Method and system for identifying multi-type monotone relationship in monotone ordered classification task
By establishing a database in monotonous and orderly classification tasks and using machine learning algorithms for domain division and data relationship comparison, multiple monotonic relationships in the task are identified, which solves the problem of insufficient accuracy in the identification of multi-type monotonic relationships in the existing technology, and improves the accuracy and efficiency of classification tasks.
Patent Information
- Application Number
- CN202510068153.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-16
- Publication Date
- 2025-05-30
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing monotonic and orderly classification task recognition technology lacks the accuracy of identifying multi-type monotonic relationships, making it difficult to effectively deal with complex monotonic and orderly classification tasks.
By establishing a database, domain classification and data relationship comparison module, combined with machine learning algorithms, the data in the monotonic and orderly classification task is preprocessed, domain division and comparison and analysis of multiple monotonic relationships, and multiple monotonic relationships exist in the task.
It improves the accuracy and efficiency of classification tasks, enhances the adaptability and flexibility of the system, ensures the accuracy and reliability of the identification results, and makes the results more intuitive through visualization technology.
Smart Images
Figure CN120067747A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of monotonic ordered classification, and particularly relates to a method and system for identifying multiple types of monotonic relationships in monotonic ordered classification tasks. Background Art
[0002] A monotonic ordered classification task refers to a classification task in which there is a monotonic relationship between an attribute and a decision. This relationship means that as the attribute value increases or decreases, the decision value also shows a corresponding increasing or decreasing trend. For example, in credit evaluation, as the income and educational level of a depositor increase, their credit rating in the credit assessment will also increase accordingly, which is a monotonically increasing relationship.
[0003] Multiple types of monotonic relationships refer to various monotonically increasing or decreasing relationships that may exist in a monotonic ordered classification task. These relationships may involve different attributes or features, as well as different decisions or goals. For example, in a complex monotonic ordered classification task, there may be multiple monotonic relationships between attributes and decisions at the same time, and these relationships may be increasing or decreasing.
[0004] Existing monotonic ordered classification task recognition has a high recognition rate for single-type monotonic relationships, but the recognition accuracy for multiple types of monotonic relationships is insufficient. Therefore, it is necessary to propose a method and system for identifying multiple types of monotonic relationships in monotonic ordered classification tasks. Summary of the Invention
[0005] To solve the above problems, the present invention provides a method and system for identifying multiple types of monotonic relationships in monotonic ordered classification tasks. Through a database establishment module, a domain classification module, and a data relationship comparison module, the domain of the monotonic ordered classification task is divided, and then by comparing the relationships between the data in the monotonic ordered classification task, a comparative analysis is carried out to identify multiple monotonic relationships existing in the monotonic ordered classification task.
[0006] To achieve the above object, the technical solution of the present invention is as follows: A system for identifying multiple types of monotonic relationships in monotonic ordered classification tasks, comprising:
[0007] An input module, configured to input a monotonic ordered classification task to be detected and preprocess the data in the monotonic ordered classification task to be detected.
[0008] A database establishment module, configured to establish an identification database for identifying different monotonic ordered classification tasks according to the monotonic ordered classification task input by the input module.
[0009] A domain classification module, configured to combine the data in the monotonic ordered classification task with the identification database according to a machine learning algorithm and perform domain division.
[0010] A data relationship comparison module, which is used to deeply compare and analyze by combining the domain classification results obtained in the domain classification module and the data relationships within the monotonic ordered classification task, and identify various monotonic relationships existing in the monotonic ordered classification task.
[0011] An accuracy detection module; further verifies and adjusts the various monotonic relationships identified by the data relationship comparison module.
[0012] A result output module, which uses visualization technology to output the monotonic relationships in the finally identified monotonic ordered classification task.
[0013] The technical principle of the above solution is as follows: Input the monotonic ordered classification task to be detected through the input module, and preprocess the data in the task. Establish an identification database according to the network public data set through the database establishment module. Use machine learning algorithms to perform domain division on the preprocessed data through the domain classification module. Through the data relationship comparison module, combine the results of domain classification and the data relationships within the monotonic ordered classification task to conduct in-depth comparison and analysis. By comparing the change trends between different features, this module can identify various monotonic relationships existing in the task. The accuracy detection module further verifies and adjusts the identification results of the data relationship comparison module. The result output module uses visualization technology to display the finally identified monotonic relationships.
