Real-time data processing method and device based on multi-dimensional intelligent analysis

Abstract

The invention provides a real-time data processing method and device based on multi-dimensional intelligent analysis, and relates to the technical field of data processing.The method comprises the steps that a multi-modal feature analysis model is constructed and deployed to each analysis thread; distributing independent analysis threads for different data types; acquiring a real-time data stream, and matching a corresponding data type for each piece of multi-dimensional data in the real-time data stream; the multi-dimensional data are sent to the corresponding analysis threads according to the corresponding relation between the data types and the analysis threads; calling the multi-modal feature analysis model in each analysis thread, performing feature analysis on the received multi-dimensional data, and outputting multi-modal features corresponding to the received multi-dimensional data; using a pre-deployed data missing detection model to identify whether each multi-modal feature has missing features; if yes, a corresponding data completion strategy is matched for missing feature completion; otherwise, returning to obtain the real-time data stream again.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a real-time data processing method and apparatus based on multi-dimensional intelligent analysis. Background Technology

[0002] When dealing with data processing tasks involving different data structures, especially multi-dimensional data tasks that include structured, semi-structured, and unstructured data, the processing difficulty will vary depending on the data structure.

[0003] Traditional real-time data processing methods for different data structures generally employ single, independent processing techniques. For example, for structured data: template-based wrappers leverage the stability of the data source structure to locate and extract target data using predefined rules or templates; content is parsed according to the data source format (e.g., HTML, XML); pattern matching uses regular expressions or templates to identify target data regions; and data extraction extracts the matched data and converts it into tables or other structured formats. For unstructured data: methods such as word embedding are used to convert text into numerical feature vectors.

[0004] However, existing methods cannot achieve unified classification and parsing of multidimensional heterogeneous data when processing batches of data packets, which can easily lead to thread lag and low efficiency in processing multidimensional heterogeneous data. Furthermore, there is a lack of follow-up and optimization strategies for the parsed data, making it impossible to achieve automated data completion and optimization, resulting in difficulty in obtaining complete and high-quality data later. Summary of the Invention

[0005] To address the shortcomings of existing methods in processing batches of data packets, which fail to achieve unified classification and parsing of multidimensional heterogeneous data, leading to thread lag and low efficiency in processing such data, and further hindering automated data completion and optimization, resulting in difficulties in obtaining complete and high-quality data later, this invention provides a real-time data processing method and apparatus based on multidimensional intelligent parsing.

[0006] The technical solutions provided by the embodiments of the present invention are as follows: First aspect: This invention provides a real-time data processing method based on multi-dimensional intelligent analysis, comprising: S1: Construct a multimodal feature parsing model and deploy it to each parsing thread; S2: Allocate independent parsing threads for different data types; S3: Acquire the real-time data stream and match the corresponding data type for each multi-dimensional data in the real-time data stream; S4: Send each multi-dimensional data in the real-time data stream to the corresponding parsing thread according to the correspondence between the data type and the parsing thread; S5: Call the multimodal feature parsing model in each parsing thread to perform feature parsing on the received multidimensional data and output the multimodal features corresponding to the received multidimensional data; S6: Using a pre-deployed data missing detection model, identify whether each of the multimodal features has missing features; if so, match the corresponding data completion strategy to complete the missing features; otherwise, return to step S3.

[0007] The second aspect: This invention provides a real-time data processing device based on multi-dimensional intelligent analysis, comprising: The building module is used to construct the multimodal feature parsing model and deploy it to each parsing thread; The allocation module is used to allocate independent parsing threads for different data types; The acquisition module is used to acquire real-time data streams and match corresponding data types for each multi-dimensional data in the real-time data streams. The sending module is used to send each multi-dimensional data in the real-time data stream to the corresponding parsing thread according to the correspondence between the data type and the parsing thread; The calling module is used to call the multimodal feature parsing model in each parsing thread to perform feature parsing on the received multidimensional data and output the multimodal features corresponding to the received multidimensional data; The identification module is used to identify whether each of the multimodal features has missing features using a pre-deployed data missing detection model; if so, it matches the corresponding data completion strategy to complete the missing features; otherwise, it returns to the acquisition module.

[0008] Third aspect: The present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the real-time data processing method based on multi-dimensional intelligent parsing as described in the first aspect.

