Data feature management method, computer program product and electronic equipment

Abstract

The invention belongs to the technical field of big data, and particularly relates to a data feature management method, a computer program product and electronic equipment. The data feature management method comprises the following steps: acquiring service feature data in a plurality of scenes; classifying the business feature data according to a classification model to obtain a classification result, the classification result comprising the type of each piece of business feature data; performing evaluation processing according to the business feature data of each type to obtain classification evaluation data of the classification model on the business feature data; performing optimization processing on the classification model according to the classification evaluation data to obtain an optimized classification model; and performing data feature management based on the optimized classification model. Therefore, the data management efficiency can be improved.

Description

Technical Field

[0001] This invention relates to the field of big data technology, and in particular to a data feature management method, computer program product, and electronic device. Background Technology

[0002] In today's society, there is an increasing amount of unstructured or structured data from various sources. Traditional data processing applications are no longer sufficient to handle large or complex datasets, which has led to the emergence of big data. Big data usually contains data that exceeds the processing capacity of traditional software within an acceptable timeframe, and the emergence of big data has spurred novel research on a wide range of topics.

[0003] However, big data lacks statistical capabilities; it merely observes and tracks events. While some data feature management methods exist in the big data field, each has significant shortcomings. For example, current traditional feature classification and management methods lack a closed-loop chain that optimizes feature classification algorithms from downstream application results, leading to inaccurate overall classification and a lack of feedback mechanisms. They even rely on verbal communication, resulting in information loss or inaccurate feedback. Furthermore, fine-grained data classification in the big data field depends on personnel's business knowledge, resulting in low data processing efficiency. Summary of the Invention

[0004] This application aims to at least solve the technical problems existing in the prior art and provide a data feature management method, computer program product, and electronic device.

[0005] In a first aspect, this application provides a data feature management method, comprising: acquiring business feature data in multiple scenarios; classifying the business feature data according to a classification model to obtain classification results, wherein the classification results include the types of each business feature data; evaluating each type of business feature data to obtain classification evaluation data of each business feature data by the classification model; optimizing the classification model according to the classification evaluation data to obtain an optimized classification model; and performing data feature management based on the optimized classification model.

[0006] In one embodiment, acquiring business feature data under multiple scenarios includes: acquiring raw data under multiple scenarios; and performing feature extraction processing on the raw data to obtain the business feature data.

[0007] In one embodiment, the step of performing feature extraction processing on the original data to obtain the business feature data includes: determining a corresponding feature extraction strategy based on the data type of the original data; and performing feature extraction on the original data according to the corresponding feature extraction strategy to obtain the business feature data.

[0008] In one embodiment, the step of evaluating and processing the business feature data of each type to obtain the classification evaluation data of each business feature data by the classification model includes: determining the business feature data of each scenario corresponding to each scenario from the business feature data according to the scenario type of each scenario; and processing the business feature data of each scenario corresponding to each scenario based on a preset evaluation model to obtain the classification evaluation data of each business feature data by the classification model.

[0009] In one embodiment, the step of processing business feature data of corresponding types for each scenario based on a preset evaluation model to obtain classification evaluation data of each business feature data by the classification model includes: inputting each of the business feature data into a preset scenario model corresponding to each scenario to obtain evaluation data of each of the business feature data; inputting each of the business feature data into a preset plurality of indicator monitoring models to obtain a plurality of indicator data monitored by the classification model; and using at least one of the evaluation data and the indicator data as the classification evaluation data.

[0010] In one embodiment, optimizing the classification model based on the classification evaluation data to obtain an optimized classification model includes: obtaining a model score based on the classification evaluation data; and adjusting the parameters in the classification model according to a preset adjustment algorithm in response to the model score being less than a preset score threshold, thereby obtaining the optimized classification model.

[0011] In one embodiment, optimizing the classification model based on the classification evaluation data to obtain an optimized classification model includes: obtaining a model score based on the classification evaluation data; in response to the model score being less than a preset score threshold, obtaining adjustment information for the classification model based on the classification evaluation data; and adjusting the classification model based on the adjustment information to obtain the optimized classification model.

[0012] In one embodiment, the data feature management based on the optimized classification model includes: acquiring current feature data; inputting the current feature data into the optimized classification model to obtain a classification result of the current feature data; and storing the current feature data in a corresponding data management database according to the classification result of the current feature data.