[0014] The above solution has the following beneficial effects:
[0015] 1. Improve the accuracy and efficiency of the classification task. Through the modular processing flow and advanced machine learning algorithms, the data relationship comparison module can more accurately identify and analyze various monotonic relationships in the monotonic ordered classification task. At the same time, the introduction of steps such as preprocessing and domain classification also improves the efficiency of data processing, enabling the data relationship comparison module to complete the classification task more quickly.
[0016] 2. Enhance the adaptability and flexibility of the system. By establishing an identification database and using machine learning algorithms, the system can perform adaptive processing for different types of monotonic ordered classification tasks.
[0017] Furthermore, the preprocessing includes removing noise and redundant information from the data, and performing standardized conversion on the data format of the monotonic ordered classification task.
[0018] Beneficial effects: The preprocessed data is purer, reducing the interference of noise and redundant information, which helps to improve the accuracy of subsequent processing. Converting the original data into a standard format makes subsequent processing more convenient, reducing the complexity and development cost of the system.
[0019] Furthermore, the recognition database is constructed based on a publicly available network dataset, and an English recognition database and a Chinese recognition database are established for English and Chinese data respectively.
[0020] Beneficial effects: By establishing recognition databases for Chinese and English data respectively, the system can handle tasks in different languages, improving the scope of application of the system. The recognition database provides rich data support, which is conducive to the domain classification module to divide the monotonic and ordered tasks into different domains.
[0021] Furthermore, the machine learning algorithm divides the monotonic and ordered classification tasks into different professional domains by analyzing the feature information in the monotonic and ordered classification tasks.
[0022] Beneficial effects: Through domain division, the data relationship comparison module can more accurately understand the task background and data characteristics, thereby improving the recognition accuracy. And through domain division, the data relationship comparison module can adaptively process monotonic and ordered classification tasks in different domains, improving the flexibility and adaptability of data processing for monotonic and ordered classification tasks.
[0023] Furthermore, the feature information includes keywords, tags, numerical features, and time series features.
[0024] Beneficial effects: Keywords and tags can directly reflect the core content and characteristics of the monotonic and ordered classification tasks, helping the system to more accurately understand the task background and data meaning. Numerical features provide quantitative data support, enabling the system to conduct more in-depth analysis and recognition based on numerical relationships.
[0025] Furthermore, the precision detection module is equipped with a detection method for the statistical model, and the output result of the data relationship comparison module is deeply verified and corrected through the detection method of the statistical model.
[0026] Beneficial effects: Through in-depth verification and correction, the precision detection module can ensure the accuracy of the recognition results, reduce the misjudgment rate, and enhance the reliability of the recognition results through the precision detection module.
[0027] Furthermore, the detection methods include but are not limited to linear regression, logistic regression, and decision trees.
[0028] Beneficial effects: Linear regression is suitable for predicting the output of continuous values. In the case of relatively regular data distribution, it can find a better fitting line, thereby providing relatively accurate prediction results.
[0029] Furthermore, it also includes an outlier detection module, which is used to identify outliers in the data and then delete and replace the outliers.
[0030] Beneficial effects: By deleting or replacing outliers, the system can further improve data quality and reduce the interference of noise on the recognition results. The outlier detection module enhances the robustness of the system, enabling the system to better handle complex and changing data environments.
[0031] Further, the visualization technology includes one or more of line charts, scatter plots, and heat maps.
[0032] Beneficial effects: The visualization technology makes the recognition results more intuitive and easy to understand, facilitating user understanding and application. Through the graphical display method, the system can provide a better user experience, enabling users to obtain the required information more easily.
[0033] Further, a method for identifying multiple types of monotonic relationships in monotonic ordered classification tasks includes:
[0034] Step 1: Input the monotonic ordered classification task to be detected through the input module, and then preprocess the monotonic ordered classification task to remove data noise and redundant information.
[0035] Step 2: Through the database establishment module, use publicly available network datasets to establish different recognition databases for Chinese and English.
[0036] Step 3: Through the domain classification module, use machine learning algorithms to analyze the characteristic information of the monotonic ordered task, and then divide the domain of the monotonic ordered task.