[0009] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this embodiment of the invention, by constructing a multimodal feature parsing model and allocating independent parsing threads for different data types, the system can efficiently process multidimensional heterogeneous data, avoiding thread stalling and inefficiency. Real-time data streams are accurately assigned to corresponding threads after type matching, improving processing efficiency. Simultaneously, multimodal features are extracted from the data through feature parsing, and missing data is automatically filled in using a data missing detection model, ensuring the integrity and high quality of the parsed data. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A flowchart illustrating a real-time data processing method based on multi-dimensional intelligent analysis provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a real-time data processing device based on multi-dimensional intelligent analysis, provided in an embodiment of the present invention. Detailed Implementation

[0012] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0013] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0014] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0015] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0016] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0017] Reference manual attached Figure 1 The diagram illustrates a flowchart of a real-time data processing method based on multi-dimensional intelligent analysis provided by an embodiment of the present invention.

[0018] This invention provides a real-time data processing method based on multi-dimensional intelligent parsing. This method can be implemented by a real-time data processing device based on multi-dimensional intelligent parsing, which can be a terminal or a server. The processing flow of the real-time data processing method based on multi-dimensional intelligent parsing may include the following steps: S1: Construct a multimodal feature parsing model and deploy it to each parsing thread.

[0019] Multimodal feature parsing models are used to process and fuse features from different data sources, such as text, images, audio, video, structured and unstructured data. These different types of data typically have different representations and structures, so the goal of multimodal feature parsing models is to fuse them into a unified feature space for effective analysis and processing.

[0020] In one possible implementation, the multimodal feature parsing model is specifically as follows:

[0021] in, F This represents a multimodal feature parsing model, where Concat represents the feature concatenation operation. MIP This represents a multilayer perceptron. D s Represents a structured data matrix. GAT This represents a graph attention network. D h Represents a semi-structured data graph. Transformer- CNN Indicates combination Transformer and CNN Hybrid models for parsing unstructured data D u This represents unstructured data.

[0022] In this invention, feature concatenation (Concat) is used to fuse data from different modalities, integrating information from various data sources into a unified feature representation. This concatenation operation reduces the complexity of data transmission and processing, ensures efficient information flow, and helps improve the efficiency of subsequent processing tasks (such as classification and prediction). Furthermore, combining different data processing mechanisms (MIP, GAT, Transformer-CNN, etc.) with feature concatenation of multimodal data not only maximizes the preservation and utilization of potential information from different data types but also improves the model's accuracy, efficiency, and generalization ability.

[0023] S2: Allocate independent parsing threads for different data types.

[0024] Optionally, the data types include structured data, semi-structured data, and unstructured data.

[0025] In parallel computing and multithreaded processing environments, a parsing thread is an independent execution unit used to handle data parsing tasks. Each parsing thread is responsible for processing, parsing, or transforming specific types of data. By using parsing threads, efficient processing and real-time analysis of multimodal data can be achieved, reducing processing latency and improving the overall performance of the system.

[0026] In this invention, parallel data processing can be achieved by allocating independent parsing threads for each data type. Each thread focuses on processing a specific type of data, avoiding interference between different data types. This fully utilizes multi-core processors or distributed computing resources, improving the overall processing efficiency of the system. Simultaneously, each parsing thread processes different types of data, avoiding competition and resource contention between different data types, reducing the risk of thread stalls, and ensuring that the parsing task for each type of data can proceed independently and smoothly.

[0027] S3: Acquire real-time data streams and match the corresponding data types for each multi-dimensional data in the real-time data stream.

[0028] Multidimensional data refers to data from different data sources that contains multiple features, such as a single data item that may include text, images, and tabular data.

[0029] It's important to note that real-time data streams typically originate from various data sources, such as sensors, network requests, log files, and database streams. This data is usually dynamic, constantly updated, and requires immediate processing. Each data item can be multi-dimensional and of different types (e.g., structured, semi-structured, and unstructured data). Therefore, data classification algorithms or strategies are needed to determine the type of each data item.

[0030] In this invention, by classifying each data item according to its type and matching it to the corresponding parsing thread, it is ensured that data of different data types can be processed correctly. This avoids the low processing efficiency caused by mixed data types, thereby improving the overall processing speed.