[0013] Secondly, this application provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the data feature management method of any one of claims 1-7.

[0014] Thirdly, this application provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the data feature management method as described in any one of claims 1 to 7.

[0015] The data feature management method, computer program product, and electronic device provided in this application evaluate and process various types of business feature data to obtain classification evaluation data for each business feature data using a classification model; then, the classification model is optimized based on the classification evaluation data to obtain an optimized classification model; and data feature management is performed based on the optimized classification model. Therefore, the classification model can be optimized in real time based on the classification results, improving classification accuracy and thus enhancing the accuracy of data management.

[0016] In addition, classifying business feature data using a classification model before data management is more efficient than manual classification. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a data feature management method provided by the present invention;

[0018] Figure 2 This is an application diagram illustrating how the present invention acquires business feature data;

[0019] Figure 3 This is another application diagram of acquiring business feature data provided by the present invention;

[0020] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0021] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0022] The execution entity of the data feature management method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in the embodiments of this application: a server, a terminal, etc. In other words, the data feature management method can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0023] Specifically, combined Figure 1 As shown in the figure, this embodiment provides a data feature management method, which includes the following steps:

[0024] Step S110: Obtain business feature data from multiple scenarios.

[0025] Multiple business scenarios can include government affairs scenarios, urban governance scenarios, etc.

[0026] Business characteristic data refers to the characteristic data of business operations in a given scenario.

[0027] It can acquire business characteristic data across multiple scenarios. For example, it can collect characteristic data of business operations in multiple scenarios in real time to obtain business characteristic data for multiple scenarios. Alternatively, it can retrieve business characteristic data for multiple scenarios from a pre-set database.

[0028] Step S120: Classify the business feature data according to the classification model to obtain the classification results, which include the type of each business feature data.

[0029] A classification model is a model that classifies data into different types. One such model is the trained BERT, which is a pre-trained language model based on Transformer.

[0030] In one embodiment, during the training of the classification model, the semantic information and contextual relationships of the feature data samples are obtained through natural language processing (NLP) methods, and feature representations of the feature data samples are generated, such as output vectors labeled [CLS], and the semantic information of the entire feature data sample is output.

[0031] In a preferred embodiment, the loss function of the classification model is used to measure the difference between the probability distribution predicted by the model and the probability distribution of the true labels, and the loss function is as follows:

[0032] L loss =-Σ i y i *log(y' i +Δ i )

[0033] Among them, y i It is the built-in probability distribution of the true label of sample i. i This is for y i The weighted values ​​of the probability variables set can be modified through a device or system, y' i The predicted probability distribution is obtained through a fully connected (dense) layer and a softmax activation function. The weights of the fully connected layer are predefined variables, and the activation function does not need to be modified.

[0034] In another preferred embodiment, the LightGBM algorithm is used as the feature representation vector classification label during the training process of the classification model, and weight values ​​and different calculation rules are set for each data type. The multi-class log loss function of the model is expressed by the following formula:

[0035]

[0036] Where N is the number of samples; K is the number of categories; y ij y' is the true label of category j for sample i; ik λ is the probability that the model predicts the class k of sample i, obtained through the built-in softmax activation function; ij The weights are the rule loss values, which can be set individually by the device or initialized uniformly. `rule(,)` is a separately set rule function, a matrix where `rule(i,j)` represents the rule function value for class `j` of sample `i` in `rule(,)`, used to balance λ. ij The dynamically set rules can be implemented without modifying λ. ij In this case, λ can be adjusted by setting the value of rule(i,j) on the device page. ij `rule(i,j)` represents the initial matrix values, each of which can be random or manually set on the device page. The types of business characteristic data can include transaction type, behavior type, user type, and hobby type, etc.

[0037] The business feature data is classified according to the classification model to obtain the classification results. Specifically, each business feature data is input into the classification model to obtain the classification results for each business feature data.

[0038] The classification model includes classification reasoning patterns and associated feature representation vectors. Based on multiple classification algorithms and preset classification rules and thresholds, the feature representation vectors are classified to obtain the classification results.

[0039] The feature representation vector is a high-dimensional feature vector that represents the semantic information of the entire feature data, reflecting the statistical and semantic characteristics obtained by the feature in the classification model. For example, for input feature data such as ["File name: example.txt","File size: 1024MB","File type: txt","Author: Zhang San","Creation time: 2024-05-08","Key information: Government data, Confidential data"], BERT generates a 768-dimensional feature representation vector.

