needs to be encrypted to protect its security and privacy. This process will comprehensively consider multiple factors such as the sensitivity of the data, the security of the transmission environment, and the needs of the recipient, so as to make a decision on whether to encrypt and how to encrypt. The transmission encryption decision label is a tag or identifier that is generated based on the result of the transmission encryption decision and attached to the data to be transmitted. This label indicates which data parts need to be encrypted, as well as the encryption method and key that may be used. It helps to ensure that the correct data parts are encrypted during the data transmission process, and also provides the recipient with key information required for decryption. The data to be encrypted refers to the data part that is identified as needing encryption protection during the transmission encryption decision process. Such data may contain sensitive information, such as personal privacy, commercial secrets or key business data, and therefore needs to be encrypted to ensure its security and confidentiality. Before being sent, the data to be encrypted will be processed by an encryption algorithm and converted into an unreadable ciphertext form to prevent unauthorized access and leakage.
[0096] In detail, the cloud computing processing system will conduct an in-depth analysis of the target session event information through a dedicated transmission encryption network. This analysis process is mainly to make a transmission encryption decision, that is, to determine which data is sensitive and needs to be encrypted. This decision will be made based on a series of complex algorithms and security protocols to ensure that only the data parts that really need to be protected will be encrypted. After the decision is made, the system will generate a transmission encryption decision label for the target session event information. This label not only indicates which data is to be encrypted, but may also contain key information such as encryption methods and keys. This label will be transmitted along with the data to ensure that the recipient of the data can accurately decrypt and read the original data. In addition, step 170 also ensures that only the sensitive part (i.e., the data to be encrypted) in the target session event information will be encrypted, thereby effectively protecting the privacy and security of the data, while also improving the efficiency and accuracy of data transmission. In this way, the cloud computing processing system can provide solid security for data transmission in a complex and changing network environment.
[0097] By using the embodiments of the present application, an innovative text data processing and encryption method is provided, which has the significant advantage of being able to intelligently identify and process complex business conversation text data. By introducing the first text parsing network, the method can accurately parse the subject of each business conversation text and obtain key user identity information and conversation event information accordingly. This not only improves the accuracy of data processing, but also greatly enhances the efficiency and intelligence level of information extraction.
[0098] In addition, the embodiment of the present application constructs a comprehensive semantic difference analysis framework by calculating the first semantic difference between each session event information and the user identity information, as well as the second semantic difference of each session event information itself. This framework enables the system to more deeply understand the nuances of the session content, thereby accurately evaluating the uniqueness and importance of each session event information.
[0099] More importantly, this method determines the global semantic difference of each session event information by integrating these semantic differences, and then identifies the most distinctive session event information as the target session event information. This precise identification mechanism is of vital importance for subsequent data processing, analysis and application.
[0100] Finally, the target session event information is encrypted through the transmission encryption network to ensure the transmission security of sensitive information. This step not only improves the confidentiality of data transmission, but also provides strong technical support for protecting user privacy and data security.
[0101] In summary, the embodiments of the present application have demonstrated excellent performance and wide applicability in text data processing and encryption, bringing new breakthroughs to the field of modern information security.
[0102] In an optional embodiment, the obtaining of the first transmission text data to be encrypted includes: obtaining the second transmission text data to be encrypted, wherein the second transmission text data includes at least two business conversation texts; parsing each business conversation text in the second transmission text data through a second text parsing network to obtain a sensitivity parsing result of each business conversation text, wherein the sensitivity parsing result is used to reflect the information sensitivity of the business conversation text; based on the sensitivity parsing result of each business conversation text, taking the business conversation text whose sensitivity parsing result in the second transmission text data belongs to the sensitive information category as the business conversation text in the first transmission text data.
[0103] In detail, the cloud computing processing system first obtains the second transmission text data to be encrypted. These data contain at least two business conversation texts, which can be chat records within the enterprise, customer service conversations, transaction negotiation records, etc. These text data are crucial for business operations and customer relationship maintenance, but they may also contain sensitive information, so encryption processing is required to protect data security.
[0104] Next, the cloud computing processing system performs a deep analysis of each business conversation text in the second transmission text data through a second text analysis network. This second text analysis network can be a deep learning model based on natural language processing (NLP), which can understand the semantics and context in the text and analyze the sensitivity of each business conversation text accordingly. The analysis result, i.e., the sensitivity analysis result, can directly reflect the information sensitivity of the business conversation text, such as whether it contains personal privacy, commercial secrets, or other content that needs to be kept confidential.
[0105] After obtaining the sensitivity analysis results of each business conversation text, the cloud computing processing system will further screen these data. Based on the sensitivity analysis results, it selects the business conversation texts that are determined to be sensitive information. These selected texts, due to their high sensitivity, become the focus of the subsequent encryption processing. They are integrated into the first transmission text data, which is a collection of conversation texts containing highly sensitive information.
[0106] Through the above steps, the cloud computing processing system can accurately identify sensitive information that needs to be encrypted, laying a solid foundation for subsequent data transmission encryption. This data screening method based on sensitivity analysis not only improves the pertinence and efficiency of data encryption, but also greatly enhances the security of data transmission.
[0107] This design, by introducing a second text parsing network to perform sensitivity analysis on business conversation text, achieves accurate identification and screening of sensitive information. This innovative data processing process not only improves the accuracy and efficiency of data encryption, but also provides strong technical support for enterprises and organizations to protect sensitive information during data transmission. Through this method, the cloud computing processing system can ensure that only truly sensitive data will be encrypted, thereby optimizing resource allocation and improving the cost-effectiveness of data encryption. At the same time, this intelligent data processing method also greatly reduces the risk of human errors and leaks, and enhances the overall level of information security.
[0108] Under some other design ideas, the method also includes: obtaining a first transmission text data example set, wherein the first transmission text data example set includes at least one first transmission text data example, and each first transmission text data example corresponds to a sensitivity authentication result; for each first transmission text data example in the first transmission text data example set, obtaining a sensitivity heat map corresponding to each first transmission text data example based on the second text parsing network; determining a first training error based on the sensitivity heat map corresponding to each first transmission text data example and the sensitivity authentication result; and using the first training error to optimize the neuron bias weight of the second text parsing network.