[0031] S4: Send each multi-dimensional data in the real-time data stream to the corresponding parsing thread according to the correspondence between the data type and the parsing thread.

[0032] In one possible implementation, S4 specifically includes: S401: Pre-configure the corresponding data type classification function, thread pool mapping table, and data distribution routing function for each parsing thread:

[0033] in, t i express D s , D h , D u The Middle i The type label for each data item, TypeClassifier, indicates the classifier configured for the data format. d i express D s , D h , D u The Middle i One data item.

[0034] The Thread Pool Mapping Table is a structured data model used to manage and allocate thread pools, particularly in multithreaded and parallel computing systems. It helps the system dynamically allocate and manage thread pools based on different tasks or data types. Its core purpose is to map different types of tasks (or data) to appropriate thread pools, thereby improving resource utilization efficiency, avoiding thread contention, and ensuring efficient system operation.

[0035]

[0036] in, T This represents the thread pool mapping table. T 1 represents D s The parsing thread pool, T 2 indicates D h The parsing thread pool,T 3 indicates D u The parsing thread pool.

[0037] In this invention, by pre-configuring a thread pool mapping table, the system can dynamically allocate workloads to appropriate thread pools based on tasks or data types. This method ensures efficient utilization of thread pools and avoids resource waste.

[0038] The data distribution routing function is a functional module used to efficiently manage and distribute data in a distributed system. Its core task is to correctly route data items in a real-time data stream to the appropriate processing unit (such as a parsing thread, compute node, or service) according to predefined rules or logic. By using the data distribution routing function, the system can ensure that different types of data are correctly distributed to the corresponding processing paths, greatly improving the system's efficiency, stability, and scalability.

[0039]

[0040] Here, Route() represents the data distribution routing function.

[0041] In this invention, different data types require different processing resources. The data distribution routing function can select the most suitable parsing thread pool based on the data type (such as structured, semi-structured, and unstructured data) to optimize the allocation of computing resources.

[0042] S402: Use Apache-Kafka or Flink to receive real-time data streams, and process the real-time data streams into chunks according to time windows to obtain data items. d i Multidimensional data:

[0043] in, D w Indicates data items within the time window d i Multidimensional data, t start Indicates the starting time window. t end The time window indicates the end of the window, and the timestamp represents the timestamp.

[0044] Apache Kafka is an open-source stream processing platform used for high-throughput, distributed messaging, log aggregation, and real-time data stream processing. Originally developed by LinkedIn and later becoming part of the Apache Software Foundation, it is widely used in big data systems.

[0045] Apache Flink is an open-source stream processing framework originally developed by Data Artisans, a Berlin-based technology company (later acquired by Alibaba), and now part of the Apache Software Foundation. Flink supports not only streaming data processing but also batch processing, enabling it to handle both real-time and offline data streams. It is widely used in real-time data analytics, big data processing, and event-driven applications.

[0046] In this invention, by using tools such as Apache Kafka or Flink to shard the data, the system can respond to each data item more quickly, ensuring efficient transmission and processing of real-time data streams. Each data item can be quickly routed to the corresponding parsing thread, reducing latency. Simultaneously, sharding the data stream according to time windows enables the system to process real-time data streams sequentially, improving processing speed and response time.

[0047] S403: For data items d i Parallel classification and routing of multi-dimensional data.

[0048] In one possible implementation, S403 specifically includes: The following formula is used to analyze the data items. d i Parallel classification and routing of multi-dimensional data:

[0049] in, Represents any symbol; TypeClassifier represents the classifier configured for the data format. T j Indicates and D s , D h , D u The corresponding thread pool, j =1,2,3.

[0050] In this invention, through data classification functions and data distribution routing functions, the system can dynamically and intelligently classify and route data according to the type of real-time data stream. This allows the system to flexibly handle different data types and can be extended to include new data types and parsing methods as needed.

[0051] S5: Call the multimodal feature parsing model in each parsing thread to perform feature parsing on the received multidimensional data and output the multimodal features corresponding to the received multidimensional data.

