AI-based intelligent error correction systems and methods based on contextual semantics

Abstract

A system and method for AI-based intelligent errata correction based on contextual semantics includes an errata module comprising multiple errata lists; a speech-to-text module that receives an input speech message and converts the speech message into multiple text messages; and an aggregation module that connects the speech-to-text module and the errata module. The aggregation module receives and integrates the text messages and errata lists into a prompting process as input, uses an artificial intelligence model to correct the text messages, and outputs a corrected result.

Claims

1. A system for intelligent error correction based on AI and semantic context, comprising: An errata module, including multiple errata lists; a speech-to-text module connected to the errata module, the speech-to-text module receiving an input speech message and converting the speech message into multiple text messages, wherein the speech-to-text module includes multiple speech-to-text models to improve the accuracy of the converted text messages; and an aggregation module connected to the speech-to-text module and the errata module, the aggregation module receiving the text messages and the errata lists, including the text messages and the errata lists in a prompting process, the prompting process serving as input to an artificial intelligence model, using the artificial intelligence model to correct the text messages, and outputting a corrected result.

2. The AI-based intelligent errata system as described in claim 1, wherein the errata table includes a dynamically generated errata table and a static errata table.

3. The AI-based intelligent errata system described in claim 2, wherein the dynamically generated errata table is an errata rule dynamically generated based on the contextual content of the voice message.

4. The AI-based intelligent errata system described in claim 2, wherein the static errata table is existing data, including dictionaries or organizational rosters.

5. The AI-based intelligent error correction system as described in claim 1, wherein the speech-to-text module includes at least one first speech-to-text model with prompting engineering capabilities and at least one second speech-to-text model without prompting engineering capabilities.

6. The AI-based intelligent error correction system as described in claim 1, wherein the aggregation module selects the best result from the text messages using a majority vote or a reliability assessment method for the correction process.

7. The AI-based intelligent error correction system as described in claim 5, wherein the artificial intelligence model takes the contextual content of the voice message and the error correction table as input to correct the text message to generate the correction result.

8. The AI-based intelligent error correction system as described in claim 1 further includes an input / output module connected to the speech-to-text module and the aggregation module, receiving the voice message and transmitting it to the speech-to-text module, and outputting the correction result.

9. A method for intelligent error correction based on AI and semantic context, comprising the following steps: receiving an input voice message, and using a speech-to-text module to convert the voice message into multiple text messages, wherein, The speech-to-text module includes multiple speech-to-text models to improve the accuracy of the converted text messages; it uses a collection module to receive multiple errata sheets and the text messages, includes the text messages and errata sheets in a prompting process, the prompting process serves as input to an artificial intelligence model, the artificial intelligence model performs correction processing on the text messages, and outputs a correction result.

10. The AI-based intelligent errata method as described in claim 9, wherein the errata tables include a dynamically generated errata table and a static errata table.

11. The AI-based intelligent errata method for semantic contextualization as described in claim 10, wherein the dynamically generated errata table is an errata rule dynamically generated based on the contextual content of the voice message.

12. The AI-based intelligent errata method as described in claim 10, wherein the static errata table is existing data, including dictionaries or organizational rosters.

13. The AI-based intelligent error correction method according to claim 11, wherein the speech-to-text module includes at least one first speech-to-text model with prompting engineering capability and at least one second speech-to-text model without prompting engineering capability.

14. The AI-based intelligent error correction method as described in claim 9, wherein the aggregation module selects the best result from the text messages using a majority vote or a reliability assessment method for the correction process.

15. The AI-based intelligent errata correction method as described in claim 9, wherein the AI ​​model takes the contextual content of the voice message and the errata table as input to correct the text message to generate the correction result.

Images