17526806. HYBRID TRANSFORMER-BASED DIALOG PROCESSOR simplified abstract (MICROSOFT TECHNOLOGY LICENSING, LLC)

From WikiPatents
Jump to navigation Jump to search

HYBRID TRANSFORMER-BASED DIALOG PROCESSOR

Organization Name

MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor(s)

Jinchao Li of Redmond WA (US)

Lars H. Liden of Seattle WA (US)

Baolin Peng of Bellevue WA (US)

Thomas Park of Seattle WA (US)

Swadheen Kumar Shukla of Kirkland WA (US)

Jianfeng Gao of Woodinville WA (US)

HYBRID TRANSFORMER-BASED DIALOG PROCESSOR - A simplified explanation of the abstract

This abstract first appeared for US patent application 17526806 titled 'HYBRID TRANSFORMER-BASED DIALOG PROCESSOR

Simplified Explanation

The patent application describes a system and method for determining a response to a query in a dialog. Here are the key points:

  • An entity extractor is used to extract rules and conditions associated with the query and determine a specific task.
  • A transformer-based dialog embedding is generated by pre-training a transformer using dialog corpora that include multiple tasks.
  • A task-specific classifier generates a set of candidate responses based on the rules and conditions associated with the task.
  • The transformer-based dialog embedding also generates a set of candidate responses to the query.
  • The classifier can accommodate changes made to a task by an interactive dialog editor, acting as machine teaching.
  • A response generator uses an optimization function to generate a response based on the candidate responses from both the classifier and the transformer-based dialog embedding.

Potential applications of this technology:

  • Chatbots and virtual assistants that can provide accurate and contextually relevant responses in a dialog.
  • Customer service systems that can handle a wide range of queries and tasks.
  • Interactive tutoring systems that can provide personalized feedback and guidance.

Problems solved by this technology:

  • The system can handle complex queries and tasks by combining rule-based classification with a data-driven, generative model.
  • It can adapt to changes in rules and conditions associated with a task, allowing for easy updates and improvements.
  • The response generator optimizes the selection of the best response from a set of candidate responses, ensuring high-quality output.

Benefits of this technology:

  • Improved accuracy and relevance of responses in a dialog, leading to better user experiences.
  • Flexibility to accommodate changes and updates in rules and conditions associated with a task.
  • Efficient and effective generation of responses by leveraging both rule-based classification and data-driven models.


Original Abstract Submitted

Systems and methods are provided for determining a response to a query in a dialog. An entity extractor extracts rules and conditions associated with the query and determines a particular task. The disclosed technology generates a transformer-based dialog embedding by pre-training a transformer using dialog corpora including a plurality of tasks. A task-specific classifier generates a first set of candidate responses based on rules and conditions associated with the task. The transformer-based dialog embedding generates a second set of candidate responses to the query. The classifier accommodates changes made to a task by an interactive dialog editor as machine teaching. A response generator generates a response based on the first and second sets of candidate responses using an optimization function. The disclosed technology leverages both a data-driven, generative model (a transformer) based on dialog corpora and a user-driven, task-specific rule-based classifier that accommodating updates in rules and conditions associated with a particular task.

Retrieved from "http://wikipatents.org/index.php?title=17526806._HYBRID_TRANSFORMER-BASED_DIALOG_PROCESSOR_simplified_abstract_(MICROSOFT_TECHNOLOGY_LICENSING,_LLC)&oldid=24209"