[0052] It should be noted that the system calls multimodal feature parsing models in different parsing threads to parse the real-time received multidimensional data, thereby extracting the multimodal features corresponding to each data point. Each parsing thread uses a specially designed parsing model to process the data according to its type, and finally outputs multimodal features that match the data.

[0053] In this invention, the multimodal feature parsing model processes data independently in each parsing thread, enabling feature output in a shorter time, which meets the requirements of real-time data stream processing. By reducing processing latency and accelerating data stream transmission, the system's response speed is improved. Simultaneously, data type parsing is performed in independent threads, avoiding data queuing and waiting, thereby reducing overall processing latency, which is crucial, especially in applications requiring real-time responses.

[0054] S6: Using a pre-deployed data missing detection model, identify whether any multimodal features are missing. If so, match the corresponding data completion strategy to complete the missing features. Otherwise, return to step S3.

[0055] Among them, the missing data detection model is a machine learning model used to identify missing or incomplete data in a dataset or data stream. Its core task is to automatically identify the missing parts of the data by analyzing the features or patterns of the data, thereby providing support for subsequent data processing (such as data completion, imputation, or repair).

[0056] In one possible implementation, the method for generating the data missing detection model specifically includes: S601: Collect multiple sets of multi-dimensional data.

[0057] S602: Using a multimodal feature analysis model, analyze each group of multidimensional data to obtain multiple first multimodal features corresponding to each group of multidimensional data.

[0058] S603: Select the second multimodal features that have missing features in each of the first multimodal features, and mark the missing features of the second multimodal features.

[0059] S604: Use a data completion strategy to annotate the missing features in the labeled second multimodal features.

[0060] S605: Based on the labeled second multimodal features, construct a feature set and divide the feature set into a training set and a validation set according to a preset ratio.

[0061] It should be noted that those skilled in the art can set the preset ratio according to actual needs, and this invention does not limit that.

[0062] S606: Input the training set into the data missing detection model based on the random forest model, perform classification learning of multimodal features and corresponding labeled features and labeling strategies, and train the data missing detection model.

[0063] Random Forest is an ensemble learning method primarily used for classification and regression tasks. It improves the accuracy and robustness of the model by constructing multiple decision trees and combining their results. While based on decision trees, Random Forest differs from a single decision tree by integrating multiple decision trees to reduce overfitting and improve predictive performance.

[0064] In this invention, by utilizing a random forest model for missing data detection, the system can accurately identify which features are missing and learn from training data to automatically determine missing patterns and completion strategies. The ensemble learning characteristics of random forests enable them to handle complex feature relationships, improving the accuracy of missing data detection.

[0065] S607: Validate the missing data detection model's performance using a validation set. If validation passes, deploy and apply the missing data detection model. Otherwise, repeat the training process.

[0066] In this invention, a data missing detection model enables the system to automatically detect missing portions of data. This is crucial for complex multimodal data, as different types of data (such as images, text, and structured data) may exhibit different data missing patterns. Ensuring data integrity is fundamental to high-quality analysis; missing data will not affect subsequent processing. Furthermore, once missing data is detected, the system can automatically match an appropriate completion strategy for processing, avoiding human intervention and errors in human judgment, thereby improving the accuracy and reliability of the data.

[0067] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this embodiment of the invention, by constructing a multimodal feature parsing model and allocating independent parsing threads for different data types, the system can efficiently process multidimensional heterogeneous data, avoiding thread stalling and inefficiency. Real-time data streams are accurately assigned to corresponding threads after type matching, improving processing efficiency. Simultaneously, multimodal features are extracted from the data through feature parsing, and missing data is automatically filled in using a data missing detection model, ensuring the integrity and high quality of the parsed data.

[0068] Reference manual attached Figure 2 The diagram shows a schematic of the structure of a real-time data processing device based on multi-dimensional intelligent analysis provided by the present invention.

[0069] The present invention also provides a real-time data processing device 20 based on multi-dimensional intelligent analysis, comprising: Module 201 is used to build a multimodal feature parsing model and deploy it to each parsing thread.

[0070] In one possible implementation, the multimodal feature parsing model is specifically as follows:

[0071] in, F This represents a multimodal feature parsing model, where Concat represents the feature concatenation operation. MIP This represents a multilayer perceptron. D s Represents a structured data matrix. GAT This represents a graph attention network. D h Represents a semi-structured data graph. Transformer- CNN Indicates combination Transformer and CNN Hybrid models for parsing unstructured data D u This represents unstructured data.