[0109] Under other design ideas of the embodiments of the present application, the method also covers an important training and optimization process, which aims to improve the recognition accuracy of the second text parsing network's sensitivity to text data.
[0110] First, the cloud computing processing system obtains a first transmission text data example set. This set contains at least one first transmission text data example, and each example corresponds to a known sensitivity authentication result. These examples can be data that has been manually annotated or previously processed by the system, and are used as reference standards for training and optimizing models.
[0111] Next, for each example in the first transmission text data example set, the cloud computing processing system will use the second text parsing network to process and generate a sensitivity heat map corresponding to each example. This heat map can intuitively show the distribution of the sensitivity of each part of the text, and is the visualization result of the second text parsing network's text sensitivity analysis.
[0112] With these sensitivity heat maps, the cloud computing processing system will compare them with the corresponding sensitivity certification results. By calculating the difference between the two, the system can determine the first training error. This error reflects the accuracy of the second text parsing network in sensitivity recognition. The smaller the error, the stronger the network's recognition ability.
[0113] Finally, the cloud computing processing system will use this first training error to optimize the neuron bias weights of the second text parsing network. This is a process of adjusting network parameters through the back-propagation algorithm, the purpose of which is to reduce the training error and improve the network's ability to predict and recognize future data. Through continuous iteration and optimization, the performance of the second text parsing network in identifying text sensitivity will be significantly improved.
[0114] In this way, the cloud computing processing system obtains a set of examples of the first transmitted text data and uses these examples to train and optimize the second text parsing network, thereby improving the network's accuracy in identifying text sensitivity. This not only enhances the level of intelligence in data processing, but also provides more reliable technical support for protecting sensitive information. At the same time, through continuous training and optimization, the performance of the second text parsing network will continue to improve, providing a solid foundation for the cloud computing processing system to process more complex and sensitive data in the future.
[0115] In some alternative embodiments, the parsing of each business conversation text in the first transmission text data by a first text parsing network to obtain a topic discrimination label of each business conversation text and a conversation text semantics of each business conversation text includes: for each business conversation text in the first transmission text data, obtaining the conversation text semantics of each business conversation text by a text feature embedding branch included in the first text parsing network; for the conversation text semantics of each business conversation text, obtaining the downsampled semantics of each business conversation text by a semantic downsampling branch included in the first text parsing network; for the downsampled semantics of each business conversation text, obtaining a topic heat map of each business conversation text by a discrimination branch included in the first text parsing network; and determining a topic discrimination label of each business conversation text based on the topic heat map of each business conversation text.
[0116] This embodiment involves further training and optimizing the second text parsing network. This process is accomplished by utilizing the first set of transmitted text data examples, aiming to improve the recognition accuracy of the network's sensitivity to text data.
[0117] First, the cloud computing processing system obtains a set of at least one first transmitted text data example, which are carefully selected and each example corresponds to a clear sensitivity certification result. This certification result is based on previous data analysis or expert judgment and represents the actual sensitivity level of the text data.
[0118] Then, for each example in the first set of transmitted text data examples, the cloud computing processing system uses the second text parsing network for processing. During the processing, the second text parsing network generates a sensitivity heat map, which can intuitively show the sensitivity distribution of different parts of the text. In this way, it is clear which parts contain sensitive information and which parts are relatively safe.
[0119] Next, the cloud computing processing system will compare the generated sensitivity heat map with the corresponding sensitivity certification result. By calculating the difference between the two, the system can determine a first training error. This error reflects the accuracy of the second text parsing network in identifying the sensitivity of the text; if the error is large, it means that the network's recognition ability needs to be improved.
[0120] In order to reduce this error, the cloud computing processing system will use optimization algorithms such as gradient descent to adjust the neuron bias weights of the second text parsing network. Through continuous iteration and optimization, the performance of the network will gradually improve, and the recognition of text sensitivity will become more accurate.
[0121] By introducing the first transmission text data example set to train and optimize the second text parsing network, the cloud computing processing system can significantly improve the recognition accuracy of the sensitivity of text data. The implementation of this method not only enhances the level of intelligence in the data processing process, but also provides a more reliable technical means for protecting sensitive information. At the same time, as the training continues, the performance of the second text parsing network will continue to improve, allowing the cloud computing processing system to show higher adaptability and accuracy when facing complex and changeable text data.
[0122] In addition, the sensitivity heat map is an innovative data visualization technology in the embodiment of this application, which is designed to reveal the distribution of sensitive information in text data. Through this technology, the sensitivity of different areas in the text can be intuitively displayed, helping users to quickly identify key information and ensure the security and privacy of data.
[0123] In the cloud computing processing system, the generation of the sensitivity heat map begins with the acquisition of the first set of transmitted text data examples. These data examples are equipped with corresponding sensitivity certification results, providing an accurate reference for subsequent sensitivity analysis. With the help of the powerful second text parsing network, the system deeply analyzes each text data example and accurately captures the sensitive information therein.
[0124] As the analysis progresses, the system begins to build a sensitivity heat map. In this map, colorful areas represent different degrees of sensitivity. Dark areas mean high sensitivity, while light areas are relatively safe. This intuitive visual presentation allows users to understand the distribution of sensitive information in the text at a glance.
[0125] The application value of the sensitivity heat map is not limited to this. By comparing with the known sensitivity certification results, the system can reveal the deviation of the second text parsing network in sensitivity recognition, that is, the first training error. This error becomes a key indicator for optimizing network performance. The cloud computing processing system will use this error to accurately adjust the network's neuron bias weights, thereby improving its ability to predict and recognize future data.
[0126] Therefore, the sensitivity heat map not only plays a role in data visualization in the cloud computing processing system, but also serves as a powerful assistant for network optimization. It helps users intuitively understand the sensitivity characteristics of the data, and also provides strong support for improving the recognition accuracy of the network. Through this comprehensive application, the cloud computing processing system demonstrates a higher level of intelligence and adaptability when processing complex text data.
