Processing method for monitoring analysis system

Abstract

The invention provides a processing method for a monitoring analysis system. According to the method, comment data on a public network platform in a preset time period range are acquired to generate a comment data set, and then a user node set is generated according to user information of each piece of comment data in the comment data set; generating a feature vector edge set among the user nodes in the user node set according to each piece of comment data in the comment data set, thereby generating a comment data monitoring graph corresponding to the comment data according to the user node set and the feature vector edge set; according to the method, the target comment data is determined from the comment data set according to the comment data monitoring graph, so that the target comment data can be accurately determined from the comment data set based on the comment data monitoring graph, and key comment information published by users with transmissibility can be quickly focused.

Description

Technical Field

[0001] The present application relates to data processing technology, and in particular to a processing method for a monitoring and analysis system. Background Art

[0002] With the rapid development of the Internet, public network platforms such as social media, forums, and blogs have become important channels for people to express their opinions and share their views. The comment data on these platforms contains a large amount of user feedback and market dynamics information, which is of great analytical value to enterprises and individuals.

[0003] Traditional comment data analysis methods often rely on simple text mining techniques, such as keyword search and sentiment analysis. Although these methods can provide some basic information, they are obviously insufficient in capturing the complex relationships between comment data, understanding the propagation paths of comments, and identifying key comment nodes.

[0004] Therefore, faced with a large amount of comment data, how to efficiently extract useful information, especially those comments that are spreadable and influential, has become a problem that needs to be solved urgently. Summary of the invention

[0005] The present application provides a processing method for a monitoring and analysis system, which is based on a comment data monitoring graph and can accurately determine target comment data from a comment data set so as to quickly focus on key comment information posted by users with dissemination potential.

[0006] In a first aspect, the present application provides a processing method for a monitoring and analysis system, comprising: Acquire comment data on a public network platform within a preset time period to generate a comment data set, wherein the comment data includes comment time, comment information, and user information; Generate a user node set according to the user information of each comment data in the comment data set, wherein each user node in the user node set is used to represent the corresponding user information in the comment data set; Generating a feature vector edge set between each user node in the user node set according to each comment data in the comment data set, wherein each feature edge in the feature vector edge set is used to characterize a feature relationship between any two comment data in the comment data set; Generate a comment data monitoring graph corresponding to the comment data according to the user node set and the feature vector edge set; Target comment data is determined from the comment data set according to the comment data monitoring map.

[0007] In the above scheme, by obtaining the comment data on the public network platform within the preset time period, a large amount of user comment information can be systematically collected, covering a wide range of time spans and diverse data sources, ensuring the comprehensiveness and timeliness of the data. Among them, the comment data set not only contains the comment time and comment information, but also covers the user information, providing multi-dimensional data support for subsequent data analysis. Then, a user node set is generated based on the user information in the comment data set, and each user information is abstracted as a user node, realizing the structured representation of the comment data. Then, by generating a feature vector edge set between each user node in the user node set, the feature relationship between any two comment data in the comment data set can be accurately captured. Among them, the feature vector edge not only represents the association between the comments, but also contains information such as the direction and strength of the relationship. The capture of this feature relationship is of great significance for understanding the propagation path of the comment data and identifying the key comment nodes, and provides strong support for the subsequent determination of the target comment data. Then, a comment data monitoring map is generated based on the user node set and the feature vector edge set. The map presents the user nodes and the feature relationships between them in the form of a topological structure, so that the complex structure and internal laws of the data can be effectively expressed through the topological structure. Based on the comment data monitoring map, the target comment data can be accurately determined from the comment data set, so as to quickly focus on the key comment information posted by users with dissemination potential.

[0008] Optionally, generating a feature vector edge set between each user node in the user node set according to each comment data in the comment data set includes: Determining a comment time difference and user association information according to the first comment data and the second comment data in the comment data set, wherein the comment time difference is determined according to the first comment time in the first comment data and the second comment time in the second comment data, and the user association information is determined according to the first user information in the first comment data and the second user information in the second comment data; If the comment time difference satisfies a preset comment time difference condition and the user association information satisfies a preset user association relationship condition, an initial feature vector edge is generated between a first user node corresponding to the first comment data and a second user node corresponding to the second comment data, wherein the vector direction of the initial feature vector edge is from the user node corresponding to the comment data with an earlier comment time in the first comment data and the second comment data to another user node; If the comment time difference does not satisfy the preset comment time difference condition and / or the user association information does not satisfy the preset user association relationship condition, then no initial feature vector edge is generated between the first user node corresponding to the first comment data and the second user node corresponding to the second comment data; After the initial feature vector edge is generated between the first user node and the second user node, the feature weight value of the initial feature vector edge is determined according to the first comment information in the first comment data and the second comment information in the second comment data, so as to generate a corresponding feature vector edge according to the initial feature vector edge and the feature weight value.

[0009] In the above scheme, the comment time difference and user association information are first determined based on the first comment data and the second comment data in the comment data set. This step provides a reference for the subsequent feature vector edge generation by calculating the comment time difference and evaluating the association relationship between users. The calculation of the comment time difference helps to identify the time order between comments, so as to understand the propagation speed and trend of comments. At the same time, the determination of user association information can reveal the potential connections between commentators, such as friend relationships, social circles, etc., which is crucial for analyzing the influence and propagation path of comments. After determining the comment time difference and user association information, the preset comment time difference conditions and user association relationship conditions are used for screening. Only when both conditions are met at the same time, the initial feature vector edge is generated between the first user node and the second user node. This conditional screening mechanism ensures that the generated feature vector edge has a high degree of relevance and accuracy, avoids the generation of irrelevant or weakly related edges, and thus improves the quality of the graph and the analysis efficiency. The generated initial feature vector edge has a clear vector direction, from the user node with an earlier comment time to the user node with a later comment time. This directional representation not only reflects the propagation direction of the comment, but also helps to understand the flow trend of the comment between users. Through directional vector edges, the diffusion path of comments in the user network can be more intuitively displayed, providing strong support for subsequent public opinion monitoring and trend prediction. After the initial feature vector edge is generated, the feature weight value is further determined based on the comment information. The feature weight value reflects the similarity or correlation between comments and is a key indicator for evaluating the importance of feature vector edges. By assigning weight values ​​to feature vector edges, the correlation strength and influence between comments can be more accurately measured, which is of great significance for identifying key comments and mining potential topics. Finally, the corresponding feature vector edge is generated based on the initial feature vector edge and its feature weight value. This step combines multiple factors such as comment time difference, user association information, and comment information to ensure that the generated feature vector edge meets the preset conditions and has high relevance and accuracy. Through this comprehensive edge optimization mechanism, the generated comment data monitoring map can more accurately reflect the comment relationship and dissemination trend between users, providing strong support for subsequent data analysis and decision making.

