Ensemble learning-based prompt enhancement for extracting public relations

Abstract

A system and method for extracting relationships from text data is provided, including collecting labeled text data from diverse sources containing sentences annotated with detailed grammatical structures, including digital archives and online repositories (502). Initial relationship data is generated from grammatical structures by applying advanced parsing and machine learning using advanced rule-based algorithms (506). A training set is generated to increase the diversity and complexity of the relationship dataset by applying data augmentation techniques to the initial relationship data (508). A neural network model is trained using an array of semantically equivalent but syntactically diverse prompt templates designed to test and improve the model's linguistic capabilities (510). The final relationship extraction output is determined by implementing a voting-based decision system that integrates statistical analysis and utilizes a weighted voting mechanism to optimize extraction accuracy and reliability (512).

Claims

1. A method for extracting relationships from text data, implemented on a computer, Collecting labeled text data from diverse sources, including digital archives and online repositories, that contain texts annotated with detailed grammatical structures (502), By applying advanced syntactic analysis and machine learning techniques using advanced rule-based algorithms, the system systematically generates initial relational data from the grammatical structure (506), Applying one of several data augmentation techniques to the initial relational data generates a training set to increase the diversity and complexity of the relational dataset (508), Training a neural network model using a comprehensive array of semantically equivalent but syntactically diverse prompt templates designed to test and improve the model's linguistic capabilities (510), A method comprising determining the final relational extraction output by integrating statistical analysis and implementing a voting-based decision system that utilizes a weighted voting mechanism to optimize extraction accuracy and reliability (512).

2. In the method according to claim 1, The method for collecting the aforementioned labeled text data includes preprocessing operations to remove noise and transform the data to standardize the format across various text sources.

3. In the method according to claim 1, Furthermore, the method includes integrating the extracted relationships into an enterprise data management system to provide automated data retrieval and enhance functionality, including search accuracy and content recommendations within enterprise databases and enterprise resource planning (ERP) systems, thereby improving operational efficiency and data utilization based on the insights gained from the derived relationships.

4. In the method according to claim 1, The method for generating the training set includes using a machine learning model to automatically determine the optimal combination of synthetic and adversarial examples, thereby maximizing the model's robustness to unknown data.

5. In the method according to claim 1, Training the aforementioned neural network model further includes repeating the training cycle multiple times, followed by an evaluation phase after each training cycle in which the model is adjusted based on performance metrics such as accuracy and loss reduction.

6. In the method according to claim 1, A method for determining the final relation extraction output includes applying an ensemble learning technique that combines multiple model predictions to reduce variance and improve decision accuracy.

7. In the method according to claim 1, A method for improving data accessibility and searchability in a relatively large-scale information system, further comprising integrating and utilizing the extracted relationships, and automatically tagging and classifying new input text data.

8. A system for extracting relationships from text data, Processor device (104) and The system has a memory (110) for storing instructions, and when an instruction is executed by the processor device (104), the system receives the instruction. The system collects labeled text data from diverse sources, including digital archives and online repositories, that contain texts annotated with detailed grammatical structures (502), By applying advanced syntactic analysis and machine learning techniques using advanced rule-based algorithms, initial relational data is systematically generated from the grammatical structure (506), By applying one of several data augmentation techniques to the initial relational data, a training set is generated to increase the diversity and complexity of the relational dataset (508). A neural network model is trained using a comprehensive array of semantically equivalent but syntactically diverse prompt templates designed to test and improve the model's linguistic capabilities (510). A system that determines the final relation extraction output by implementing a voting-based decision system that integrates statistical analysis and utilizes a weighted voting mechanism to optimize extraction accuracy and reliability (512).

9. In the system described in claim 8, The system for collecting the aforementioned labeled text data includes preprocessing operations that remove noise and transform the data to standardize the format across various text sources.

10. In the system described in claim 8, The aforementioned instructions further integrate the extracted relationships into the enterprise data management system to provide automated data retrieval and enhance functionality, including search accuracy and content recommendations within enterprise databases and enterprise resource planning (ERP) systems, thereby improving operational efficiency and data utilization based on insights from the derived relationships.

11. In the system described in claim 8, The generation of the aforementioned training set is a system that uses a machine learning model to automatically determine the optimal combination of synthetic and adversarial examples, thereby maximizing the model's robustness to unknown data.

12. In the system described in claim 8, The system involves training the aforementioned neural network model by repeatedly executing multiple training cycles, followed by an evaluation phase after each training cycle in which the model is adjusted based on performance metrics such as accuracy and loss reduction.

13. In the system described in claim 8, The system for determining the final relation extraction output includes applying ensemble learning techniques that combine multiple model predictions to reduce variance and improve decision accuracy.

14. In the system described in claim 8, The aforementioned instructions further enable the system to integrate and utilize the extracted relationships in a relatively large-scale information system, automatically tagging and classifying new input text data to improve data accessibility and searchability.

15. A computer program product for extracting relationships from text data, the computer program product includes a computer-readable storage medium in which program instructions are embodied, and the program instructions are transmitted by a hardware processor. Collecting labeled text data from diverse sources, including digital archives and online repositories, that contain texts annotated with detailed grammatical structures (502), By applying advanced syntactic analysis and machine learning techniques using advanced rule-based algorithms, the system systematically generates initial relational data from the grammatical structure (506), Applying one of several data augmentation techniques to the initial relational data generates a training set to increase the diversity and complexity of the relational dataset (508), Training a neural network model using a comprehensive array of semantically equivalent but syntactically diverse prompt templates designed to test and improve the model's linguistic capabilities (510), A computer program product capable of determining the final relational extraction output (512) by implementing a voting-based decision system that integrates statistical analysis and utilizes a weighted voting mechanism to optimize extraction accuracy and reliability.

16. In the computer program product described in claim 15, The collection of the aforementioned labeled text data is a computer program product that includes preprocessing operations to remove noise and transform the data to standardize the format across various text sources.

17. In the computer program product described in claim 15, Furthermore, the computer program product includes instructions to integrate the extracted relationships into the enterprise data management system, provide automated data retrieval, enhance functionality including search accuracy and content recommendations within the enterprise database and enterprise resource planning (ERP) system, and improve operational efficiency and data utilization based on the derived relationship insights.

18. In the computer program product described in claim 15, The generation of the aforementioned training set is a computer program product that uses a machine learning model to automatically determine the optimal combination of synthetic and adversarial examples, thereby maximizing the model's robustness to unknown data.

19. In the computer program product described in claim 15, The system for training the aforementioned neural network model includes repeatedly executing training cycles, and after each training cycle, adjusting the model based on performance metrics such as accuracy and loss reduction.

20. In the computer program product described in claim 15, A computer program product that further includes instructions for integrating and utilizing the extracted relationships in a relatively large-scale information system, automatically tagging and classifying new input text data, and improving data accessibility and searchability.

Images