[0037] Step 4: Through the data relationship comparison module, combine the results of domain classification and the internal data relationship of the monotonic ordered classification task for in-depth comparison and analysis.
[0038] Step 5: Through the accuracy detection module, adopt a detection method based on a statistical model to deeply verify and correct the recognition results of the data relationship comparison module.
[0039] Step 6: Through the result output module, use visualization technology to output the monotonic relationships in the finally recognized monotonic ordered classification task.
[0040] Beneficial effects: The preprocessing process effectively removes data noise and redundant information, improves data quality, and provides a more reliable basis for subsequent analysis and recognition. At the same time, the preprocessing process also simplifies the data format, making subsequent processing more efficient and reducing the computational cost. By establishing different recognition databases for Chinese and English, the system can handle tasks in multiple languages, enhancing the flexibility of the system. By using publicly available network datasets, the system can continuously learn and update recognition rules, improving its adaptability to new tasks. Using machine learning algorithms to analyze the feature information of monotonic ordered tasks and perform domain division helps the system understand the task background and data features more accurately, thereby improving the recognition accuracy. Combining the domain classification results and the internal data relationships of the tasks for in-depth comparison and analysis can comprehensively identify various monotonic relationships existing in the tasks, improving the comprehensiveness of recognition. Using a detection method based on a statistical model to deeply verify and correct the recognition results of the data relationship comparison module ensures the accuracy and reliability of the recognition results. Using visualization technology to output the monotonic relationships in the finally recognized monotonic ordered classification tasks makes the results more intuitive and understandable.
[0041] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. Brief Description of the Drawings
[0042] Figure 1 It is a structural block diagram of a system for identifying multiple types of monotonic relationships in a monotonic ordered classification task in an embodiment of the present invention.
[0043] Figure 2 It is a full-process structural block diagram of a system for identifying multiple types of monotonic relationships in a monotonic ordered classification task in an embodiment of the present invention.
[0044] Figure 3 It is a method step diagram for identifying multiple types of monotonic relationships in a monotonic ordered classification task in an embodiment of the present invention. Detailed Description of the Embodiments
[0045] The following is a further detailed description through specific embodiments:
[0046] Embodiment 1:
[0047] As shown in the Figure 1 - Figure 2 accompanying drawings: A system for identifying multiple types of monotonic relationships in a monotonic ordered classification task includes an input module, a database establishment module, a domain classification module, a data relationship comparison module, an accuracy detection module, and a result output module.
[0048] Among them, the input module is used to input the monotonic and ordered classification task to be detected and preprocess the data in the monotonic and ordered classification task to be detected; the preprocessing includes removing the noise and redundant information in the data and performing a standardized conversion on the data format of the monotonic and ordered classification task.
[0049] The database establishment module is used to establish an identification database for identifying different monotonic and ordered classification tasks according to the monotonic and ordered classification tasks input by the input module; the identification database is constructed based on the network public data set, and an English identification database and a Chinese identification database are respectively established for English and Chinese data.
[0050] The domain classification module is used to combine the data in the monotonic and ordered classification task with the identification database according to the machine learning algorithm and perform domain division; the machine learning algorithm divides the monotonic and ordered classification task into different professional domains by analyzing the feature information in the monotonic and ordered classification task. The feature information includes keywords, labels, numerical features, and time series features.
[0051] The data relationship comparison module is used to deeply compare and analyze by combining the domain classification results obtained in the domain classification module and the data relationship inside the monotonic and ordered classification task to identify various monotonic relationships existing in the monotonic and ordered classification task.
[0052] The accuracy detection module; further verifies and adjusts the various monotonic relationships identified by the data relationship comparison module; the accuracy detection module is provided with a detection method of a statistical model, and deeply verifies and corrects the output result of the data relationship comparison module through the detection method of the statistical model. The detection methods include, but are not limited to, linear regression, logistic regression, and decision tree.
[0053] The result output module outputs the monotonic relationships in the finally identified monotonic and ordered classification task by using visualization technology. The visualization technology includes one or more of line charts, scatter plots, and heat maps.
[0054] The functions of each module will be explained in detail as follows:
[0055] Input module, the user inputs the monotonic and ordered classification task to be detected through the input module.