[0010] Optionally, determining the feature weight value of the initial feature vector edge according to the first comment information in the first comment data and the second comment information in the second comment data includes: Respectively extracting keywords from the first comment information and the second comment information to form a corresponding first keyword set and a second keyword set; generating a keyword similarity matrix according to the first keyword set and the second keyword set; Determine a corresponding first keyword weight set according to the first keyword set, and determine a corresponding second keyword weight set according to the second keyword set; The feature weight value is determined according to the first keyword weight set, the second keyword weight set, and the keyword similarity matrix.

[0011] In the above scheme, keywords are first extracted from the first comment information and the second comment information respectively to form the corresponding first keyword set and the second keyword set. This step can effectively extract the core content of the comment, which provides a basis for subsequent analysis. The construction of the keyword set not only simplifies the complexity of the comment information, but also retains the core semantics of the comment, making the analysis more focused and accurate. Based on the first keyword set and the second keyword set, a keyword similarity matrix is ​​generated. The similarity matrix provides a basis for evaluating the relevance between comments by quantifying the similarity between keywords. Through similarity calculation, the commonalities and differences in the comments can be accurately identified, which provides a basis for the subsequent determination of feature weight values. In addition, the corresponding first keyword weight set and the second keyword weight set are further determined according to the first keyword set and the second keyword set. The allocation of weights takes into account factors such as the importance and frequency of occurrence of keywords in the comments, making the weight of keywords more reasonable and accurate. Through the reasonable allocation of weights, the influence of keywords in the comments can be more objectively evaluated, which provides a basis for the subsequent calculation of feature weight values. After determining the first keyword weight set, the second keyword weight set, and the keyword similarity matrix, the feature weight values ​​are calculated by combining these factors so that the feature weight values ​​can not only reflect the semantic similarity between comments, but also take into account the differences in the importance of keywords. By comprehensively considering multiple factors, more accurate and comprehensive feature weight values ​​can be generated, providing support for the subsequent construction of comment data monitoring maps. Through the above steps, the precise extraction of keywords in comment information, similarity calculation, weight allocation, and comprehensive consideration of feature weight values ​​are achieved. These steps are progressive and together improve the accuracy and depth of data analysis. Accurate feature weight values ​​help to more accurately represent the feature relationships between user nodes in the comment data monitoring map, providing strong data support for subsequent target comment data determination, public opinion monitoring, etc.

[0012] Optionally, determining a corresponding first keyword weight set according to the first keyword set, and determining a corresponding second keyword weight set according to the second keyword set includes: Determine the first keyword weight set according to the word frequency, inverse document frequency and the number of occurrences of each keyword in the first keyword set in the comment data set; The second keyword weight set is determined according to the word frequency, the inverse document frequency and the number of occurrences of each keyword in the second keyword set in the comment data set.

[0013] In the above scheme, when determining the keyword weight, the three dimensions of word frequency, inverse document frequency and the number of occurrences of the keyword in the comment data set are comprehensively considered. This multi-dimensional consideration makes the allocation of keyword weight more comprehensive and objective. Word frequency reflects the importance of the keyword in a single comment, inverse document frequency reflects the discrimination of the keyword in the entire comment data set, and the number of occurrences further strengthens the universality and influence of the keyword. Through the comprehensive evaluation of these three indicators, the actual weight of the keyword can be measured more accurately. Through the comprehensive calculation of word frequency, inverse document frequency and number of occurrences, the deviation that may be caused by a single indicator can be avoided, thereby improving the accuracy of keyword weight. For example, relying solely on word frequency may ignore some low-frequency but highly relevant keywords, while the introduction of inverse document frequency can make up for this deficiency and highlight the uniqueness of these keywords. At the same time, the consideration of the number of occurrences also enhances the stability of weight allocation, making the weight value more consistent with the actual distribution of keywords in the comment data set. Accurate keyword weight allocation helps to more finely characterize the semantic relationship between comments in subsequent data analysis. Among them, keyword weight not only affects the generation of feature vector edges and weight assignment, but also directly affects the construction quality of the comment data monitoring map. By assigning reasonable weights to keywords, the nodes and edges in the graph can more accurately reflect the actual structure and characteristics of the comment data, providing stronger support for subsequent target comment data determination, public opinion monitoring, etc.

[0014] Optionally, determining the comment time difference and user association information according to the first comment data and the second comment data in the comment data set includes: Determine the comment time difference according to the time difference between the first comment time and the second comment time; Determine a user association feature value according to the first user information and the second user information; Correspondingly, when the comment time difference is greater than the preset comment time difference threshold and the user association feature value is greater than the preset user association feature threshold, it is determined that the comment time difference satisfies the preset comment time difference condition and the user association information satisfies the preset user association relationship condition.

[0015] In the above scheme, the time difference between the first comment time and the second comment time is first calculated to determine the comment time difference, which helps to accurately identify the time interval between comments and provide basic data for the subsequent analysis of the propagation speed and trend of comments. In addition, through the accurate calculation of the time difference, the time correlation between comments can be more objectively evaluated, providing a basis for the subsequent generation of feature vector edges. Then, the user association feature value is determined according to the first user information and the second user information to quantify the degree of association between users. The calculation of the user association feature value may involve multiple dimensions, such as social relations, interaction frequency, common interests, etc. between users. By evaluating the user association feature value, the strength of association between users can be more accurately judged, providing support for the subsequent analysis of the diffusion path of comments in the user network. After determining the comment time difference and the user association feature value, the preset comment time difference threshold and user association feature threshold are used for screening. Only when the comment time difference is greater than the preset comment time difference threshold and the user association feature value is greater than the preset user association feature threshold, the condition is considered to be met. This condition screening mechanism ensures that the generated feature vector edge has a high degree of relevance and importance, avoids the generation of irrelevant or weakly related edges, and thus improves the quality and analysis efficiency of the comment data monitoring map. It can be seen that by accurately calculating the time difference between comments and evaluating the user association feature value, combined with a strict conditional screening mechanism, more accurate and meaningful feature vector edges can be generated. These feature vector edges not only reflect the time sequence and user association between comments, but also consider the importance of comment content and keywords, providing more accurate data support for subsequent public opinion monitoring.

[0016] Optionally, determining target comment data from the comment data set according to the comment data monitoring graph includes: Acquire multiple comment propagation paths from the comment data monitoring graph, wherein the comment propagation path includes a target user node sequence and a target feature vector edge for connecting two adjacent target user nodes in the target user node sequence, and the target feature vector edge points from a previous target user node in the target user node sequence to a next target user node; Determine the comment data corresponding to the target user node on the comment propagation path as the target comment data.