[0127] In the next step, the topic discrimination label of each business conversation text is determined based on the topic heat map of each business conversation text, including: comparing the various thermal weights included in the topic heat map of each business conversation text to obtain the maximum thermal weight in the topic heat map of each business conversation text; for each business conversation text, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the first thermal unit distribution, then the topic discrimination label of the business conversation text is determined as the first topic category; for each business conversation text, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the second thermal unit distribution, then the topic discrimination label of the business conversation text is determined as the second topic category, wherein the business conversation text belonging to the second topic category is the user identity information; for each business conversation text, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the third thermal unit distribution, then the topic discrimination label of the business conversation text is determined as the third topic category, wherein the business conversation text belonging to the third topic category is the conversation event information.
[0128] In the next step, the cloud computing processing system will determine the topic discriminant label based on the topic heat map of each business conversation text. This process is carried out by analyzing the heat weight in the heat map.
[0129] First, the cloud computing processing system conducts an in-depth analysis of the topic heat map of each business conversation text. These topic heat maps are generated through the previous steps, and they intuitively show the heat distribution of different topic keywords in the business conversation text. Each topic heat map consists of multiple heat units, each of which represents a specific topic keyword and indicates the importance and frequency of the keyword in the text through color or value.
[0130] Next, the system will compare the heat weights included in the topic heat map of each business conversation text. This comparison process is to find the maximum heat weight in the heat map, that is, the most significant and most frequently appearing topic keyword. This maximum heat weight will serve as the key basis for determining the topic discrimination label.
[0131] After determining the maximum thermal weight, the cloud computing processing system will determine the topic type of the business conversation text according to the thermal unit distribution corresponding to this weight. Specifically, if the maximum thermal weight corresponds to the first thermal unit distribution, then the topic discrimination label of the business conversation text will be determined as the first topic type. Similarly, if the maximum thermal weight corresponds to the second thermal unit distribution or the third thermal unit distribution, then the topic discrimination label will be determined as the second topic type or the third topic type, respectively.
[0132] It is worth noting that the second and third subject categories of the embodiment of the present application have specific meanings. Business conversation texts belonging to the second subject category are identified as user identity information, which means that these texts contain keywords or information related to the user identity. Business conversation texts belonging to the third subject category are identified as conversation event information, that is, these texts mainly describe events or activities that occurred during the conversation.
[0133] Through the above steps, the cloud computing processing system can accurately assign a topic discrimination label to each business conversation text, thereby realizing the automatic identification and classification of the conversation text topic. This automated topic discrimination method not only improves the processing efficiency, but also provides strong support for subsequent data analysis and mining. In this way, the business conversation text topic discrimination method based on the topic heat map has significant beneficial effects. It can help the cloud computing processing system to more accurately understand and classify business conversation texts, thereby providing users with more accurate and personalized services. At the same time, this method can also effectively extract and utilize key information in the conversation text, providing strong support for the company's data analysis and decision-making.
[0134] In some further exemplary embodiments, the determining of the topic discrimination label of each business conversation text based on the topic heat map of each business conversation text comprises: comparing the thermal weights included in the topic heat map of each business conversation text to obtain the maximum thermal weight in the topic heat map of each business conversation text; for each business conversation text, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the first thermal unit distribution, determining the topic discrimination label of the business conversation text as the first topic category; for each business conversation text, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the second thermal unit distribution, determining the topic discrimination label of the business conversation text as the first topic category. The discrimination label is determined as the second topic category, wherein the business conversation text belonging to the second topic category is the user identity information; for each of the business conversation texts, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the third thermal unit distribution, the topic discrimination label of the business conversation text is determined as the third topic category, wherein the business conversation text belonging to the third topic category is the conversation event information; for each of the business conversation texts, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the fourth thermal unit distribution, the topic discrimination label of the business conversation text is determined as the fourth topic category, wherein the business conversation text belonging to the fourth topic category is question and answer text information.
[0135] In some further exemplary embodiments, the cloud computing processing system determines the topic discrimination labels of each business conversation text by deeply analyzing the topic heat map of the text. This process is crucial for data classification and subsequent information processing.
[0136] First, the cloud computing processing system will conduct a detailed analysis of the topic heat map of each business conversation text. These topic heat maps are previously generated by a specific algorithm and can intuitively show the heat distribution of different topic keywords in the text. The system focuses on the heat weights in the heat map, which represent the importance and frequency of different topic keywords in the text.
[0137] Next, the system will compare the heat weights of each topic heat map of each business conversation text. The purpose of this comparison is to find the maximum heat weight in each heat map, that is, the most prominent and significant topic keyword.
[0138] Once the maximum thermal weight is determined, the cloud computing processing system will assign a topic discrimination label to each business conversation text according to the thermal unit distribution corresponding to this weight. Specifically: if the maximum thermal weight corresponds to the first thermal unit distribution, then the topic discrimination label of the business conversation text will be determined as the first topic category; if the maximum thermal weight corresponds to the second thermal unit distribution, then the topic discrimination label of the text will be determined as the second topic category, and this type of text usually contains user identity information; if the maximum thermal weight corresponds to the third thermal unit distribution, then the topic discrimination label of the text will be determined as the third topic category, and this type of text mainly involves conversation event information.
[0139] In addition, if the maximum thermal weight corresponds to the newly introduced fourth thermal unit distribution, then the topic discrimination label of the text will be determined as the fourth topic category, and this type of text is identified as question-and-answer text information, that is, interactive content containing questions and answers.
[0140] In this way, the cloud computing processing system can accurately assign a suitable topic discrimination label to each business conversation text, thereby achieving fine classification of text topics. This classification method not only improves the accuracy of data processing, but also provides strong support for subsequent data mining and analysis. In this way, the cloud computing processing system can understand the text content more deeply and provide more accurate classification results, thereby providing users and enterprises with more efficient and targeted services. At the same time, through refined topic classification, the system can more effectively extract and utilize key information in the text, providing strong support for data analysis, user behavior research and other fields.
