US 9734825

US Patent 9734825, titled "Methods and apparatus for determining a domain based on the content and context of a natural language utterance," was issued on August 15, 2017. The filing date for this patent was May 15, 2014. Robert A. Kennewick, David Locke, Michael R. Kennewick, SR., Michael R. Kennewick, JR., Richard Kennewick, and Tom Freeman are listed as the inventors. The current assignee is Dialect LLC, though it was originally assigned to Nuance Communications Inc.

Abstract:
The patent describes systems and methods for receiving user-generated natural language utterances, such as queries and/or commands, and executing them. A parser is used to determine a domain for a natural language utterance based on its content and context. The system then employs one or more domain agents, associated with particular domains, to receive and respond to these queries and/or commands.

Plain-language overview of each independent claim:

• Claim 1: This claim describes a computer-implemented method. It involves receiving a natural language utterance from a user and parsing it to determine an approximate meaning. Based on this meaning, the method determines a domain for the utterance, such as by using a real-time scoring system. If the confidence level of this domain determination is too low, the system asks the user to verify its understanding. Once the domain is sufficiently determined, the method invokes a domain agent (from a plurality of such agents, each associated with a specific domain) to process the utterance. The domain agent then generates and executes one or more machine-processable queries or commands based on the utterance, taking into account user profiles, context, and domain-specific information. Finally, a natural language response or action is generated and presented to the user.

• Claim 11: This claim outlines a system that includes a speech recognition module, a parser, and an event manager. The speech recognition module captures the user's natural language utterance. The parser determines a domain for the utterance based on its content and context, potentially using a real-time scoring system. The system can request user verification if the confidence in the domain determination is low. The event manager, in communication with user profiles, personality modules, agents, an update manager, and databases, manages the invocation of a domain agent based on the determined domain. This agent then generates and executes queries or commands and, subsequently, generates a natural language response.

• Claim 18: This claim details a non-transitory computer-readable storage medium containing instructions. When executed by a processor, these instructions cause the processor to perform steps similar to Claim 1. Specifically, the steps include receiving and parsing a natural language utterance to determine an approximate meaning and a domain (e.g., via a real-time scoring system). If confidence is low, user verification is requested. A domain agent is then invoked to generate and execute machine-processable queries or commands, considering user profiles, context, and domain-specific information. Finally, a natural language response or action is generated and presented to the user.

No results were found in the CAFC 2026 dockets specifically for US Patent 9734825. Searches for "CAFC 2026 dockets" and "Federal Circuit" yielded general information about CAFC cases and updates, but no direct mention of this specific patent.