[0017] In the above scheme, multiple comment propagation paths are obtained from the comment data monitoring map. These paths intuitively characterize the diffusion process and direction of comments in the user network in a topological manner. Through the comment propagation path, it is possible to analyze how the comments spread from the initial user to other users, as well as the key nodes and paths in the propagation process, which provides a basis for subsequent analysis. After determining the comment propagation path, the comment data corresponding to the target user node on the path is further located as the target comment data. This positioning method ensures that the target comment data is closely related to the comment propagation path and has a significant impact, providing data support for subsequent analysis and decision-making. By directly obtaining the comment propagation path and target comment data from the comment data monitoring map, the cumbersome data screening and filtering steps in traditional data analysis are avoided, and the efficiency of data analysis is improved. At the same time, since the comment propagation path and target comment data are generated based on the topological comment data monitoring map, they have higher accuracy and reliability. The accurately located target comment data provides decision makers with clearer and more comprehensive information such as public opinion dynamics and user feedback. Based on this information, decision makers can respond and make decisions more quickly, such as adjusting marketing strategies, optimizing product features, etc., thereby improving the company's competitiveness and market response speed.

[0018] In a second aspect, the present application provides a monitoring and analysis system, comprising: An acquisition module, used to acquire comment data on a public network platform within a preset time period to generate a comment data set, wherein the comment data includes comment time, comment information, and user information; A processing module, configured to generate a user node set according to user information of each comment data in the comment data set, wherein each user node in the user node set is used to represent corresponding user information in the comment data set; The processing module is further used to generate a feature vector edge set between each user node in the user node set according to each comment data in the comment data set, wherein each feature edge in the feature vector edge set is used to characterize a feature relationship between any two comment data in the comment data set; The processing module is further used to generate a comment data monitoring graph corresponding to the comment data according to the user node set and the feature vector edge set; The processing module is further used to determine target comment data from the comment data set according to the comment data monitoring map.

[0019] Optionally, the processing module is specifically used to: Determining a comment time difference and user association information according to the first comment data and the second comment data in the comment data set, wherein the comment time difference is determined according to the first comment time in the first comment data and the second comment time in the second comment data, and the user association information is determined according to the first user information in the first comment data and the second user information in the second comment data; If the comment time difference satisfies a preset comment time difference condition and the user association information satisfies a preset user association relationship condition, an initial feature vector edge is generated between a first user node corresponding to the first comment data and a second user node corresponding to the second comment data, wherein the vector direction of the initial feature vector edge is from the user node corresponding to the comment data with an earlier comment time in the first comment data and the second comment data to another user node; If the comment time difference does not satisfy the preset comment time difference condition and / or the user association information does not satisfy the preset user association relationship condition, then no initial feature vector edge is generated between the first user node corresponding to the first comment data and the second user node corresponding to the second comment data; After the initial feature vector edge is generated between the first user node and the second user node, the feature weight value of the initial feature vector edge is determined according to the first comment information in the first comment data and the second comment information in the second comment data, so as to generate a corresponding feature vector edge according to the initial feature vector edge and the feature weight value.

[0020] Optionally, the processing module is specifically used to: Respectively extracting keywords from the first comment information and the second comment information to form a corresponding first keyword set and a second keyword set; generating a keyword similarity matrix according to the first keyword set and the second keyword set; Determine a corresponding first keyword weight set according to the first keyword set, and determine a corresponding second keyword weight set according to the second keyword set; The feature weight value is determined according to the first keyword weight set, the second keyword weight set, and the keyword similarity matrix.

[0021] Optionally, the processing module is specifically used to: Determine the first keyword weight set according to the word frequency, inverse document frequency and the number of occurrences of each keyword in the first keyword set in the comment data set; The second keyword weight set is determined according to the word frequency, the inverse document frequency and the number of occurrences of each keyword in the second keyword set in the comment data set.

[0022] Optionally, the processing module is specifically used to: Determine the comment time difference according to the time difference between the first comment time and the second comment time; Determine a user association feature value according to the first user information and the second user information; Correspondingly, when the comment time difference is greater than the preset comment time difference threshold and the user association feature value is greater than the preset user association feature threshold, it is determined that the comment time difference satisfies the preset comment time difference condition and the user association information satisfies the preset user association relationship condition.

[0023] Optionally, the processing module is specifically used to: Acquire multiple comment propagation paths from the comment data monitoring graph, wherein the comment propagation path includes a target user node sequence and a target feature vector edge for connecting two adjacent target user nodes in the target user node sequence, and the target feature vector edge points from a previous target user node in the target user node sequence to a next target user node; Determine the comment data corresponding to the target user node on the comment propagation path as the target comment data.

[0024] In a third aspect, the present application provides an electronic device, including: processor; and, A memory, configured to store executable instructions of the processor; The processor is configured to perform any possible method described in the first aspect by executing the executable instructions.

[0025] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement any possible method described in the first aspect.

[0026] The processing method of the monitoring and analysis system provided in the present application obtains comment data on a public network platform within a preset time period to generate a comment data set, then generates a user node set based on user information of each comment data in the comment data set, and then generates a feature vector edge set between each user node in the user node set based on each comment data in the comment data set, thereby generating a comment data monitoring map corresponding to the comment data based on the user node set and the feature vector edge set, so as to determine the target comment data from the comment data set based on the comment data monitoring map, thereby realizing the ability to accurately determine the target comment data from the comment data set based on the comment data monitoring map, so as to quickly focus on the key comment information posted by users with dissemination potential. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0028] Figure 1 This is a flowchart of a processing method of a monitoring and analysis system according to an exemplary embodiment of the present application; Figure 2 is a flowchart of a processing method of a monitoring and analysis system according to another exemplary embodiment of the present application; Figure 3 is a schematic diagram of the structure of a monitoring and analysis system according to an exemplary embodiment of the present application; Figure 4 It is a schematic diagram of the structure of an electronic device according to an exemplary embodiment of the present application.

[0029] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0030] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0031] In order to solve the above problems, the embodiments provided in this application mainly include the following technical concepts: Multi-dimensional data collection and integration: First, by obtaining comment data on public network platforms within a preset time period, a multi-dimensional comment data set containing comment time, comment information, and user information was constructed. This data collection method ensures the comprehensiveness and timeliness of the data, providing a solid foundation for subsequent data analysis.

[0032] Structured representation of comment data: In order to better understand and analyze comment data, the user information in the comment data set is abstracted into user nodes to generate a user node set. Each user node not only contains the user's unique identification information, but also reflects the potential connection between users through the connection relationship between nodes. This structured representation makes comment data more intuitive, easy to process and analyze.

[0033] Accurately capture feature relationships: The concept of feature vector edges is further proposed to characterize the feature relationship between any two comment data in the comment data set. Feature vector edges not only contain the association information between comments, but also reflect the direction and strength of the relationship through vector direction and weight value. This method of accurately capturing feature relationships provides strong support for understanding the propagation path of comment data and identifying key comment nodes.

[0034] Construction of comment data monitoring graph: Based on the user node set and feature vector edge set, a comment data monitoring graph is generated. The graph presents the user nodes and their feature relationships in a topological structure, so that the complex structure and inherent laws of the data can be effectively expressed through the graph. This visual expression method not only helps users to intuitively understand the comment data, but also facilitates subsequent data analysis.