[0141] Under other possible design ideas, the method also includes: obtaining a second transmission text data example set, wherein the second transmission text data example set includes at least one second transmission text data example, and each second transmission text data example corresponds to a topic discrimination training annotation; for each second transmission text data example in the second transmission text data example set, obtaining a topic heat map corresponding to each second transmission text data example based on the first text parsing network; determining a second training error based on the topic heat map corresponding to each second transmission text data example and the topic discrimination training annotation; and using the second training error to optimize the neuron bias weight of the first text parsing network.
[0142] The cloud computing processing system introduces a more in-depth and continuous training and optimization process in its advanced topic discrimination function. This process is not simply carried out in steps, but a complete system that is interconnected and interdependent.
[0143] First, the system will comprehensively collect and organize a collection of second-transfer text data examples. These data are not randomly selected, but carefully considered to ensure their diversity and representativeness in order to train a more robust and general model. Each text data is accompanied by detailed topic discrimination training annotations, which are obtained by industry experts after rigorous analysis, providing a solid foundation for training.
[0144] Next, the cloud computing processing system uses the powerful first text parsing network to deeply analyze the text data. This network has been carefully designed and tuned to keenly capture every nuance in the text and generate accurate topic heat maps. These heat maps are not just graphics, but also an intuitive display of the deep meaning of the text, providing strong support for subsequent training and optimization.
[0145] The system then begins to compare the generated topic heat map with the accompanying training annotations. This process is not a simple comparison, but a deep analysis of the differences and similarities between the two through complex algorithms and models. These differences are quantified into specific values, namely the second training error, which accurately reflects the accuracy and reliability of the network in topic discrimination.
[0146] With this error value, the cloud computing processing system can start targeted optimization. This process is not achieved overnight, but requires repeated iterations and adjustments. The system will fine-tune the neuron bias weights of the first text parsing network according to the size and direction of the error. This fine-tuning is not done blindly, but is based on deep theoretical knowledge and practical experience to ensure that every adjustment can bring the network's performance to a higher level.
[0147] Through this continuous process, the cloud computing processing system not only significantly improves the accuracy of its topic identification, but also provides more solid and efficient support for various text analysis applications through continuous optimization and improvement. This in-depth and continuous training and optimization process is the core competitiveness of the cloud computing processing system in the field of topic identification.
[0148] Furthermore, the key steps and specific formulas involved in the above embodiments are as follows.
[0149] First, the system obtains a collection of multiple second transmission text data examples, and each example is equipped with corresponding topic discrimination training annotations. This is the basis for ensuring the accuracy of training. Then, using the first text parsing network, the system generates a topic heat map for each text example. This process is achieved through complex calculations, including steps such as identifying keywords and phrases in the text and quantifying the strength of the topic. The final generated topic heat map graphically displays the distribution and importance of topics in the text.
[0150] In the key step of determining the second training error, the system uses a specific error calculation formula. The topic vector predicted by the topic heat map is P, and the corresponding training annotation vector is T. The error E can be obtained by calculating the difference between the two. A commonly used method is to calculate the Euclidean distance or cosine similarity between the two. For example, the calculation formula of the Euclidean distance can be expressed as: (E=sqrt{sum_{i=1}^{n}(P_i-T_i)^2}), where n is the dimension of the topic vector. This formula can quantify the difference between prediction and reality, and provide direction for subsequent optimization.
[0151] Finally, the system uses the calculated second training error to optimize the neuron bias weights of the first text parsing network. The back propagation algorithm and gradient descent optimization method are used here. In each iteration, the weights of the network are updated according to the gradient information of the error. The specific weight update formula can be: (w_{new}=w_{old}-etacdotnablaE), where (eta) is the learning rate, which controls the step size of the weight update; (nablaE) is the gradient of the error E with respect to the weight w, indicating the direction of the weight adjustment. Through this optimization process, the performance of the network gradually improves, and the accuracy of topic discrimination is also enhanced.
[0152] In this way, through continuous iteration and optimization, the performance of the system can be continuously improved, providing more accurate and reliable support for various text analysis tasks.
[0153] In some other preferred embodiments, the determining of X first semantic differences and X second semantic differences based on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information includes: performing a first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a first semantic difference between each conversation event information and the user identity information; performing a second semantic difference operation on the conversation text semantics of each conversation event information and the conversation text semantics corresponding to other conversation event information to obtain a second semantic difference corresponding to each conversation event information, wherein the other conversation event information is the remaining conversation event information among the X conversation event information except the conversation event information. Then, the determining of the global semantic difference corresponding to each conversation event information based on the X first semantic differences and the X second semantic differences includes: integrating the first semantic difference and the second semantic difference corresponding to each conversation event information for each conversation event information to obtain a global semantic difference corresponding to the conversation event information.
[0154] In this embodiment, first, the cloud computing processing system obtains the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information. These semantic information are extracted from the text through deep learning and natural language processing technology, and they effectively express the meaning and context of the text.
[0155] Next, the system performs a series of operations on this semantic information. First, it performs a first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information. The purpose of this operation is to measure the semantic difference between each conversation event information and the user identity information. Through complex algorithms and models, the system accurately calculates this difference, which is called the first semantic difference.
[0156] In addition to the first semantic difference, the system also performs a second semantic difference operation. This time, it compares the difference between the conversation text semantics of each conversation event information and the conversation text semantics of other conversation event information. Here, "other conversation event information" refers to all conversation event information other than the currently processed conversation event information among the X conversation event information. The result of this operation reveals the semantic difference between each conversation event information and other conversation event information, which is called the second semantic difference.
[0157] After obtaining the first semantic difference and the second semantic difference of each session event information, the cloud computing processing system performs the final step of integration. For each session event information, the system comprehensively considers the first semantic difference and the second semantic difference corresponding to the session event information, and integrates these differences into a global semantic difference value through a specific algorithm and model. This global semantic difference value comprehensively reflects the semantic uniqueness and importance of the session event information in the entire session.
