Prior art

Prior Art Analysis for US Patent 10,510,341

This analysis covers the prior art references cited by the USPTO examiner during the prosecution of US patent 10,510,341. The focus is on whether any single reference anticipates the independent claims (1, 12, and 19) under 35 U.S.C. § 102, which requires that every element of the claim be present in that single reference.

The core invention of US 10,510,341 revolves around a method for a conversational interface that:

• Builds multiple hypotheses about a user's intent from an utterance.
• Uses both short-term knowledge (from the current conversational session) and long-term knowledge (user-centric history, preferences) to inform these hypotheses.
• Assigns a degree of certainty (e.g., "sure," "not sure") to each hypothesis.
• Generates an adaptive response based on the certainty level, which may involve asking clarifying questions.
• Frames the response to guide the user toward an utterance that is easier for the system to recognize.

---

Analysis of Cited References

Based on the patent's file history, the following patents were cited as relevant prior art.

1. US Patent 7,640,160 B2 (Yankelovich et al.)

• Full Citation: US 7,640,160 B2, "Systems and methods for responding to natural language speech utterance"
• Publication Date: December 29, 2009
• Filing Date: August 5, 2005
• Brief Description: This patent, assigned to VoiceSignal Technologies, describes a system for processing natural language speech. It notably discloses a competitive model where multiple "domain agents" (e.g., a music agent, a navigation agent) compete to determine the most appropriate context for a user's utterance. The winning agent is then responsible for handling the request. The '160 patent is explicitly mentioned in the specification of '341 (Column 9).
• Potential Anticipation:
  • Does it anticipate Claim 1? Unlikely. While the '160 patent describes a sophisticated method for determining context by having domain agents compete, it does not appear to explicitly teach the combination of all key elements of claim 1. Specifically, it focuses heavily on context determination through competing agents but does not clearly describe the process of:
    1. Using both short-term (session) and long-term (user-centric) shared knowledge to generate multiple hypotheses.
    2. Assigning distinct, ranked degrees of certainty to these hypotheses (e.g., the "sure" vs. "not sure" paradigm).
    3. Systematically building an adaptive response that is explicitly framed to influence the user's next utterance for easier recognition.
  • The '160 patent's method is a foundational piece of the context determination described in the '341 patent, but it does not appear to describe the complete cooperative conversational model claimed in '341.

2. US Patent 8,073,681 B2 (Baldwin et al.)

• Full Citation: US 8,073,681 B2, "System and method for a cooperative conversational voice user interface"
• Publication Date: December 6, 2011
• Filing Date: October 16, 2006
• Brief Description: This patent shares the same title and inventors as the '341 patent. The '341 patent is a continuation in a long chain of applications that ultimately claims priority back to the application that issued as this '681 patent.
• Potential Anticipation:
  • This patent is not prior art under 35 U.S.C. § 102. A patent cannot be anticipated by another patent from its own family with a later filing date but an earlier priority date. The '681 patent represents an earlier version of the same core invention. The claims in '341 may be different in scope or wording, but they rely on the same original disclosure.

3. US Patent 7,634,409 B2 (Baldwin et al.)

• Full Citation: US 7,634,409 B2, "Dynamic speech sharpening"
• Publication Date: December 15, 2009
• Filing Date: August 31, 2006
• Brief Description: This patent focuses on the speech recognition (ASR) part of the process. It describes a method of using phonetic dictation to recognize a stream of phonemes, which can improve the accuracy of the initial speech-to-text conversion. This is mentioned in the '341 patent specification (Column 7) as a potential technique for the ASR engine.
• Potential Anticipation:
  • Does it anticipate Claim 1? No. The '409 patent is directed at the lower-level task of speech recognition—turning audio into text. It does not disclose the higher-level conversational and contextual processing that is the subject of claim 1, such as building hypotheses, using long/short-term knowledge, assessing certainty, and generating adaptive responses.

4. US Patent Application Publication 2005/0288924 A1 (Bennett)

• Full Citation: US 2005/0288924 A1, "Spoken language interface for enterprise applications"
• Publication Date: December 29, 2005
• Filing Date: June 24, 2004
• Brief Description: This application describes a spoken language interface for accessing and manipulating data within large enterprise applications. It focuses on resolving ambiguities in user speech by considering the context of the specific enterprise application and the data it contains.
• Potential Anticipation:
  • Does it anticipate Claim 1? Unlikely. Bennett's system uses context to disambiguate, but the context is primarily derived from the application's data structure and schema. It does not appear to describe the use of both fluid, session-based short-term knowledge and evolving, user-specific long-term knowledge in the manner claimed by '341. Furthermore, it does not explicitly disclose a system of generating multiple hypotheses and ranking them with distinct degrees of certainty to then formulate a response designed to "parrot" or guide the user.

Conclusion

Based on this analysis, none of the individual prior art references cited by the examiner appear to fully anticipate the independent claims of US patent 10,510,341 under 35 U.S.C. § 102. While references like US 7,640,160 teach crucial components such as context determination, they do not disclose the complete, integrated method of the cooperative conversational model claimed in '341. The key inventive step of the '341 patent appears to be the specific combination of using short- and long-term knowledge, generating hypotheses with varying certainty levels, and using that certainty to build an adaptive, guiding response. An invalidity challenge would more likely need to rely on an obviousness argument under 35 U.S.C. § 103, which would require combining elements from multiple prior art references.