[0072] Allocation module 202 is used to allocate independent parsing threads for different data types.

[0073] The acquisition module 203 is used to acquire real-time data streams and match the corresponding data types for each multi-dimensional data in the real-time data stream.

[0074] The sending module 204 is used to send the multi-dimensional data in the real-time data stream to the corresponding parsing thread according to the correspondence between the data type and the parsing thread.

[0075] In one possible implementation, the sending module is specifically used for: Pre-configure the corresponding data type classification function, thread pool mapping table, and data distribution routing function for each parsing thread:

[0076] in, t i express D s , D h , D u The Middle i The type label for each data item, TypeClassifier, indicates the classifier configured for the data format. d i express D s , D h , D u The Middle i Data items

[0077] in, T This represents the thread pool mapping table. T 1 represents D s The parsing thread pool, T 2 indicates D h The parsing thread pool, T 3 indicates D u The parsing thread pool.

[0078]

[0079] Here, Route() represents the data distribution routing function.

[0080] Use Apache-Kafka or Flink to receive real-time data streams, and then segment the real-time data streams according to time windows to obtain data items. d i Multidimensional data:

[0081] in, D w Indicates data items within the time window d i Multidimensional data, t start Indicates the starting time window. t endThe time window indicates the end of the window, and the timestamp represents the timestamp.

[0082] For data items d i Parallel classification and routing of multi-dimensional data.

[0083] In one possible implementation, the sending module is specifically used for: The following formula is used to analyze the data items. d i Parallel classification and routing of multi-dimensional data:

[0084] in, Represents any symbol; TypeClassifier represents the classifier configured for the data format. T j Indicates and D s , D h , D u The corresponding thread pool, j =1,2,3.

[0085] Module 205 is used to call the multimodal feature parsing model in each parsing thread to perform feature parsing on the received multidimensional data and output the multimodal features corresponding to the received multidimensional data.

[0086] The identification module 206 uses a pre-deployed data missing detection model to identify whether any multimodal features are missing. If so, it matches the corresponding data completion strategy to complete the missing features. Otherwise, it returns to the acquisition module 203.

[0087] In one possible implementation, the method for generating the data missing detection model specifically includes: Collect multiple sets of multi-dimensional data.

[0088] By using a multimodal feature analysis model, the multidimensional data of each group is analyzed to obtain multiple first multimodal features corresponding to each group of multidimensional data.

[0089] Select the second multimodal features that have missing features in each of the first multimodal features, and mark the missing features of the second multimodal features.

[0090] Using a data completion strategy, missing features in the labeled second multimodal features are annotated.

[0091] Based on the labeled second multimodal features, a feature set is constructed, and the feature set is divided into a training set and a validation set according to a preset ratio.

[0092] The training set is input into a data missing detection model based on a random forest model to perform classification learning of multimodal features and corresponding labeled features and labeling strategies, thereby training the data missing detection model.

[0093] Use the validation set to verify the performance of the missing data detection model. If the validation passes, deploy and apply the missing data detection model. Otherwise, repeat the training process.

[0094] The real-time data processing device 20 based on multi-dimensional intelligent analysis provided by the present invention can execute the above-mentioned real-time data processing method based on multi-dimensional intelligent analysis and achieve the same or similar technical effects. To avoid repetition, the present invention will not elaborate further.

[0095] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this embodiment of the invention, by constructing a multimodal feature parsing model and allocating independent parsing threads for different data types, the system can efficiently process multidimensional heterogeneous data, avoiding thread stalling and inefficiency. Real-time data streams are accurately assigned to corresponding threads after type matching, improving processing efficiency. Simultaneously, multimodal features are extracted from the data through feature parsing, and missing data is automatically filled in using a data missing detection model, ensuring the integrity and high quality of the parsed data.

[0096] It should be understood that the processor in the embodiments of the present invention can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0097] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0098] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0099] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0100] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0101] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0102] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0103] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0104] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0105] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0106] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0107] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0108] This invention provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the real-time data processing method based on multi-dimensional intelligent parsing as described in the method embodiment.