[0040] Step S130: Evaluate and process the business characteristic data of each type to obtain the classification evaluation data of each business characteristic data by the classification model.

[0041] The classification assessment data can be used for classification assessment scores.

[0042] Input the business characteristic data of each type into the preset target evaluation model to obtain the classification evaluation data of each business characteristic data by the classification model.

[0043] Step S140: Optimize the classification model based on the classification evaluation data to obtain the optimized classification model.

[0044] In response to the classification evaluation data falling below a preset threshold, the parameters in the classification model are adjusted to obtain an optimized classification model. Parameters can include tree depth, learning rate, regularization terms, etc.

[0045] Step S150: Perform data feature management based on the optimized classification model.

[0046] Data feature management is based on the optimized classification model. The process involves: acquiring the current feature data; inputting the current feature data into the optimized classification model to obtain the classification result; and storing the current feature data in the corresponding data management database based on the classification result.

[0047] Furthermore, the classification results of the current feature data are evaluated to obtain classification evaluation data for the current feature data. If the classification evaluation data satisfies the optimization conditions of the classification model, the optimized classification model is further optimized, achieving iterative optimization. Therefore, continuous real-time optimization of the classification model during data management improves both the accuracy of the classification model and the accuracy of data management.

[0048] Acquire business feature data from multiple scenarios, including: acquiring raw data from multiple scenarios; and performing feature extraction processing on the raw data to obtain business feature data.

[0049] Raw data includes vector data, metadata, etc. Metadata may include file name, size, type, author, creation time, key information, etc.

[0050] Obtain raw data from multiple scenarios. Specifically, obtain source data from multiple scenarios, preprocess the source data, and obtain the raw data.

[0051] The appropriate data source acquisition method is determined based on the type of scenario, and the source data for each scenario is acquired according to the corresponding method. The data source acquisition method can be real-time acquisition or offline access acquisition. For example, data from the data source can be integrated through real-time big data technology or offline batch processing via scheduled tasks. Real-time big data technology can be Flink + message middleware or CDC (Central Data Console) technology, while offline batch processing can be Spark or Hive.

[0052] The source data is preprocessed to obtain the raw data. Specifically, the metadata in the source data is extracted and used as the raw data.

[0053] For example, if the source data includes structured or semi-structured data, the metadata information and actual data are obtained from the metadata table in the structured or semi-structured data; if the source data includes unstructured data, the unstructured data is converted into vector data using vectorization technology, and the metadata information in the vector data is obtained.

[0054] In one embodiment, unstructured data can be audio, video, images, documents, etc. TF-IDF is used to extract word frequencies and inverse document frequencies from the text in the document, forming vector values ​​and metadata to obtain the original data corresponding to the document. Image feature extraction algorithms are used to extract key points and descriptive information from the image, forming vectors and image metadata to obtain the original data corresponding to the image. Image feature extraction algorithms can be SIFT (Scale Invariant Feature Transform), SURF (Speed-Up Robust Feature Transform), etc. Mel spectrogram transformation is used to process the audio, capturing the audio frequencies and audio metadata to obtain the original data corresponding to the audio. For each frame of the video, vectors are extracted using an image feature extraction algorithm, and the vectors of all frames and the image metadata are combined. A convolutional neural network (CNN) is used to extract visual features and video metadata. The vectors of all frames, image metadata, visual features, and video metadata are used as the original data corresponding to the video.

[0055] It should be noted that business feature data is stored in the feature data pool, and unstructured data and structured data are stored together.

[0056] Combination Figure 2 As shown, third-party source data can be obtained through files, interfaces, system imports, data entry, and databases, and integrated into the management system. Various types of business source data can also be generated through business systems, such as events, contingency plans, risk points, equipment information, and user behavior. The source data is written to the database and sent to a message middleware. Batch unified calculations are performed on the source data to obtain vector features and metadata. The metadata undergoes feature extraction, transformation, annotation, and classification to obtain raw data. Features are extracted from the raw data to obtain business feature data. This business feature data is stored in a feature management database. Features are selected from the feature data pool and the feature management database for model training to obtain a classification model. The classification results of the model are evaluated using an evaluation algorithm to obtain classification evaluation data. If the classification evaluation data meets the optimization conditions of the classification model, feature feedback is provided.

