17453840. ENHANCING NATURAL LANGUAGE PROCESSING ACCURACY IN COMPUTER SYSTEMS simplified abstract (International Business Machines Corporation)

From WikiPatents
Jump to navigation Jump to search

ENHANCING NATURAL LANGUAGE PROCESSING ACCURACY IN COMPUTER SYSTEMS

Organization Name

International Business Machines Corporation

Inventor(s)

Chulaka Gunasekara of New Hyde Park NY (US)

Guy Feigenblat of Givataym (IL)

Benjamin Sznajder of Jerusalem (IL)

Sachindra Joshi of Gurgaon (IN)

ENHANCING NATURAL LANGUAGE PROCESSING ACCURACY IN COMPUTER SYSTEMS - A simplified explanation of the abstract

This abstract first appeared for US patent application 17453840 titled 'ENHANCING NATURAL LANGUAGE PROCESSING ACCURACY IN COMPUTER SYSTEMS

Simplified Explanation

The patent application proposes a method to improve abstract summarization using question and answer rewards. Here are the key points:

  • The approach involves generating questions and answers for a generated summary using a question and answer generator.
  • The received answers for the generated questions are evaluated by comparing them with answers for an original summary.
  • A reward is calculated based on the similarity between the answers associated with the generated summary and the original summary.
  • The generation model is updated by applying the calculated reward to further train the summary generation model.

Potential Applications:

  • This technology can be applied in various fields where abstract summarization is important, such as news articles, research papers, and legal documents.
  • It can be used in content generation for websites, blogs, and social media platforms.
  • The method can be integrated into search engines to provide more accurate and concise summaries of search results.

Problems Solved:

  • Traditional abstract summarization methods may not always generate accurate and relevant summaries.
  • Evaluating the quality of generated summaries can be subjective and time-consuming.
  • The proposed approach addresses these issues by using question and answer rewards to improve the summarization process.

Benefits:

  • The method enhances the accuracy and relevance of generated summaries by incorporating question and answer rewards.
  • It reduces the need for manual evaluation of summaries, saving time and effort.
  • By training the summary generation model with calculated rewards, the system can continuously improve its summarization capabilities.


Original Abstract Submitted

In an approach to improve abstract summarization with question and answer rewards embodiments generate, by a question and answer generator, questions and answers corresponding to a generated summary. Further, embodiments evaluate received answers for the generated questions by analyzing received answers associated with the generated summary against answers received for an original summary, and calculate a reward based on the similarity between answers associated with generated summary and the original summary. Additionally, embodiments update the generation model by applying the calculated reward to further train the summary generation model.

Retrieved from "http://wikipatents.org/index.php?title=17453840._ENHANCING_NATURAL_LANGUAGE_PROCESSING_ACCURACY_IN_COMPUTER_SYSTEMS_simplified_abstract_(International_Business_Machines_Corporation)&oldid=22395"