[0035] Rapid determination of target comment data: By analyzing the comment data monitoring map, the target comment data can be accurately determined from the comment data set. The target comment data usually has high dissemination and influence, and can reflect the mainstream trend of public opinion. This method of quickly determining the target comment data provides strong support for public opinion monitoring, market analysis, etc., and helps enterprises and individuals to grasp market trends and user feedback in a timely manner.

[0036] Figure 1 FIG. 1 is a flow chart of a processing method of a monitoring and analysis system according to an exemplary embodiment of the present application. Figure 1 As shown, the processing method of the monitoring and analysis system provided in this embodiment includes: S101, obtaining comment data on a public network platform within a preset time period to generate a comment data set.

[0037] In this step, the comment data on the public network platform within a preset time period is obtained to generate a comment data set, wherein the comment data includes comment time, comment information and user information.

[0038] First, determine a preset time period, such as the past week or month. Then, collect comment data within this time period from public network platforms (such as Weibo, news websites, etc.). The collected comment data should include comment time, comment information (i.e., the content of the comment posted by the user), and user information (such as user name, user ID, etc.). These comment data are organized into a data set for subsequent processing and analysis. It is worth noting that the above-mentioned public network platform refers to a network platform that can legally collect information. If user authorization is required, the user's authorization must be obtained before obtaining the corresponding comment data.

[0039] S102: Generate a user node set according to the user information of each comment data in the comment data set.

[0040] In this step, a user node set is generated based on the user information of each comment data in the comment data set, wherein each user node in the user node set is used to represent the corresponding user information in the comment data set. The comment data set is traversed to extract the user information in each comment data. For each unique user information, a corresponding user node is created in the memory. These user nodes constitute a user node set, and each user node is used to represent the corresponding user information in the comment data set.

[0041] S103. Generate a feature vector edge set between each user node in the user node set according to each comment data in the comment data set.

[0042] In this step, a feature vector edge set between each user node in the user node set is generated according to each comment data in the comment data set, wherein each feature edge in the feature vector edge set is used to characterize the feature relationship between any two comment data in the comment data set.

[0043] Specifically, for any two comment data in the comment data set, the comment time difference and user association information between them are first calculated. The comment time difference is determined by comparing the comment time of the two comment data; the user association information can be determined by comparing the concerned object, concerned object or historical comment record in the user information of the two comment data. If the comment time difference meets the preset conditions (such as less than a certain time threshold or greater than a certain time threshold, which can be set accordingly according to the purpose of monitoring) and the user association information meets the preset relationship conditions (such as belonging to the same social circle or having frequent interactions), an initial feature vector edge is generated between the user nodes corresponding to the two comment data. The vector direction of the initial feature vector edge points from the user node with an earlier comment time to the user node with a later comment time. Next, the feature weight value of the initial feature vector edge is calculated based on the comment information in the two comment data. This can be determined by extracting the keywords of the comment information, calculating the keyword similarity matrix, and combining the keyword weight. Finally, the corresponding feature vector edge is generated based on the initial feature vector edge and its feature weight value, and added to the feature vector edge set.

[0044] S104. Generate a comment data monitoring graph corresponding to the comment data according to the user node set and the feature vector edge set.

[0045] Specifically, the user node set is used as the vertex set of the graph, and the feature vector edge set is used as the edge set of the graph to generate a review data monitoring graph. The graph presents the user nodes and the feature relationships between them in a topological structure, so that the complex structure and internal laws of the data can be effectively expressed through the topological structure.

[0046] S105. Determine target comment data from the comment data set according to the comment data monitoring graph.

[0047] In the generated comment data monitoring graph, by screening the comment data that meets specific conditions as the above-mentioned target comment data, the comment propagation path can also be searched using search algorithms (such as depth-first search, breadth-first search, etc.). The comment propagation path is a sequence of user nodes connected by feature vector edges, which represents the propagation path of comments between users. After obtaining multiple such comment propagation paths from the graph, the comment data corresponding to the target user node on the path is determined as the target comment data. These target comment data often have high dissemination and influence, and are the focus of public opinion monitoring and market analysis.

[0048] In this embodiment, comment data on a public network platform within a preset time period is obtained to generate a comment data set, and then a user node set is generated based on user information of each comment data in the comment data set, and then a feature vector edge set between each user node in the user node set is generated based on each comment data in the comment data set, so as to generate a comment data monitoring map corresponding to the comment data based on the user node set and the feature vector edge set, so as to determine target comment data from the comment data set based on the comment data monitoring map, thereby realizing the ability to accurately determine target comment data from the comment data set based on the comment data monitoring map, so as to quickly focus on key comment information posted by users with dissemination potential.

[0049] Figure 2 FIG. 1 is a flow chart of a processing method of a monitoring and analysis system according to another exemplary embodiment of the present application. Figure 2 As shown, the processing method of the monitoring and analysis system provided in this embodiment includes: S201, obtaining comment data on a public network platform within a preset time period to generate a comment data set.

[0050] In this step, the comment data on the public network platform within a preset time period is obtained to generate a comment data set, wherein the comment data includes comment time, comment information and user information.

[0051] First, determine a preset time period, such as the past week or month. Then, collect comment data within this time period from public network platforms (such as Weibo, news websites, etc.). The collected comment data should include comment time, comment information (i.e., the content of the comment posted by the user), and user information (such as user name, user ID, etc.). These comment data are organized into a data set for subsequent processing and analysis. It is worth noting that the above-mentioned public network platform refers to a network platform that can legally collect information. If user authorization is required, the user's authorization must be obtained before obtaining the corresponding comment data.

[0052] S202: Generate a user node set according to the user information of each comment data in the comment data set.

[0053] In this step, a user node set is generated based on the user information of each comment data in the comment data set, wherein each user node in the user node set is used to represent the corresponding user information in the comment data set. The comment data set is traversed to extract the user information in each comment data. For each unique user information, a corresponding user node is created in the memory. These user nodes constitute a user node set, and each user node is used to represent the corresponding user information in the comment data set.

[0054] S203: Determine the comment time difference and user association information according to the first comment data and the second comment data in the comment data set.

[0055] In this step, the comment time difference and user association information are determined based on the first comment data and the second comment data in the comment data set, wherein the comment time difference is determined based on the first comment time in the first comment data and the second comment time in the second comment data, and the user association information is determined based on the first user information in the first comment data and the second user information in the second comment data.

[0056] In a possible implementation, the comment time difference is determined according to the time difference between the first comment time and the second comment time.

[0057] Using formula 3, determine the user association feature value based on the first user information and the second user information , Formula 3 is:

[0058] in, is a set of target objects that the first user is interested in in the first user information, is a set of target objects that the second user is interested in in the second user information, is a set of target objects followed by the first user in the first user information, is a set of target objects followed by the second user in the second user information, is the number of elements in the collection.