[0057] In one embodiment, combined with Figure 3 As shown, source data is retrieved from a source database, message middleware, or unstructured storage device. Batch data processing is performed on the source data to obtain vector features and metadata. The metadata undergoes feature extraction, transformation, annotation, and classification to obtain raw data. This raw data is then sent to the data warehouse and feature data parsing module. The feature data parsing module extracts features from the raw data to obtain business feature data. This business feature data is then stored in a feature data pool. The feature data parsing module sends the metadata to the metadata management module, which is interconnected with the data warehouse.

[0058] The process of extracting features from the raw data to obtain business feature data includes: determining the corresponding feature extraction strategy based on the data type of the raw data; and extracting features from the raw data according to the corresponding feature extraction strategy to obtain business feature data.

[0059] Before determining the corresponding feature extraction strategy based on the data type of the original data, the process also includes: format processing of the original data through a data parsing module to obtain raw data with a processable format, and storing the processable raw data in the feature variable pool. Thus, the data parsing module can quickly and accurately convert the original data into a processable format, which is beneficial for subsequent data processing.

[0060] The feature extraction strategy is determined based on the data type of the original data. Specifically, the original data is parsed using a preset parsing module to obtain the original data fields; the corresponding feature extraction strategy is then determined based on the data type of the original data fields.

[0061] In one embodiment, in response to the inclusion of structured data in the original data, the fields in the structured data are used as original data fields; in response to the inclusion of JSON (JavaScript Object Notation) data in the original data, the corresponding values ​​are extracted according to the key-value pairs of the JSON data to obtain the original data fields; in response to the inclusion of XML (Extensible Markup Language) format data in the original data, the field content is extracted according to the tag structure of the XML format data to obtain the original data fields.

[0062] In one embodiment, in response to the original data field being of numeric type, the corresponding feature extraction strategy is determined to be a customized function. The customized function is used to obtain the features of the original data field, resulting in business feature data. The customized function can be a function that calculates features such as mean, variance, maximum value, and minimum value. In response to the original data field being of text type, the corresponding feature extraction strategy is determined to be natural language processing technology. For example, if the original data field is of String or Text type, a word vector model such as Word2Vec or BERT is used to obtain the features of the original data field, resulting in business feature data. In response to the original data field being of a preset type, the corresponding feature extraction strategy is determined to be a preset feature extraction method. For example, if the preset type is video, visual features are extracted using a convolutional neural network (CNN) to obtain business feature data.

[0063] It should be noted that the process involves collecting source data, sending the source data to the message middleware, processing the source data obtained through the message middleware to obtain raw data, storing the raw data in a format that can be processed into the feature variable pool, selecting raw data from the feature variable pool for feature extraction to obtain business feature data, and storing the business feature data into the feature data pool.

[0064] The evaluation and processing of business feature data of various types are carried out to obtain the classification evaluation data of each business feature data by the classification model, including: determining the corresponding type of business feature data of each scenario from each business feature data according to the scenario type of each scenario; and processing the corresponding type of business feature data of each scenario based on the preset evaluation model to obtain the classification evaluation data of each business feature data by the classification model.

[0065] The preset evaluation model can include scenario evaluation models corresponding to multiple scenarios, such as e-commerce scenario evaluation model, bank loan scenario evaluation model, and urban governance scenario evaluation model.

[0066] Based on the scenario type of each scenario, the business feature data corresponding to each scenario type is determined from the business feature data. Specifically, based on the scenario type of each scenario, the corresponding target type is determined from a preset data type mapping table, and the business feature data whose type matches the target type is determined as the business feature data corresponding to the scenario.

[0067] In one embodiment, the preset data type mapping table includes e-commerce recommendation scenarios, where the corresponding business feature data types can be transactions, behaviors, users, hobbies, etc.; bank loan scenarios, where the corresponding business feature data types can be credit reports, behaviors, blacklists, transaction records, users, etc.; and urban governance scenarios, where the corresponding business feature data types can be traffic, flow, public opinions, users, etc.

[0068] Classification evaluation data can include scores from multiple dimensions, such as feature validity score, correlation score, feature usage rate score, feature change frequency score, and feature freshness score.