[0056] Specifically, the staff can input the task that needs to be monotonically and orderly classified into the input module. In this embodiment, a monotonic and ordered classification task about the rise and fall of housing prices is taken as an example.
[0057] Database establishment module, the database establishment module will establish a variety of identification databases.
[0058] Specifically, the database establishment module is constructed based on publicly available network datasets. In the monotonic and ordered classification of housing price increases and decreases, an identification database is established based on housing price datasets and stock price datasets, and an English identification database and a Chinese identification database are respectively set up for English and Chinese data.
[0059] The domain classification module uses machine learning algorithms to combine the data in the monotonic and ordered classification tasks with the identification database for domain division.
[0060] Specifically, after the database establishment module completes the establishment of the identification database, the machine learning algorithm analyzes the feature information in the monotonic and ordered classification tasks: keywords such as "housing price", "increase and decrease", and "area", labels such as "increase" and "decrease", numerical features such as the numerical values of "price" and "area", and time series features such as "time" and "month". Combining keywords, labels, numerical features, and time series, the monotonic and ordered classification tasks are divided into the real estate domain.
[0061] The data relationship comparison module combines the domain classification results in the domain classification module and the internal data relationship in the monotonic and ordered classification tasks for in-depth comparison and analysis.
[0062] Specifically, the data relationship comparison module identifies various monotonic relationships existing in the task, such as positive correlation, negative correlation, linear relationship, and non-linear relationship, by comparing the data in the monotonic and ordered classification tasks with the data in the identification database.
[0063] For example, in this embodiment, the data relationship comparison module identifies a positive correlation between housing price and time, that is, as time goes by, the housing price shows an upward trend.
[0064] The accuracy detection module is equipped with detection methods of statistical models, such as linear regression, logistic regression, and decision tree, to further verify and adjust the identification results of the data relationship comparison module.
[0065] Specifically, the accuracy detection module deeply verifies and corrects the identification results through the detection methods of the statistical models set, so as to improve the accuracy and reliability of the identification.
[0066] In this embodiment, the accuracy detection module uses a linear regression model to verify the relationship between housing price and time, and finds that the linear relationship is significant and the model fitting degree is relatively high. Therefore, it is confirmed that there is a positive correlation between housing price and time. Thus, the accuracy of the identification results identified by the data relationship comparison module is improved.
[0067] The result output module uses visualization technologies, such as line charts, scatter plots, and heat maps, to output the monotonic relationships in the finally identified monotonic and ordered classification tasks.
[0068] In this embodiment, the result output module uses a line chart to display the relationship between housing prices and time, clearly presenting the upward trend of housing prices over time.
[0069] Embodiment 2:
[0070] As Figure 2 shown, the difference from the above embodiment is that it further includes an outlier detection module, which is used to identify outliers in the data and then delete and replace the outliers.
[0071] Specifically, during the process of data information entry, if there are data entry errors, it will cause outliers to appear in the identified data. There are various methods for outlier detection. By deleting or replacing the identified outliers, the accuracy and reliability of subsequent analysis can be improved.
[0072] Embodiment 3:
[0073] As Figure 3 shown, a method for identifying multiple types of monotonic relationships in a monotonic ordered classification task includes:
[0074] Step 1: Input the monotonic ordered classification task to be detected through the input module, and then preprocess the monotonic ordered classification task to remove data noise and redundant information.
[0075] Step 2: Open the network public dataset and establish different recognition databases for Chinese and English.
[0076] Step 3: Use machine learning algorithms to analyze the feature information of the monotonic ordered task, and then divide the domain of the monotonic ordered task.
[0077] Step 4: Combine the results of domain classification and the internal data relationship of the monotonic ordered classification task for in-depth comparison and analysis.
[0078] Step 5: Adopt a detection method based on a statistical model to deeply verify and correct the recognition results of the data relationship comparison module.
[0079] Step 6: Use visualization technology to output the finally identified monotonic relationship in the monotonic ordered classification task.
[0080] The specific implementation process is as follows: First, the staff can input the monotonic ordered classification task to be detected through the input module, and then the input module will perform data preprocessing on the data of the monotonic ordered classification task. Data preprocessing is a key step to ensure data quality, including removing data noise and redundant information to make the data consistent and readable.