[0109] The present invention provides a computer-readable storage medium that can implement the steps and effects of the real-time data processing method based on multi-dimensional intelligent parsing in the above-described method embodiments. To avoid repetition, the present invention will not elaborate further.

[0110] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this embodiment of the invention, by constructing a multimodal feature parsing model and allocating independent parsing threads for different data types, the system can efficiently process multidimensional heterogeneous data, avoiding thread stalling and inefficiency. Real-time data streams are accurately assigned to corresponding threads after type matching, improving processing efficiency. Simultaneously, multimodal features are extracted from the data through feature parsing, and missing data is automatically filled in using a data missing detection model, ensuring the integrity and high quality of the parsed data.

[0111] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0112] The following points need to be explained: (1) The accompanying drawings of the embodiments of the present invention only involve the structures involved in the embodiments of the present invention. Other structures can refer to the general design.

[0113] (2) For clarity, the thickness of layers or regions is enlarged or reduced in the drawings used to describe embodiments of the invention, i.e., these drawings are not drawn to scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element or there may be intermediate elements.

[0114] (3) Where there is no conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0115] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A real-time data processing method based on multi-dimensional intelligent analysis, characterized in that, include: S1: Construct a multimodal feature parsing model and deploy the multimodal feature parsing model to each parsing thread; S2: Allocate independent parsing threads for different data types. S3: Acquire the real-time data stream and match the corresponding data type for each multi-dimensional data in the real-time data stream; S4: Send each multi-dimensional data in the real-time data stream to the corresponding parsing thread according to the correspondence between the data type and the parsing thread; S5: Call the multimodal feature parsing model in each parsing thread to perform feature parsing on the received multidimensional data and output the multimodal features corresponding to the received multidimensional data; S6: Using a pre-deployed data missing detection model, identify whether each of the multimodal features has missing features; if so, match the corresponding data completion strategy to complete the missing features; Otherwise, return to step S3.

2. The real-time data processing method based on multi-dimensional intelligent analysis according to claim 1, characterized in that, The multimodal feature parsing model is specifically as follows: ； in, F This represents a multimodal feature parsing model, where Concat represents the feature concatenation operation. MIP This represents a multilayer perceptron. D s Represents a structured data matrix. GAT This represents a graph attention network. D h Represents a semi-structured data graph. Transformer-CNN Indicates combination Transformer and CNN Hybrid models for parsing unstructured data D u This represents unstructured data.

3. The real-time data processing method based on multi-dimensional intelligent analysis according to claim 2, characterized in that, S4 specifically includes: S401: Pre-configure the corresponding data type classification function, thread pool mapping table, and data distribution routing function for each of the parsing threads: ； in, t i express D s , D h , D u The Middle i The type label for each data item, TypeClassifier, indicates the classifier configured for the data format. d i express D s , D h , D u The Middle i One data item; ； in, T This represents the thread pool mapping table. T 1 represents D s The parsing thread pool, T 2 indicates D h The parsing thread pool, T 3 indicates D u The parsing thread pool; ； Here, Route() represents the data distribution routing function; S402: Receive the real-time data stream using Apache-Kafka or Flink, and segment the real-time data stream according to a time window to obtain data items. d i Multidimensional data: ； in, D w Indicates data items within the time window d i Multidimensional data, t start Indicates the starting time window. t end The time window indicates the end of the window, and the timestamp represents the timestamp. S403: For the data item d i Parallel classification and routing of multi-dimensional data.

4. The real-time data processing method based on multi-dimensional intelligent analysis according to claim 3, characterized in that, Specifically, S403 is: The data item is processed using the following formula. d i Parallel classification and routing of multi-dimensional data: ； in, Represents any symbol; TypeClassifier represents the classifier configured for the data format. T j Indicates and D s , D h , D u The corresponding thread pool, j =1,2,3.

