US Patent Application 18334065. GROUNDED TEXT GENERATION simplified abstract
Contents
GROUNDED TEXT GENERATION
Organization Name
Microsoft Technology Licensing, LLC
Inventor(s)
Michel Galley of Seattle WA (US)
Christopher Brian Quirk of Seattle WA (US)
William Brennan Dolan of Kirkland WA (US)
GROUNDED TEXT GENERATION - A simplified explanation of the abstract
- This abstract for appeared for US patent application number 18334065 Titled 'GROUNDED TEXT GENERATION'
Simplified Explanation
This abstract describes a framework that allows for the generation of computer-generated text in a controlled and grounded manner. The framework consists of three components: a machine learning model, a grounding interface, and a control interface.
The machine learning model is trained to generate text based on given input text. It takes the input text as a starting point and produces computer-generated text as an output.
The grounding interface is a tool that the machine learning model can use to access a grounding source. This grounding source contains information related to the input text. By utilizing this interface, the model can incorporate relevant information from the grounding source into the computer-generated text.
The control interface is another tool that the machine learning model can use. It is designed to recognize a control signal, which can be used to guide the generation process. The model can adjust its output based on this control signal, allowing for more focused and specific text generation.
Overall, this framework enables the machine learning model to generate text that is both grounded in relevant information and controlled by specific signals.
Original Abstract Submitted
A controllable grounded response generation framework includes a machine learning model, a grounding interface, and a control interface. The machine learning model is trained to output computer-generated text based on input text. The grounding interface is useable by the machine learning model to access a grounding source including information related to the input text. The control interface is useable by the machine learning model to recognize a control signal. The machine learning model is configured to include information from the grounding source in the computer-generated text and focus the computer-generated text based on the control signal.