[0069] Based on a preset evaluation model, the business feature data of each scenario type is processed to obtain the classification evaluation data of each business feature data by the classification model. This includes: inputting the business feature data of each scenario type into the scenario model corresponding to each scenario to obtain the evaluation data of each business feature data; inputting the business feature data of each scenario type into multiple indicator monitoring models to obtain multiple indicator data monitored by the classification model; and using the evaluation data of each business feature data and at least one of the multiple indicator data monitored by the classification model as the classification evaluation data.

[0070] Evaluation data may include at least one of feature importance, feature usage frequency, and feature contribution. Feature importance characterizes the contribution of a feature to the model's predictions. Feature usage frequency characterizes how often a feature is used. Feature contribution characterizes the degree of contribution of a feature.

[0071] Metrics data may include accuracy, recall, and F1 score. Accuracy represents the proportion of correct predictions made by the model. Recall represents the proportion of correct predictions of the positive class. The F1 score is the harmonic mean of accuracy and recall. Other metrics for classification models may also include mean squared error (MSE) and mean absolute error (MAE).

[0072] As an example, optimizing a classification model based on classification evaluation data to obtain an optimized classification model includes: obtaining a model score based on the classification evaluation data; and adjusting the parameters in the classification model according to a preset adjustment algorithm in response to the model score being less than a preset score threshold, thereby obtaining the optimized classification model.

[0073] The model score is obtained based on the classification assessment data. Specifically, the average value of the classification assessment data is used as the model score. Alternatively, each indicator in the classification assessment data is multiplied by its corresponding preset weight, and the sum of the products is used as the model score.

[0074] In one embodiment, a feedback mechanism is triggered when the indicator data falls below a preset indicator threshold. This feedback mechanism specifically includes optimizing the classification model based on the classification evaluation data. For example, the model's accuracy is less than 0.8, recall is less than 0.6, or F1 score is less than 0.7, where F1 score is the harmonic mean of accuracy and recall.

[0075] In one embodiment, the preset adjustment algorithm can be an adaptive adjustment algorithm. The adaptive adjustment algorithm can dynamically adjust the threshold of the classification decision in the classification model according to the model's index information. The adaptive adjustment algorithm can be expressed as the following formula:

[0076] R new =R old ±Δ*(X t -X1)

[0077] Where R new It refers to the threshold or weights in the adjusted classification model, R. old X represents the threshold or weights of the classification model before adjustment, Δ is the adjustment factor, and X is the threshold or weights of the model before adjustment. t X1 is the set indicator value (such as the recall target value), and X2 is the indicator in the classification evaluation data.

[0078] As another example, optimizing a classification model based on classification evaluation data to obtain an optimized classification model includes: obtaining a model score based on the classification evaluation data; in response to a model score being less than a preset score threshold, obtaining adjustment information for the classification model based on the classification evaluation data; and adjusting the classification model based on the adjustment information to obtain the optimized classification model.

[0079] Adjustments may be made regarding issues such as excessively high learning rates or model overfitting, excessively high classification thresholds, inappropriate feature selection, or unreasonable feature weight settings.

[0080] In some embodiments, in response to a model accuracy of less than 0.8, which may be due to an excessively high learning rate or model overfitting, the learning rate of the classification model is reduced; in response to a recall of less than 0.6, which may be due to an excessively high classification threshold, the classification threshold is adjusted; and in response to an F1 score of less than 0.7, which may be due to improper feature selection or unreasonable feature weight settings, features with low contribution are removed.

[0081] The method provided by this invention improves the accuracy of the classification model by 15% and the recall rate by 10%, thus helping to enhance the effectiveness of targeted marketing and customer satisfaction for enterprises. Simultaneously, this invention reduces the cost of feature analysis, making it more beneficial for data analysis engineers and training engineers by reducing their learning and editing of business feature tags. Through dynamic annotation, feedback, and classification, it improves their work efficiency by approximately 30%.

[0082] Based on the same inventive concept, this invention also discloses a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the data feature management method provided by this invention. The computer program product should be understood as a software product that mainly implements its solution through a computer program, such as a program product integrated in the cloud or a software library.

[0083] Based on the same inventive concept, the present invention also discloses an electronic device, in one embodiment of which the electronic device includes at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to execute the data feature management method provided by the present invention.