5. The real-time data processing method based on multi-dimensional intelligent analysis according to claim 1, characterized in that, The method for generating the data missing detection model specifically includes: S601: Collect multiple sets of multi-dimensional data; S602: Using the multimodal feature analysis model, analyze each group of multidimensional data to obtain multiple first multimodal features corresponding to each group of multidimensional data; S603: Select the second multimodal features that have missing features in each of the first multimodal features, and mark the missing features of the second multimodal features; S604: Use a data completion strategy to annotate the missing features in the labeled second multimodal features; S605: Based on the labeled second multimodal features, construct a feature set, and divide the feature set into a training set and a validation set according to a preset ratio; S606: Input the training set into the data missing detection model based on the random forest model, perform classification learning of multimodal features and corresponding labeled features and labeling strategies, and train the data missing detection model. S607: Use the validation set to verify the recognition performance of the missing data detection model; if the verification passes, deploy and apply the missing data detection model; otherwise, repeat the training process.

6. A real-time data processing device based on multi-dimensional intelligent analysis, characterized in that, include: The building module is used to construct the multimodal feature parsing model and deploy it to each parsing thread; The allocation module is used to allocate independent parsing threads for different data types; The acquisition module is used to acquire real-time data streams and match corresponding data types for each multi-dimensional data in the real-time data streams. The sending module is used to send each multi-dimensional data in the real-time data stream to the corresponding parsing thread according to the correspondence between the data type and the parsing thread; The calling module is used to call the multimodal feature parsing model in each parsing thread to perform feature parsing on the received multidimensional data and output the multimodal features corresponding to the received multidimensional data; The identification module is used to identify whether each of the multimodal features has missing features using a pre-deployed data missing detection model; if so, it matches the corresponding data completion strategy to complete the missing features; otherwise, it returns to the acquisition module.

7. The real-time data processing device based on multi-dimensional intelligent analysis according to claim 6, characterized in that, The multimodal feature parsing model is specifically as follows: ； in, F This represents a multimodal feature parsing model, where Concat represents the feature concatenation operation. MIP This represents a multilayer perceptron. D s Represents a structured data matrix. GAT This represents a graph attention network. D h Represents a semi-structured data graph. Transformer-CNN Indicates combination Transformer and CNN Hybrid models for parsing unstructured data D u This represents unstructured data.

8. The real-time data processing device based on multi-dimensional intelligent analysis according to claim 7, characterized in that, The sending module is specifically used for: Pre-configure the corresponding data type classification function, thread pool mapping table, and data distribution routing function for each of the parsing threads: ； in, t i express D s , D h , D u The Middle i The type label for each data item, TypeClassifier, indicates the classifier configured for the data format. d i express D s , D h , D u The Middle i One data item; ； in, T This represents the thread pool mapping table. T 1 represents D s The parsing thread pool, T 2 indicates D h The parsing thread pool, T 3 indicates D u The parsing thread pool; ； Here, Route() represents the data distribution routing function; The real-time data stream is received using Apache Kafka or Flink, and then fragmented according to a time window to obtain data items. d i Multidimensional data: ； in, D w Indicates data items within the time window d i Multidimensional data, t start Indicates the starting time window. t end The time window indicates the end of the window, and the timestamp represents the timestamp. For the data item d i Parallel classification and routing of multi-dimensional data.

9. The real-time data processing device based on multi-dimensional intelligent analysis according to claim 8, characterized in that, The sending module is specifically used for: The data item is processed using the following formula. d i Parallel classification and routing of multi-dimensional data: ； in, Represents any symbol; TypeClassifier represents the classifier configured for the data format. T j Indicates and D s , D h , D u The corresponding thread pool, j =1,2,3.

10. The real-time data processing device based on multi-dimensional intelligent analysis according to claim 6, characterized in that, The method for generating the data missing detection model specifically includes: Collect multiple sets of multi-dimensional data; Using the multimodal feature parsing model, the multidimensional data of each group is parsed to obtain multiple first multimodal features corresponding to each group of multidimensional data; Select the second multimodal features that have missing features in each of the first multimodal features, and mark the missing features of the second multimodal features; Using a data completion strategy, feature annotation is performed on the missing features in the labeled second multimodal features; Based on the labeled second multimodal features, a feature set is constructed, and the feature set is divided into a training set and a validation set according to a preset ratio; The training set is input into a data missing detection model based on a random forest model to perform classification learning of multimodal features and corresponding labeled features and labeling strategies, thereby training the data missing detection model. The recognition performance of the data missing detection model is verified using the validation set. If the verification passes, the data missing detection model is deployed and applied; otherwise, the training process is repeated.

Images