[0158] Specifically, the calculation of global semantic differences may involve complex weighting and normalization processing to ensure that the semantic differences between different conversation events can be compared on the same scale. The formula may be as follows: (GSD_i=alphacdotFSD_i+betacdotSSD_i), where (GSD_i) represents the global semantic difference of the i-th conversation event, (FSD_i) and (SSD_i) represent its first semantic difference and second semantic difference respectively, (alpha) and (beta) are weighting coefficients used to adjust the contribution of the two differences in the global semantic difference.
[0159] In this way, the cloud computing processing system can accurately determine the global semantic difference of each session event information. This not only helps to understand the inherent structure and meaning of the session data more deeply, but also provides strong support for subsequent tasks such as session analysis and user behavior prediction. In detail, the cloud computing processing system realizes the refined analysis of session data by comprehensively applying deep learning, natural language processing and complex algorithm models. This method not only improves the accuracy of session data processing, but also provides richer and deeper information for various applications based on session data. The beneficial effect is that through the determination of global semantic differences, the system can more comprehensively capture the key information in the session, providing strong data support for intelligent decision-making and precise services.
[0160] In some further possible embodiments, the performing a first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a first semantic difference between each conversation event information and the user identity information includes: for each conversation event information, performing a semantic transmission attribute difference analysis on each semantic transmission attribute in the conversation text semantics corresponding to the conversation event information and a semantic transmission attribute of a corresponding distribution area in the conversation text semantics corresponding to the user identity information to obtain Y semantic transmission attribute differences between the conversation event information and the user identity information, where Y is an integer not less than 1; and for each conversation event information, performing a conversation information analysis on the Y semantic transmission attribute differences between the conversation event information and the user identity information to obtain the first semantic difference between the conversation event information and the user identity information.
[0161] For example, the cloud computing processing system performs an in-depth first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information. This operation process aims to accurately quantify the semantic gap between each conversation event information and the user identity information, so as to more accurately understand the context of the conversation and the user's intention.
[0162] Specifically, the cloud computing processing system first compares each semantic transmission attribute in the corresponding conversation text semantics with the semantic transmission attributes of the same or similar distribution area in the conversation text semantics corresponding to the user identity information for each conversation event information. The "semantic transmission attributes" here can be understood as the specific meaning or information points carried in the text, which are the constituent elements of the conversation text semantics. The system uses complex algorithms to compare these attributes one by one and accurately calculates the difference between the two, which is called the semantic transmission attribute difference. Such difference analysis produces a total of Y semantic transmission attribute differences, where Y is an integer not less than 1, and the specific value depends on the complexity and subdivision of the conversation text semantics.
[0163] After completing the semantic transmission attribute difference analysis, the cloud computing processing system further performs session information analysis on these differences. The purpose of this step is to comprehensively evaluate the overall semantic difference between each session event information and user identity information based on these specific attribute differences. The system may use advanced natural language processing technology and machine learning algorithms to weight and aggregate these attribute differences, and finally obtain a quantitative first semantic difference value.
[0164] This first semantic difference value not only reflects the direct semantic difference between session event information and user identity information, but also reflects the deeper connection and difference between the two. It helps the system to more accurately understand the user's true intentions and needs, thereby providing users with more personalized and accurate services.
[0165] For example, in an intelligent customer service scenario, the user's identity information indicates that he is a consumer interested in technology products. In a conversation event, the user asks about the performance of a smartphone. Through the first semantic difference operation, the cloud computing processing system can accurately capture the semantic difference between the smartphone and the user's identity information, thereby recommending products that better meet the user's interests and needs.
[0166] It can be seen that the cloud computing processing system can accurately quantify the semantic difference between each session event information and user identity information through the refined first semantic difference operation. This not only improves the system's ability to understand session data, but also provides strong data support for subsequent user behavior prediction, personalized recommendation and other applications. The beneficial effect is that through accurate semantic difference analysis, the system can gain a deeper insight into user needs and provide users with a more intimate and intelligent service experience.
[0167] In an optional design idea, performing a second semantic difference operation on the conversation text semantics of each conversation event information and the conversation text semantics corresponding to other conversation event information to obtain the second semantic difference corresponding to each conversation event information includes: for each conversation event information, performing a semantic transmission attribute difference analysis on each semantic transmission attribute in the conversation text semantics corresponding to the conversation event information and the semantic transmission attribute of the corresponding distribution area in the conversation text semantics corresponding to the other conversation event information to obtain Y semantic transmission attribute differences between each conversation event information and the other conversation event information, wherein Y is an integer not less than 1; for each conversation event information, performing conversation information analysis on the Y semantic transmission attribute differences between the conversation event information and the other conversation event information to obtain a first semantic reverse similarity between the conversation event information and the other conversation event information; for each conversation event information, performing an average processing on the first semantic reverse similarity between the conversation event information and the other conversation event information to obtain the second semantic difference corresponding to the conversation event information.
[0168] In this embodiment, the cloud computing processing system performs a refined second semantic difference operation on the conversation text semantics of each conversation event information and the conversation text semantics of other conversation event information. The purpose of this process is to measure the semantic difference between each conversation event and other conversation events, thereby revealing the internal structure and information flow of the conversation.
[0169] First, the cloud computing processing system compares each semantic transmission attribute in the corresponding conversation text semantics of each conversation event information with the semantic transmission attributes of the same or similar distribution areas in the conversation text semantics of other conversation event information. The "semantic transmission attributes" here refer to the specific meaning or information units conveyed in the text, which are the basic elements that constitute the semantics of the conversation text. Through this comparison, the system can identify the subtle differences in the semantics of each conversation event information and other conversation event information.
[0170] After performing semantic transfer attribute difference analysis, the cloud computing processing system obtains Y semantic transfer attribute differences between each session event information and other session event information. Here, Y is an integer not less than 1, which represents the number of key semantic transfer attributes identified during the analysis process. These differences reflect the uniqueness and mutual relationship of different session events at the semantic level.
[0171] Next, the system further analyzes the conversation information of these semantic transmission attribute differences. By using advanced natural language processing and machine learning technologies, the system can deeply understand the meaning behind these differences and calculate the first semantic reverse similarity between each conversation event information and other conversation event information. This similarity index actually measures the semantic proximity of different conversation events, but unlike traditional similarity, it focuses more on capturing the reverse relationship between conversation events, that is, how one conversation event semantically echoes or diverges from other conversation events.