[0081] By publicly available datasets on the network, recognition databases in Chinese and English are respectively established, which is conducive to the subsequent domain classification module to classify the domains of monotonous and ordered tasks.
[0082] Then the domain classification module can extract the feature information in the monotonous and ordered task data, and classify the monotonous and ordered tasks according to the data in different recognition databases.
[0083] Subsequently, there is the data relationship comparison module. By deeply analyzing the data relationship within the monotonous and ordered tasks through the data relationship comparison module, the monotonic relationship in the recognized monotonous and ordered tasks is judged.
[0084] Finally, after the monotonic relationship in the monotonous and ordered tasks is recognized, the monotonic relationship can be output using visualization technology for the staff to refer to.
[0085] Obviously, the above embodiments are merely examples given for clear illustration and are not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all the implementation manners here. And the obvious changes or modifications derived therefrom are still within the protection scope of the present invention.
Claims
1. A system for identifying multi-type monotonic relations in monotonic ordered classification tasks, characterized in that: include: An input module, used for inputting the monotone ordered classification task to be detected and preprocessing the data in the monotone ordered classification task to be detected; A database establishment module is used to establish a recognition database for identifying different monotonous ordered classification tasks according to the monotonous ordered classification tasks input by the input module; The domain classification module is used to combine the data in the monotonous and ordered classification task with the recognition database and perform domain division according to the machine learning algorithm; The data relationship comparison module is used to combine the domain classification results obtained in the domain classification module with the data relationship within the monotonic ordered classification task, conduct in-depth comparison and analysis, and identify various monotonic relationships existing in the monotonic ordered classification task; Precision detection module: further verify and adjust the various monotonic relationships identified by the data relationship comparison module; The result output module uses visualization technology to output the monotonic relationship in the finally identified monotonic ordered classification task.
2. The system for identifying multiple types of monotonic relations in monotonic ordered classification tasks according to claim 1, characterized in that: Preprocessing includes removing noise and redundant information from the data and standardizing the data format of the monotonous and ordered classification task.
3. The system for identifying multiple types of monotonic relations in monotonic ordered classification tasks according to claim 2, characterized in that: The recognition database is constructed based on the public data set on the Internet, and an English recognition database and a Chinese recognition database are set up for English and Chinese data respectively.
4. The system for identifying multiple types of monotonic relations in monotonic ordered classification tasks according to claim 3, characterized in that: The machine learning algorithm divides the monotonous ordered classification tasks into different professional fields by analyzing the feature information in the monotonous ordered classification tasks.
5. The system for identifying multiple types of monotonic relations in monotonic ordered classification tasks according to claim 4, characterized in that: Feature information includes keywords, labels, numerical features, and time series features.
6. The system for identifying multiple types of monotonic relations in monotonic ordered classification tasks according to claim 5, characterized in that: The accuracy detection module is equipped with a statistical model detection method, through which the output results of the data relationship comparison module are deeply verified and corrected.
7. The system for identifying multiple types of monotonic relations in monotonic ordered classification tasks according to claim 6, characterized in that: Detection methods include, but are not limited to, linear regression, logistic regression, and decision trees.
8. The system for identifying multiple types of monotonic relations in monotonic ordered classification tasks according to claim 7, characterized in that: It also includes an outlier detection module; The outlier detection module is used to identify outliers in the data and then delete and replace them.
9. The system for identifying multiple types of monotonic relations in monotonic ordered classification tasks according to claim 8, characterized in that: Visualization techniques include one or more of line graphs, scatter graphs, and heat maps.
10. A method for identifying multi-type monotonic relations in a monotonic ordered classification task, characterized in that: include: Step 1: Input the monotonous ordered classification task to be detected through the input module, and then preprocess the monotonous ordered classification task to remove data noise and redundant information; Step 2: Publicize the dataset through the Internet and establish different recognition databases for Chinese and English; Step 3: Use machine learning algorithms to analyze the characteristic information of monotonous and orderly tasks, and then divide the domain of monotonous and orderly tasks; Step 4: Combine the results of domain classification with the data relationship within the monotonous ordered classification task to conduct in-depth comparative analysis; Step 5: Use a detection method based on a statistical model to deeply verify and correct the recognition results of the data relationship comparison module; Step 6: Use visualization techniques to output the monotonic relationship in the final identified monotonic ordered classification task.