[0059] In the above scheme, the time difference of comments is one of the basic data for time series analysis. By analyzing the time difference of comments, we can understand the distribution and changes of comment data in the time dimension, and provide important basis for subsequent public opinion trend prediction, user behavior pattern recognition, etc. In real-time monitoring and analysis scenarios, the time difference of comments can help the system quickly capture the real-time dynamics of comment data, such as the outbreak of hot events, the rapid changes in user emotions, etc., so as to make timely responses and decisions. In the recommendation system, considering the time difference of comments can help optimize the recommendation algorithm and improve the timeliness and relevance of the recommendation. For example, for target objects that have generated a large number of comments in the recent period, they can be recommended to users first to meet the real-time needs of users.

[0060] The above formula 3 is used to calculate the user association feature value, which comprehensively considers the target objects that users pay attention to and the target objects that are paid attention to, as well as the number of target objects that they pay attention to and are paid attention to independently. This multi-dimensional consideration method can more comprehensively evaluate the association between users and characterize the potential connections and interaction patterns between users. By calculating the user association feature value, the user profile can be further enriched and improved. For example, adding the association information between users to the user profile helps to more accurately understand the user's interest preferences, social circles and other characteristics, thereby providing strong support for applications such as personalized recommendations and community discovery. In community discovery and analysis, the user association feature value can be used as an important indicator to measure the closeness between users. By calculating the association feature values ​​between different users, user groups with close connections can be identified, and then the structural characteristics and evolution laws of the community can be analyzed, providing a basis for community management and marketing strategy formulation. In the information dissemination network, the user association feature value can help identify key dissemination nodes and paths. By preferentially pushing information to users with high association feature values, the dissemination efficiency and coverage of information can be improved, thereby optimizing the information dissemination path and strategy.

[0061] In another possible implementation, Formula 4 may also be used to determine the user association feature value according to the first user information and the second user information: , Formula 4 is:

[0062] in, is a set of target objects for which the first user has at least one-way public comments in the first user information. is a set of target objects for which the second user has at least one-way public comments in history in the second user information, is the number of elements in the collection; Correspondingly, when the comment time difference is greater than the preset comment time difference threshold and the user association feature value When it is greater than a preset user association feature threshold, it is determined that the comment time difference meets the preset comment time difference condition and the user association information meets the preset user association relationship condition.

[0063] In the above scheme, Formula 4 quantifies the correlation between users by calculating the ratio of the intersection and union of the target objects of two users' historical one-way public comments. This quantification method helps to objectively evaluate the similarity and interaction between users. By calculating the user correlation feature value, the user portrait can be further improved. Incorporating the correlation between users into the user portrait can more comprehensively reflect the user's interest preferences, social circles and other characteristics, and provide more accurate data support for applications such as personalized recommendations and community discovery. In the information dissemination network, the user correlation feature value can be used as an important indicator to measure the efficiency of information dissemination between users. By preferentially pushing information to users with high correlation feature values, the information dissemination path can be optimized, and the coverage and dissemination speed of information can be improved. In public opinion analysis, the user correlation feature value helps to identify key opinion leaders and community structures. By analyzing the correlation between users, influential user groups and potential public opinion trends can be discovered, providing strong support for public opinion warning and response.

[0064] When the comment time difference is greater than the preset comment time difference threshold and the user association feature value is greater than the preset user association feature threshold, it can be determined that the comment time difference meets the preset conditions and the user association information meets the preset relationship conditions. This comprehensive judgment standard helps to screen out comment data that is both timely and highly relevant, and provides more accurate and valuable data support for subsequent public opinion monitoring, user behavior analysis, etc. It is worth noting that the above-mentioned preset comment time difference condition can also be set to that the comment time difference is greater than the first preset comment time difference threshold, and less than the second preset comment time difference threshold.

[0065] S204 : Generate an initial feature vector edge between a first user node corresponding to the first comment data and a second user node corresponding to the second comment data.

[0066] If the comment time difference satisfies the preset comment time difference condition and the user association information satisfies the preset user association relationship condition, an initial feature vector edge is generated between the first user node corresponding to the first comment data and the second user node corresponding to the second comment data, wherein the vector direction of the initial feature vector edge is from the user node corresponding to the comment data with an earlier comment time between the first comment data and the second comment data to another user node.

[0067] If the comment time difference does not meet the preset comment time difference condition and / or the user association information does not meet the preset user association relationship condition, no initial feature vector edge is generated between the first user node corresponding to the first comment data and the second user node corresponding to the second comment data.

[0068] Specifically, for any two comment data in the comment data set (recorded as the first comment data and the second comment data), the system first calculates the comment time difference between them and obtains their respective user information. Then, the system makes a judgment based on the preset comment time difference condition (such as the comment time difference threshold) and the user association relationship condition (such as the user association feature value threshold). Only when the comment time difference meets the preset comment time difference condition and the user association information meets the preset user association relationship condition, the next step of the initial feature vector edge generation operation will be performed.

[0069] If both of the above conditions are met, the system will generate an initial feature vector edge between the first user node corresponding to the first comment data and the second user node corresponding to the second comment data. The vector direction of this initial feature vector edge is set from the user node corresponding to the comment data with an earlier comment time to the user node with a later comment time. This directional setting helps to reflect the propagation path and temporal relationship of comments between users.

[0070] If the comment time difference does not meet the preset comment time difference condition, and / or the user association information does not meet the preset user association relationship condition, the system will not generate an initial feature vector edge between the first user node corresponding to the first comment data and the second user node corresponding to the second comment data. Such processing logic helps to ensure that the generated feature vector edges have a high degree of relevance and accuracy, and avoid the generation of irrelevant or weakly related edges.

[0071] Through the above scheme, the system fully considers the time sequence of comments and the association between users when generating the initial feature vector edges, thereby ensuring that the generated feature vector edges not only conform to the actual data characteristics, but also accurately reflect the propagation path and temporal relationship of comments in the user network. Through conditional judgment and edge generation logic, the system can generate more accurate and meaningful feature vector edges, providing strong support for subsequent data analysis and public opinion monitoring. The directional setting of the initial feature vector edges makes the nodes and edges in the comment data monitoring map more intuitively display the propagation path and temporal relationship of comments in the user network, which helps users understand the data and analysis results more quickly. By avoiding the generation of irrelevant or weakly related edges, the system can reduce unnecessary computing and analysis burdens, optimize resource utilization, and improve processing efficiency.

[0072] S205 . Determine a feature weight value of an initial feature vector edge according to the first comment information in the first comment data and the second comment information in the second comment data, so as to generate a corresponding feature vector edge according to the initial feature vector edge and the feature weight value.

[0073] Specifically, after the initial feature vector edge is generated between the first user node and the second user node, the feature weight value of the initial feature vector edge is determined according to the first comment information in the first comment data and the second comment information in the second comment data, so as to generate the corresponding feature vector edge according to the initial feature vector edge and the feature weight value.