[0172] Finally, in order to obtain a more comprehensive and concise indicator, the cloud computing processing system averages the first semantic reverse similarity between each session event information and other session event information. This step integrates multiple similarity values into a single value, namely the second semantic difference. This difference value summarizes the overall semantic difference between each session event information and all other session event information, providing an important reference for subsequent session analysis and processing.
[0173] In this way, the cloud computing processing system can accurately calculate the second semantic difference corresponding to each conversation event information. This not only helps to reveal the internal logic and structure of the conversation, but also provides rich semantic information for various applications based on conversation data (such as intelligent recommendation, sentiment analysis, etc.). In this way, the accuracy and depth of conversation analysis are improved, allowing the system to better understand the user's true intentions and needs, thereby providing users with a more personalized and high-quality service experience.
[0174] In an optional embodiment, performing a first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a first semantic difference between each conversation event information and the user identity information includes: performing a semantic reverse similarity operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a conversation text semantic reverse similarity between each conversation event information and the user identity information; performing a semantic reverse similarity operation on the encrypted statistical vector of the user identity information and the encrypted statistical vector of each conversation event information to obtain an encrypted statistical vector reverse similarity between each conversation event information and the user identity information; and integrating, for each conversation event information, the conversation text semantic reverse similarity between the conversation event information and the user identity information and the encrypted statistical vector reverse similarity between the conversation event information and the user identity information to obtain a first semantic difference between the conversation event information and the user identity information.
[0175] In this embodiment, the cloud computing processing system performs an in-depth first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information. This operation not only involves the analysis of the conversation text semantics, but also introduces the concept of encrypted statistical vectors to enhance the accuracy and security of the semantic difference calculation.
[0176] First, the cloud computing processing system performs semantic reverse similarity calculations on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information. The "semantic reverse similarity" here is an innovative indicator. It is different from the traditional positive similarity and focuses on measuring the reverse correlation between the two. Through complex algorithms and models, the system calculates the reverse similarity between each conversation event information and the user identity information at the conversation text semantic level.
[0177] In addition to the reverse similarity of the conversation text semantics, the system further performs semantic reverse similarity operations on the encrypted statistical vectors of the user identity information and the encrypted statistical vectors of each conversation event information. The "encrypted statistical vector" here is a vector representation obtained through a specific encryption algorithm, which can preserve the statistical characteristics of the data while protecting the privacy of the original data. The system uses these encrypted statistical vectors to calculate the reverse similarity between each conversation event information and the user identity information at the statistical feature level.
[0178] After obtaining the conversation text semantic reverse similarity and the encrypted statistical vector reverse similarity, the cloud computing processing system integrates the two. Through specific algorithms and models, the system combines the conversation text semantic reverse similarity and the encrypted statistical vector reverse similarity to obtain the first semantic difference between each conversation event information and the user identity information.
[0179] Specifically, the integration process can involve weighted average of the two inverse similarities, decision fusion based on rules or machine learning, etc., to ensure the reasonable weight distribution of the two similarities in the integration process. The final first semantic difference is a comprehensive indicator that integrates the information of the conversation text semantics and the encrypted statistical vector, which more accurately reflects the semantic difference between each conversation event information and the user identity information.
[0180] In this way, the cloud computing processing system can more comprehensively evaluate the semantic differences between each session event information and user identity information. This not only improves the accuracy of semantic difference calculation, but also provides more accurate data support for subsequent user behavior analysis, personalized recommendation and other applications. The beneficial effect is that by introducing the concepts of encrypted statistical vectors and semantic reverse similarity, the system can more accurately capture the subtle differences between session event information and user identity information while protecting user privacy, laying a solid foundation for providing more accurate services.
[0181] In the next step, the semantic reverse similarity operation is performed on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain the conversation text semantic reverse similarity between each conversation event information and the user identity information, including: for each conversation event information, performing semantic transmission attribute difference analysis on each semantic transmission attribute in the conversation text semantics corresponding to the conversation event information and the semantic transmission attribute of the corresponding distribution area in the conversation text semantics corresponding to the user identity information to obtain Y semantic transmission attribute differences between the conversation event information and the user identity information, where Y is an integer not less than 1; for each conversation event information, performing conversation information analysis on the Y semantic transmission attribute differences between the conversation event information and the user identity information to obtain the conversation text semantic reverse similarity between the conversation event information and the user identity information.
[0182] Then, the encrypted statistical vector of the user identity information and the encrypted statistical vector of each session event information are subjected to a semantic reverse similarity operation to obtain the encrypted statistical vector reverse similarity between each session event information and the user identity information, including: for each session event information, performing a semantic transmission attribute difference analysis on each semantic transmission attribute in the encrypted statistical vector corresponding to the session event information and the semantic transmission attribute of the corresponding distribution area in the encrypted statistical vector corresponding to the user identity information to obtain Z semantic transmission attribute differences between the session event information and the user identity information, where Z is an integer not less than 1; for each session event information, performing an encrypted statistical vector difference operation on the Z semantic transmission attribute differences between the session event information and the user identity information to obtain the encrypted statistical vector reverse similarity between the session event information and the user identity information.
[0183] In detail, the cloud computing processing system will perform a semantic reverse similarity operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information. This step aims to deeply explore the potential semantic relationship in the conversation text, so as to more accurately evaluate the similarity between the conversation event information and the user identity information.
[0184] First, for each session event information, the cloud computing processing system compares each semantic transmission attribute in the corresponding session text semantics with the semantic transmission attributes of the same or similar distribution areas in the user identity information corresponding session text semantics. The "semantic transmission attributes" here can be understood as the specific meaning or information points conveyed in the text, which are the basic elements that constitute the semantics of the session text. Through complex algorithms and models, the system can accurately identify and quantify the differences between these attributes, which are called semantic transmission attribute differences.