[0084] like Figure 4 The diagram shown is a structural schematic of an electronic device provided in an embodiment of the present invention. The electronic device may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13, and may also include a computer program, such as a data feature management method program, stored in the memory 11 and executable on the processor 10.

[0085] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., executing data feature management methods) and calls data stored in the memory 11 to perform various functions of the electronic device and process data.

[0086] The memory 11 includes at least one type of readable storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of an electronic device, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device, such as a plug-in portable hard drive, SmartMediaCard (SMC), SecureDigital (SD) card, FlashCard, etc. Furthermore, the memory 11 can include both internal and external storage units of the electronic device. The memory 11 can be used not only to store application software and various types of data installed on the electronic device, such as the code of data feature management methods, but also to temporarily store data that has been output or will be output.

[0087] The communication bus 12 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0088] Communication interface 13 is used for communication between the aforementioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, Bluetooth interface, etc.), typically used to establish communication connections between the electronic device and other electronic devices. The user interface may be a display, an input unit (such as a keyboard), or, optionally, a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface.

[0089] Figure 4 Only electronic devices with components are shown; it will be understood by those skilled in the art that... Figure 4The structure shown does not constitute a limitation on the electronic device and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0090] For example, although not shown, the electronic device may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to at least one processor 10 via a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0091] It should be understood that the embodiments are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0092] Furthermore, if the modules / units integrated into the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, a computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0093] In the description of this specification, the references to terms such as "an embodiment," "some embodiments," "example," "specific example," "a implementation," "a preferred implementation," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0094] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A data feature management method, characterized in that, include: Acquire business characteristic data from multiple scenarios; The business feature data is classified according to the classification model to obtain the classification result, which includes the type of each business feature data. The classification model is used to evaluate and process the business characteristic data of each type to obtain the classification evaluation data of each business characteristic data. The classification model is optimized based on the classification evaluation data to obtain an optimized classification model; Data feature management is performed based on the optimized classification model.

2. The method according to claim 1, characterized in that, The acquisition of business feature data in multiple scenarios includes: Acquire raw data from multiple scenarios; The original data is subjected to feature extraction processing to obtain the business feature data.

3. The method according to claim 2, characterized in that, The process of extracting features from the original data to obtain the business feature data includes: Determine the corresponding feature extraction strategy based on the data type of the original data; The original data is subjected to feature extraction according to the corresponding feature extraction strategy to obtain the business feature data.

4. The method according to claim 1, characterized in that, The evaluation process based on various types of business characteristic data yields the classification evaluation data for each type of business characteristic data obtained by the classification model, including: Based on the scenario type of each scenario, determine the corresponding type of business feature data for each scenario from the business feature data; Based on the preset evaluation model, the business feature data corresponding to each scenario are processed to obtain the classification evaluation data of each business feature data by the classification model.

5. The method according to claim 4, characterized in that, The process of processing business feature data corresponding to each scenario based on a preset evaluation model to obtain classification evaluation data for each business feature data by the classification model includes: Each of the aforementioned business feature data is input into a preset scenario model corresponding to each scenario to obtain evaluation data for each of the aforementioned business feature data; The business characteristic data are input into multiple preset indicator monitoring models to obtain multiple indicator data for monitoring the classification model. At least one of the aforementioned evaluation data and the aforementioned indicator data shall be used as the classification evaluation data.

6. The method according to claim 1, characterized in that, The step of optimizing the classification model based on the classification evaluation data to obtain an optimized classification model includes: The model score is obtained based on the classification and evaluation data. In response to the model score being less than a preset score threshold, the parameters in the classification model are adjusted according to a preset adjustment algorithm to obtain the optimized classification model.

7. The method according to claim 1, characterized in that, The step of optimizing the classification model based on the classification evaluation data to obtain an optimized classification model includes: The model score is obtained based on the classification and evaluation data. In response to the model score being less than a preset score threshold, adjustment information for the classification model is obtained based on the classification evaluation data; The classification model is adjusted based on the adjustment information to obtain the optimized classification model.

8. The method according to claim 1, characterized in that, The data feature management based on the optimized classification model includes: Obtain current feature data; The current feature data is input into the optimized classification model to obtain the classification result of the current feature data; The current feature data is stored in the corresponding data management database based on the classification result of the current feature data.

9. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of any one of the data feature management methods of claims 1-7.

10. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the data feature management method as described in any one of claims 1 to 7.

Images