[0074] In a possible implementation, keywords of the first comment information and the second comment information are extracted respectively to form a corresponding first keyword set. And the second keyword set ,in, The first keyword set The number of keywords in The second keyword set The number of keywords in According to the first keyword collection Combined with the second keyword Generate keyword similarity matrix , where the keyword similarity matrix for:

[0075] Among them, the keyword similarity matrix In The first keyword set Middle Keyword and second keyword set Middle The similarity between keywords; According to the first keyword collection Determine the corresponding first keyword weight set , according to the second keyword set Determine the corresponding second keyword weight set ; Using formula 1, and based on the first keyword weight set , Second keyword weight set And the keyword similarity matrix Determine feature weights , where Formula 1 is:

[0076] in, is a preset constant.

[0077] In the above scheme, by extracting the keywords of the first comment information and the second comment information and generating a keyword similarity matrix, the content correlation between the two comments can be quantified. Each element in the keyword similarity matrix represents the similarity between the corresponding keywords in the two comments. The calculation of this similarity is based on the matching at the semantic level, which can more accurately reflect the substantive connection between the comment content. When calculating the feature weight value, not only the similarity between the keywords is considered, but also a keyword weight set is introduced to represent the importance of each keyword in the first comment information and the second comment information. This processing method enables the importance of different keywords in the comment to be considered when calculating the feature weight value, avoids the limitation of treating all keywords equally, and improves the accuracy and rationality of the weight value calculation. In addition, the above formula 1 effectively balances the influence of keyword similarity and keyword weight on the feature weight value. Through multiplication and exponential operation, formula 1 not only considers the similarity between keywords, but also weights the similarity through the weight value, so that the feature weight value can more comprehensively reflect the comprehensive relationship between the two comments. At the same time, the preset constant in the formula is used to avoid the calculation problem caused by the similarity being 0, ensuring the stability and reliability of the calculation. By quantifying the correlation between comments and considering the importance of keywords, the feature weight values ​​between comment data can be determined more accurately, thereby improving the accuracy of data analysis. In addition, since a structured approach is used to process comment data, the data analysis process is more efficient and can quickly process large-scale data sets to meet the needs of real-time monitoring and analysis.

[0078] In other words, the above formula 1 not only considers the similarity between keywords, but also introduces keyword weights, realizing the comprehensive quantification of the relationship between comment data. This comprehensive consideration method makes the calculated feature weight value more accurately reflect the actual correlation strength between comments, and improves the accuracy of data analysis. By introducing a preset constant, formula 1 effectively avoids the calculation problem caused when the keyword similarity is 0. This processing method ensures the stability and reliability of the calculation, and can obtain a reasonable feature weight value even in extreme cases. The exponential part in formula 1 realizes normalization processing, so that the final feature weight value is between 0 and 1 (or adjusted to other reasonable ranges according to specific circumstances), which is convenient for subsequent data processing and comparison. Since formula 1 adopts a structured calculation method, it can efficiently process large-scale data sets. When processing a large amount of comment data, the feature weight value between each comment pair can be quickly calculated, which provides strong support for subsequent monitoring and analysis work. By quantifying the feature weight value between comment data, formula 1 supports in-depth analysis of complex relationships. For example, in public opinion monitoring, key comment nodes and propagation paths can be identified based on feature weight values, providing a basis for public opinion guidance and control.

[0079] Further, for the above-mentioned first keyword set Determine the corresponding first keyword weight set , according to the second keyword set Determine the corresponding second keyword weight set , then it may specifically include: Using formula 2, and based on the first keyword set Determine the first keyword weight set , according to the second keyword set Determine the corresponding second keyword weight set , where Formula 2 is:

[0080] in, The first keyword set Middle Keywords The word frequency, The first keyword set Middle Keywords The inverse document frequency of The first keyword set Middle Keywords The number of occurrences in the review dataset, The second keyword set Middle Keywords The word frequency, The second keyword set Middle Keywords The inverse document frequency of The second keyword set Middle Keywords The number of occurrences in the review dataset.

[0081] In the above scheme, Formula 2 comprehensively considers three factors: word frequency, inverse document frequency, and the number of times a keyword appears in the comment data set. These factors jointly determine the weight of the keyword. Word frequency reflects the importance of a keyword in a single comment, inverse document frequency reflects the distinction of the keyword in the entire comment data set, and the number of occurrences further emphasizes the prevalence and importance of the keyword in the data set. This multi-dimensional consideration method makes the calculation of keyword weight more accurate and can better reflect the actual role of keywords in comment data.

[0082] Since keyword weight is an important basis for the subsequent calculation of feature weight values ​​(as shown in Formula 1), improving the accuracy of keyword weight will directly enhance the accuracy of data analysis. In the monitoring and analysis system, this means that key information and trends in comment data can be more accurately identified, providing stronger support for decision-making.

[0083] The various factors in Formula 2 can be adjusted or optimized according to actual needs to adapt to different analysis scenarios and data characteristics. For example, in some cases, more emphasis can be placed on the role of word frequency or inverse document frequency, while in other cases, more emphasis may be placed on the number of times a keyword appears in a data set. This flexibility enables the system to better adapt to different analysis needs and improve processing efficiency and effectiveness.

[0084] In addition, the calculation process of Formula 2 is relatively simple and efficient, and can support the processing of large-scale comment data sets. In the monitoring and analysis system, this means that a large amount of comment data can be quickly processed, key information can be extracted and analyzed, so as to meet the needs of real-time monitoring and rapid response. By calculating the keyword weight and combining it with the feature weight value calculation (as shown in Formula 1), the system can better discover the knowledge patterns and potential trends in the comment data. This is of great significance to fields such as public opinion monitoring and market analysis, and can help decision makers to promptly discover and respond to potential problems and opportunities.

[0085] S206. Obtain multiple comment propagation paths from the comment data monitoring graph.

[0086] In this step, multiple comment propagation paths are obtained from the comment data monitoring graph, wherein the comment propagation path includes a target user node sequence and a target feature vector edge for connecting two adjacent target user nodes in the target user node sequence, and the target feature vector edges all point from the previous target user node in the target user node sequence to the next target user node.

[0087] Specifically, the system builds a comment data monitoring graph based on the comment data set. The graph consists of user nodes and feature vector edges, where user nodes represent the publishers of comment data, and feature vector edges are generated based on the time difference between comment data and user association information, reflecting the propagation path and temporal relationship of comments between users.

[0088] In the comment data monitoring graph, the system extracts multiple comment propagation paths. These paths consist of a series of target user node sequences, each node in the sequence represents a user, and the feature vector edge used to connect two adjacent nodes indicates the direction of the comment propagation from the previous user to the next user.

[0089] S207: Determine the comment data corresponding to the target user node on the comment propagation path as the target comment data.

[0090] The system traverses each extracted comment propagation path and analyzes each target user node on the path one by one. For each target user node on the path, the system maps it back to the original comment data set and finds the comment data corresponding to the user node. The system determines the comment data corresponding to all target user nodes on the comment propagation path as the target comment data. These comment data have high attention and influence because they are widely disseminated in the user network.