[0185] After completing the analysis of Y semantic transmission attribute differences (where Y is an integer not less than 1, representing the number of key semantic transmission attributes identified), the cloud computing processing system further performs session information analysis on these differences. The purpose of this step is to comprehensively evaluate the reverse similarity between each session event information and user identity information at the level of session text semantics based on these specific attribute differences. The system may use advanced natural language processing technology and machine learning algorithms to deeply process these attribute differences and ultimately obtain a quantified session text semantic reverse similarity value.
[0186] At the same time, the cloud computing processing system also performs semantic reverse similarity operations on the encrypted statistical vectors of user identity information and the encrypted statistical vectors of each session event information. The encrypted statistical vector is a vector representation obtained through a specific encryption algorithm. It ensures the security of the data while retaining the statistical characteristics of the data.
[0187] In this step, the system also compares and analyzes each semantic transmission attribute in the encrypted statistical vector corresponding to each session event information with the semantic transmission attributes of the same or similar distribution area in the encrypted statistical vector corresponding to the user identity information. Through complex calculations, the system obtains Z semantic transmission attribute differences (Z is an integer not less than 1), which reflect the subtle differences between the session event information and the user identity information at the encrypted statistical vector level.
[0188] Next, the cloud computing processing system performs encrypted statistical vector difference operations on these semantic transmission attribute differences. Through specific algorithms and models, the system combines these differences to obtain a quantitative encrypted statistical vector reverse similarity value. This value not only reflects the similarity of the statistical characteristics of the session event information and the user identity information, but also takes into account the impact of encryption processing on the data.
[0189] In summary, through semantic reverse similarity calculation and encrypted statistical vector reverse similarity calculation, the cloud computing processing system can more comprehensively evaluate the similarity between each session event information and user identity information. This not only improves the accuracy of semantic analysis, but also provides more accurate data support for subsequent user behavior prediction, personalized recommendation and other applications.
[0190] In addition, in order to more intuitively demonstrate the calculation methods of these two reverse similarities, the following formula is introduced:
[0191] Conversation text semantic reverse similarity = 1 / (1+∑|diff_text(i)| / Y);
[0192] The reverse similarity of the encrypted statistical vector = 1 / (1+∑|diff_vector(j)| / Z).
[0193] Among them, diff_text(i) represents the difference of the i-th semantic transmission attribute in the conversation text semantics, and diff_vector(j) represents the difference of the j-th semantic transmission attribute in the encrypted statistical vector. Both formulas use the concept of inverse similarity, that is, the smaller the difference, the higher the similarity.
[0194] In this way, the cloud computing processing system can more accurately capture the deep connection between session event information and user identity information. The beneficial effect is that it improves the accuracy and depth of semantic analysis, allowing the system to understand user needs more deeply and provide users with a more accurate and personalized service experience.
[0195] Further, Figure 2 Schematic diagram of a cloud computing processing system 200 provided in an embodiment of the present application. Figure 2 The cloud computing processing system 200 shown includes a processor 210, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.
[0196] Alternatively, if Figure 2 As shown, the cloud computing processing system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method in the embodiment of the present application.
[0197] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .
[0198] Alternatively, if Figure 2 As shown, the cloud computing processing system 200 may further include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, to send information or data to other devices, or to receive information or data sent by other devices.
[0199] Optionally, the cloud computing processing system 200 may implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with a storage engine in each method of the embodiments of the present application, which will not be described in detail here for the sake of brevity.
[0200] It should be understood that the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiment can be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor may be a general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to be executed, or the hardware and software modules in the decoding processor are combined and executed. The software module may be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.
[0201] It can be understood that the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAM bus random access memory (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.
[0202] It should be understood that the above-mentioned memory is exemplary but not restrictive. For example, the memory in the embodiments of the present application may also be static random access memory (static RAM, SRAM), dynamic random access memory (dynamic RAM, DRAM), synchronous dynamic random access memory (synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (double data rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (synch link DRAM, SLDRAM) and direct memory bus random access memory (Direct Rambus RAM, DR RAM), etc. That is to say, the memory in the embodiments of the present application is intended to include but not limited to these and any other suitable types of memory.
[0203] Based on the above, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program implements the above method when running.
[0204] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations.
Claims
1. A data transmission method for a cloud computer, characterized in that: Applied to a cloud computing processing system, the method comprises: Acquire first transmission text data to be encrypted, wherein the first transmission text data includes at least two business session texts; Parsing each business conversation text in the first transmission text data through a first text parsing network to obtain a topic discrimination label of each business conversation text and a conversation text semantics of each business conversation text, wherein the topic discrimination label is used to reflect the topic type to which the business conversation text belongs; Based on the topic discrimination label of each business conversation text, obtaining user identity information and X conversation event information from the first transmission text data, wherein the user identity information and the conversation event information both belong to the business conversation text, and X is an integer not less than 1; Determine X first semantic differences and X second semantic differences based on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information, wherein the X first semantic differences include a first semantic difference between each conversation event information and the user identity information, and the X second semantic differences include a second semantic difference corresponding to each conversation event information; Determining a global semantic difference corresponding to each session event information based on the X first semantic differences and the X second semantic differences, wherein the global semantic difference is used to reflect the distinctiveness of the session event information; Acquire the maximum global semantic difference from the global semantic differences corresponding to each session event information, and use the session event information corresponding to the maximum global semantic difference as the target session event information; A transmission encryption decision is made on the target session event information through a transmission encryption network to obtain a transmission encryption decision tag of the target session event information, wherein the transmission encryption decision tag is used to determine the data to be encrypted in the target session event information.
2. The data transmission method of cloud computer according to claim 1, characterized in that: The step of obtaining the first transmission text data to be encrypted comprises: Acquire second transmission text data to be encrypted, wherein the second transmission text data includes at least two business conversation texts; Parsing each business conversation text in the second transmission text data through a second text parsing network to obtain a sensitivity parsing result of each business conversation text, wherein the sensitivity parsing result is used to reflect the information sensitivity of the business conversation text; Based on the sensitivity analysis result of each business conversation text, the business conversation text whose sensitivity analysis result in the second transmission text data belongs to the sensitive information category is used as the business conversation text in the first transmission text data.