[0091] It is worth noting that, based on the above embodiment, after obtaining multiple comment propagation paths from the comment data monitoring graph, the same target user nodes on different comment propagation paths can also be merged according to the multiple comment propagation paths to generate a comment propagation network. Then, characteristic user nodes are determined according to the comment propagation network, and the characteristic user nodes include original comment user nodes, key propagation user nodes, and active propagation user nodes, wherein the original comment user nodes include the starting user nodes of each comment propagation path in the comment propagation network, the key propagation user nodes include user nodes whose number of characteristic vector edges pointing to other user nodes in the comment propagation network is greater than a preset number threshold, and the active propagation user nodes include user nodes whose vector directions point to other user nodes in the comment propagation network and at the same time there are other user nodes pointing to the node.

[0092] The system first traverses all extracted comment propagation paths, identifies and merges the same target user nodes on different paths. This step is achieved by comparing the unique identifiers of user nodes (such as user IDs) to ensure that each user is represented only once in the network. After merging the same user nodes, the system constructs a comment propagation network based on the remaining nodes and feature vector edges. The network is a directed graph in which nodes represent users and edges represent the propagation paths and directions of comments between users. The system traverses each comment propagation path in the comment propagation network and marks the starting user node of the path as the original comment user node. These nodes are the starting point of comment propagation and are of great significance for understanding the origin of public opinion. The system counts the number of feature vector edges of each user node in the comment propagation network whose vector direction points to other user nodes. When the number of feature vector edges of a node exceeds the preset number threshold, the node is marked as a key user node for propagation. These nodes play an important transit role in the comment propagation process and have a significant impact on the spread of public opinion. The system further analyzes the user nodes in the comment propagation network to find those nodes that both point to other user nodes and are pointed to by other user nodes at the same time. These nodes are marked as active user nodes. They show high interactivity in the comment dissemination network and play an important role in the fermentation and spread of public opinion.

[0093] In the above scheme, after obtaining multiple comment propagation paths, a comment propagation network is generated by merging the same target user nodes on different paths to effectively integrate the scattered comment propagation information and construct a global comment propagation view. The construction of the comment propagation network enables the originally isolated comment propagation paths to be interconnected, forming a more complete and coherent information network, which helps to more comprehensively understand the propagation pattern and trend of comments in the user group. On the basis of the comment propagation network, characteristic user nodes are further determined, including original comment user nodes, key propagation user nodes and active propagation user nodes. The original comment user node is the starting point of comment propagation, and its identification helps to trace the source and initial impact of the comment. The key propagation user node and the active propagation user node represent the important transit stations and active participants in the comment propagation process, respectively. Their identification helps to deeply understand the diffusion mechanism and influence distribution of comments in the user network. In addition, different types of characteristic user nodes are distinguished by setting different standards (such as the number of characteristic vector edges with vector directions pointing to other user nodes, whether there are other user nodes pointing to the node, etc.). This multi-dimensional user classification method makes the analysis more refined and accurate, and can better capture the roles and contributions of different users in the comment propagation process. After identifying the characteristic user nodes, the monitoring and analysis system can pay more attention to these nodes and allocate resources in a more targeted manner. For example, for key user nodes and active user nodes, the system can strengthen the monitoring of their comment content and communication behavior in order to timely discover potential risks or opportunities, which helps to improve monitoring efficiency and reduce unnecessary waste of resources. Based on the analysis results of the comment propagation network and characteristic user nodes, enterprises can formulate marketing strategies, crisis public relations plans, etc. more scientifically. For example, by understanding the characteristics and influence of the original comment user nodes, enterprises can better locate the target audience and formulate precision marketing plans; by analyzing the behavior patterns of key user nodes and active user nodes, enterprises can more effectively guide public opinion and optimize crisis response strategies. The analysis of comment propagation networks and characteristic user nodes not only helps the current monitoring and analysis tasks, but also provides a basis for subsequent knowledge discovery and application. For example, by analyzing historical comment data, enterprises can discover information such as the changing trend of user interests and potential market demand, thereby promoting the innovation of products and services; at the same time, these analysis results can also provide valuable references and references for academic research, social public opinion analysis and other fields.

[0094] Figure 3 FIG. 1 is a schematic diagram of a monitoring and analysis system according to an exemplary embodiment of the present application. Figure 3 As shown, the monitoring and analysis system 300 provided in this embodiment includes: An acquisition module 310 is used to acquire comment data on a public network platform within a preset time period to generate a comment data set, wherein the comment data includes comment time, comment information, and user information; A processing module 320, configured to generate a user node set according to user information of each comment data in the comment data set, wherein each user node in the user node set is used to represent corresponding user information in the comment data set; The processing module 320 is further used to generate a feature vector edge set between each user node in the user node set according to each comment data in the comment data set, wherein each feature edge in the feature vector edge set is used to characterize a feature relationship between any two comment data in the comment data set; The processing module 320 is further used to generate a comment data monitoring graph corresponding to the comment data according to the user node set and the feature vector edge set; The processing module 320 is further configured to determine target comment data from the comment data set according to the comment data monitoring graph.

[0095] Optionally, the processing module 320 is specifically configured to: Determining a comment time difference and user association information according to the first comment data and the second comment data in the comment data set, wherein the comment time difference is determined according to the first comment time in the first comment data and the second comment time in the second comment data, and the user association information is determined according to the first user information in the first comment data and the second user information in the second comment data; If the comment time difference satisfies a preset comment time difference condition and the user association information satisfies a preset user association relationship condition, an initial feature vector edge is generated between a first user node corresponding to the first comment data and a second user node corresponding to the second comment data, wherein the vector direction of the initial feature vector edge is from the user node corresponding to the comment data with an earlier comment time in the first comment data and the second comment data to another user node; If the comment time difference does not satisfy the preset comment time difference condition and / or the user association information does not satisfy the preset user association relationship condition, then no initial feature vector edge is generated between the first user node corresponding to the first comment data and the second user node corresponding to the second comment data; After the initial feature vector edge is generated between the first user node and the second user node, the feature weight value of the initial feature vector edge is determined according to the first comment information in the first comment data and the second comment information in the second comment data, so as to generate a corresponding feature vector edge according to the initial feature vector edge and the feature weight value.

[0096] Optionally, the processing module 320 is specifically configured to: Respectively extracting keywords from the first comment information and the second comment information to form a corresponding first keyword set and a second keyword set; generating a keyword similarity matrix according to the first keyword set and the second keyword set; Determine a corresponding first keyword weight set according to the first keyword set, and determine a corresponding second keyword weight set according to the second keyword set; The feature weight value is determined according to the first keyword weight set, the second keyword weight set, and the keyword similarity matrix.

[0097] Optionally, the processing module 320 is specifically configured to: Determine the first keyword weight set according to the word frequency, inverse document frequency and the number of occurrences of each keyword in the first keyword set in the comment data set; The second keyword weight set is determined according to the word frequency, the inverse document frequency and the number of occurrences of each keyword in the second keyword set in the comment data set.