3. The data transmission method of cloud computer according to claim 2, characterized in that: The method further comprises: Obtaining a first transmission text data example set, wherein the first transmission text data example set includes at least one first transmission text data example, and each first transmission text data example corresponds to a sensitivity authentication result; For each first transmission text data example in the first transmission text data example set, obtaining a sensitivity heat map corresponding to each first transmission text data example according to the second text parsing network; Determine a first training error according to the sensitivity heat map corresponding to each first transmission text data example and the sensitivity authentication result; The first training error is used to optimize the neuron bias weights of the second text parsing network.
4. The data transmission method of cloud computer according to claim 1, characterized in that: The step of parsing each business conversation text in the first transmission text data through a first text parsing network to obtain a topic discrimination label of each business conversation text and a conversation text semantics of each business conversation text includes: For each business conversation text in the first transmission text data, acquiring conversation text semantics of each business conversation text through a text feature embedding branch included in the first text parsing network; For the conversation text semantics of each business conversation text, obtaining the downsampled semantics of each business conversation text through the semantic downsampling branch included in the first text parsing network; For the downsampled semantics of each business conversation text, obtaining a topic heat map of each business conversation text through a discriminant branch included in the first text parsing network; Based on the topic heat map of each business conversation text, a topic discrimination label of each business conversation text is determined.
5. The data transmission method of cloud computer according to claim 4, characterized in that: The determining of the topic identification label of each business conversation text based on the topic heat map of each business conversation text includes: Comparing the thermal weights included in the topic thermal map of each business conversation text to obtain the maximum thermal weight in the topic thermal map of each business conversation text; For each business conversation text, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the first thermal unit distribution, the subject discrimination label of the business conversation text is determined as the first subject category; For each of the business conversation texts, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the second thermal unit distribution, the subject discrimination label of the business conversation text is determined as the second subject category, wherein the business conversation text belonging to the second subject category is the user identity information; For each business conversation text, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the third thermal unit distribution, the topic discrimination label of the business conversation text is determined as the third topic category, wherein the business conversation text belonging to the third topic category is the conversation event information.
6. The data transmission method of cloud computer according to claim 4, characterized in that: The determining of the topic identification label of each business conversation text based on the topic heat map of each business conversation text includes: Comparing the thermal weights included in the topic thermal map of each business conversation text to obtain the maximum thermal weight in the topic thermal map of each business conversation text; For each business conversation text, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the first thermal unit distribution, the subject discrimination label of the business conversation text is determined as the first subject category; For each of the business conversation texts, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the second thermal unit distribution, the subject discrimination label of the business conversation text is determined as the second subject category, wherein the business conversation text belonging to the second subject category is the user identity information; For each of the business conversation texts, if the maximum thermal weight in the subject heat map of the business conversation text corresponds to the third thermal unit distribution, the subject discrimination label of the business conversation text is determined as the third subject category, wherein the business conversation text belonging to the third subject category is the conversation event information; For each business conversation text, if the maximum thermal weight in the topic heat map of the business conversation text corresponds to the fourth thermal unit distribution, the topic discrimination label of the business conversation text is determined as the fourth topic category, wherein the business conversation text belonging to the fourth topic category is question and answer text information.
7. The data transmission method of a cloud computer according to any one of claims 1 to 6, characterized in that: The method further comprises: Obtaining a second set of transmitted text data examples, wherein the second set of transmitted text data examples includes at least one second transmitted text data example, each second transmitted text data example corresponding to a topic discriminative training annotation; For each second transmission text data example in the second transmission text data example set, obtaining a topic heat map corresponding to each second transmission text data example according to the first text parsing network; Determining a second training error based on the topic heat map corresponding to each second transmission text data example and the topic discrimination training annotation; The second training error is used to optimize the neuron bias weights of the first text parsing network.
8. The data transmission method of cloud computer according to claim 1, characterized in that: The determining of X first semantic differences and X second semantic differences based on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information includes: performing a first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a first semantic difference between each conversation event information and the user identity information; performing a second semantic difference operation on the conversation text semantics of each conversation event information and the conversation text semantics corresponding to other conversation event information to obtain a second semantic difference corresponding to each conversation event information, wherein the other conversation event information is the remaining conversation event information among the X conversation event information except the conversation event information; The determining, based on the X first semantic differences and the X second semantic differences, a global semantic difference corresponding to each session event information includes: for each session event information, integrating the first semantic difference and the second semantic difference corresponding to the session event information to obtain a global semantic difference corresponding to the session event information.
9. The data transmission method of cloud computer according to claim 8, characterized in that: The performing a first semantic difference operation on the conversation text semantics of the user identity information and the conversation text semantics of each conversation event information to obtain a first semantic difference between each conversation event information and the user identity information includes: For each session event information, performing semantic transmission attribute difference analysis on each semantic transmission attribute in the session text semantics corresponding to the session event information and the semantic transmission attribute of the corresponding distribution area in the session text semantics corresponding to the user identity information, to obtain Y semantic transmission attribute differences between the session event information and the user identity information, where Y is an integer not less than 1; For each session event information, performing session information analysis on Y semantic transmission attribute differences between the session event information and the user identity information to obtain a first semantic difference between the session event information and the user identity information; The performing a second semantic difference operation on the conversation text semantics of each conversation event information and the conversation text semantics corresponding to other conversation event information to obtain the second semantic difference corresponding to each conversation event information includes: For each of the session event information, a semantic transmission attribute difference analysis is performed on each semantic transmission attribute in the session text semantics corresponding to the session event information and the semantic transmission attributes of the corresponding distribution area in the session text semantics corresponding to each of the other session event information, to obtain Y semantic transmission attribute differences between each of the session event information and each of the other session event information, wherein Y is an integer not less than 1; For each session event information, performing session information analysis on Y semantic transmission attribute differences between the session event information and other session event information to obtain a first semantic reverse similarity between the session event information and other session event information; For each session event information, the first semantic reverse similarities between the session event information and each of the other session event information are averaged to obtain a second semantic difference corresponding to the session event information.
10. A cloud computing processing system, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 9.