[0098] Optionally, the processing module 320 is specifically configured to: Determine the comment time difference according to the time difference between the first comment time and the second comment time; Determine a user association feature value according to the first user information and the second user information; Correspondingly, when the comment time difference is greater than the preset comment time difference threshold and the user association feature value is greater than the preset user association feature threshold, it is determined that the comment time difference satisfies the preset comment time difference condition and the user association information satisfies the preset user association relationship condition.

[0099] Optionally, the processing module 320 is specifically configured to: Acquire multiple comment propagation paths from the comment data monitoring graph, wherein the comment propagation path includes a target user node sequence and a target feature vector edge for connecting two adjacent target user nodes in the target user node sequence, and the target feature vector edge points from a previous target user node in the target user node sequence to a next target user node; Determine the comment data corresponding to the target user node on the comment propagation path as the target comment data.

[0100] Figure 4 is a schematic diagram of the structure of an electronic device according to an exemplary embodiment of the present application. Figure 4 As shown, an electronic device 400 provided in this embodiment includes: a processor 401 and a memory 402; wherein: The memory 402 is used to store computer programs, and the memory may also be a flash memory.

[0101] The processor 401 is used to execute the execution instructions stored in the memory to implement each step in the above method. For details, please refer to the relevant description in the above method embodiment.

[0102] Optionally, the memory 402 may be independent or integrated with the processor 401 .

[0103] When the memory 402 is a device independent of the processor 401, the electronic device 400 may further include: The bus 403 is used to connect the memory 402 and the processor 401 .

[0104] This embodiment further provides a readable storage medium, in which a computer program is stored. When at least one processor of an electronic device executes the computer program, the electronic device executes the methods provided in the above-mentioned various implementation modes.

[0105] This embodiment also provides a program product, which includes a computer program stored in a readable storage medium. At least one processor of the electronic device can read the computer program from the readable storage medium, and at least one processor executes the computer program so that the electronic device implements the methods provided in the above various embodiments.

[0106] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the claims.

[0107] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A processing method for a monitoring and analysis system, characterized in that: include: Acquire comment data on a public network platform within a preset time period to generate a comment data set, wherein the comment data includes comment time, comment information, and user information; Generate a user node set according to the user information of each comment data in the comment data set, wherein each user node in the user node set is used to represent the corresponding user information in the comment data set; Generating a feature vector edge set between each user node in the user node set according to each comment data in the comment data set, wherein each feature edge in the feature vector edge set is used to characterize a feature relationship between any two comment data in the comment data set; Generate a comment data monitoring graph corresponding to the comment data according to the user node set and the feature vector edge set; Target comment data is determined from the comment data set according to the comment data monitoring map.

2. The processing method for monitoring and analyzing system according to claim 1, characterized in that: The step of generating a feature vector edge set between each user node in the user node set according to each comment data in the comment data set includes: Determining a comment time difference and user association information according to the first comment data and the second comment data in the comment data set, wherein the comment time difference is determined according to the first comment time in the first comment data and the second comment time in the second comment data, and the user association information is determined according to the first user information in the first comment data and the second user information in the second comment data; If the comment time difference satisfies a preset comment time difference condition and the user association information satisfies a preset user association relationship condition, an initial feature vector edge is generated between a first user node corresponding to the first comment data and a second user node corresponding to the second comment data, wherein the vector direction of the initial feature vector edge is from the user node corresponding to the comment data with an earlier comment time in the first comment data and the second comment data to another user node; The feature weight value of the initial feature vector edge is determined according to the first comment information in the first comment data and the second comment information in the second comment data, so as to generate a corresponding feature vector edge according to the initial feature vector edge and the feature weight value.

3. The processing method for monitoring and analyzing system according to claim 2, characterized in that: If the comment time difference does not satisfy the preset comment time difference condition and / or the user association information does not satisfy the preset user association relationship condition, no initial feature vector edge is generated between the first user node corresponding to the first comment data and the second user node corresponding to the second comment data.

4. The processing method for monitoring and analyzing system according to claim 3, characterized in that: The determining the feature weight value of the initial feature vector edge according to the first comment information in the first comment data and the second comment information in the second comment data includes: Respectively extracting keywords from the first comment information and the second comment information to form a corresponding first keyword set and a second keyword set; generating a keyword similarity matrix according to the first keyword set and the second keyword set; Determine a corresponding first keyword weight set according to the first keyword set, and determine a corresponding second keyword weight set according to the second keyword set; The feature weight value is determined according to the first keyword weight set, the second keyword weight set, and the keyword similarity matrix.

5. The processing method for monitoring and analyzing system according to claim 4, characterized in that: The step of determining a corresponding first keyword weight set according to the first keyword set, and determining a corresponding second keyword weight set according to the second keyword set includes: Determine the first keyword weight set according to the word frequency, inverse document frequency and the number of occurrences of each keyword in the first keyword set in the comment data set; The second keyword weight set is determined according to the word frequency, the inverse document frequency and the number of occurrences of each keyword in the second keyword set in the comment data set.

6. The processing method for monitoring and analyzing system according to any one of claims 2 to 5, characterized in that: The determining of the comment time difference and user association information according to the first comment data and the second comment data in the comment data set includes: Determine the comment time difference according to the time difference between the first comment time and the second comment time; Determine a user association feature value according to the first user information and the second user information; Correspondingly, when the comment time difference is greater than the preset comment time difference threshold and the user association feature value is greater than the preset user association feature threshold, it is determined that the comment time difference satisfies the preset comment time difference condition and the user association information satisfies the preset user association relationship condition.

7. The processing method for monitoring and analyzing system according to any one of claims 2 to 5, characterized in that: The determining target comment data from the comment data set according to the comment data monitoring graph includes: Acquire multiple comment propagation paths from the comment data monitoring graph, wherein the comment propagation path includes a target user node sequence and a target feature vector edge for connecting two adjacent target user nodes in the target user node sequence, and the target feature vector edge points from a previous target user node in the target user node sequence to a next target user node; Determine the comment data corresponding to the target user node on the comment propagation path as the target comment data.

8. A monitoring and analysis system, characterized in that: include: An acquisition module, used to acquire comment data on a public network platform within a preset time period to generate a comment data set, wherein the comment data includes comment time, comment information, and user information; A processing module, configured to generate a user node set according to user information of each comment data in the comment data set, wherein each user node in the user node set is used to represent corresponding user information in the comment data set; The processing module is further used to generate a feature vector edge set between each user node in the user node set according to each comment data in the comment data set, wherein each feature edge in the feature vector edge set is used to characterize a feature relationship between any two comment data in the comment data set; The processing module is further used to generate a comment data monitoring graph corresponding to the comment data according to the user node set and the feature vector edge set; The processing module is further used to determine target comment data from the comment data set according to the comment data monitoring map.

9. An electronic device, characterized in that: include: processor; as well as, A memory, configured to store executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Images