Obviousness

Obviousness Analysis for US Patent 7,818,176 Under 35 U.S.C. § 103

This analysis identifies combinations of prior art references that would render the claims of US Patent 7,818,176 obvious to a person having ordinary skill in the art (PHOSITA) at the time of the invention (i.e., prior to the February 6, 2007 filing date), along with the motivation for such combinations. The independent claims of US7818176 are Claims 1, 14, 27, and 40.

Combinations for Claims 1 and 27 (Core Method and System)

Claims 1 and 27 broadly describe a system and method for receiving a natural language utterance, recognizing words and phrases (including mapping phonemes to syllables and generating a preliminary interpretation), interpreting them at a conversational language processor to establish context, and then selecting and presenting an advertisement based on that context.

Combination: US Patent 7,398,209 B2 (Kennewick et al.) in combination with US Patent Application Publication 2005/0288925 A1 (Van et al.).

• US 7,398,209 B2 (Kennewick et al.): This patent, explicitly cited as foundational in the '176 patent, comprehensively teaches a system and method for responding to natural language speech utterances. It details receiving utterances, recognizing words and phrases, and interpreting them using a conversational language processor that establishes context and manages dialogue to fulfill user requests. It also describes the underlying speech recognition mechanics, including dynamically updated dictionaries and phrase tables, which are pertinent to the recognition steps in claims 1 and 27. However, the '209 patent focuses on executing commands or retrieving information as a direct response to a user's request, not on selecting and presenting advertisements.
• US 2005/0288925 A1 (Van et al.): This prior art specifically addresses providing targeted advertisements based on audio input. It discloses capturing user speech, analyzing it to identify keywords or topics, and then selecting and delivering relevant advertisements. While its approach to speech analysis might be less sophisticated than the "conversational language processor" of the '209 patent (potentially relying more on keyword spotting), it clearly teaches the concept of associating speech input with advertisement selection and presentation.

Motivation for Combination:
A PHOSITA in the field of voice user interfaces and advertising, seeking to improve the relevance and effectiveness of voice-based advertising, would have been motivated to combine the sophisticated natural language understanding and contextual awareness of the system described in US 7,398,209 with the known concept of speech-based advertisement targeting taught by US 2005/0288925.

The motivation is to leverage the robust conversational capabilities of US 7,398,209 (which allows for a deeper interpretation of user intent and context beyond simple keywords) to enhance the targeting and relevance of advertisements as broadly disclosed in US 2005/0288925. A system that can understand the context of a user's request (as described in '209) rather than just isolated keywords (as might be inferred from '925) would predictably lead to more accurate and valuable advertisement selections. This would be a predictable result of combining known elements to improve a known function (ad targeting) by applying a more advanced form of analysis (conversational NLP).

Therefore, a PHOSITA would find it obvious to integrate an advertisement selection and presentation module (as taught by Van et al.) into the conversational language processing system of Kennewick et al. to provide contextually relevant advertisements in response to natural language utterances.

Combinations for Claims 14 and 40 (Adaptive Misrecognition Method and System)

Claims 14 and 40 extend the core method and system by adding an "adaptive misrecognition" capability. This involves detecting when the conversational language processor incorrectly interpreted words or phrases based on a predetermined event (e.g., subsequent input proximate in time, stopping/overriding a request, or repeating an utterance) and subsequently reinterpreting the words or phrases.

Combination: The combination described for Claims 1 and 27 (US 7,398,209 B2 + US 2005/0288925 A1) in further combination with US Patent 7,620,549 B2 (Kennewick et al.).

• US 7,620,549 B2 (Kennewick et al.): This patent, also from the same original assignee and inventors as the '176 patent, specifically teaches a "System and method of supporting adaptive misrecognition in conversational speech". It details the detection of predetermined events (such as subsequent inputs or repetitions) that indicate a prior misinterpretation, and the subsequent reinterpretation of the user's utterance. This patent directly discloses the adaptive misrecognition features recited in claims 14 and 40.

Motivation for Combination:
Given the established combination for claims 1 and 27 (i.e., a conversational voice system that also serves advertisements), a PHOSITA would have a clear motivation to incorporate known solutions for improving the accuracy and robustness of such a system. Speech recognition and natural language processing systems are inherently prone to errors. US 7,620,549 B2 directly addresses this known problem by providing a mechanism for adaptive misrecognition.

It would be obvious to a PHOSITA to integrate the adaptive misrecognition techniques from US 7,620,549 B2 into the voice-based advertising system formed by combining US 7,398,209 B2 and US 2005/0288925 A1. The motivation would be to enhance the user experience and the accuracy of the system by correcting misinterpretations, thereby ensuring that both the user's requested action is correctly performed and that the selected advertisements are indeed relevant to the user's actual intent. This is the application of a known technique (adaptive misrecognition) to a known system (a voice-based advertising system) to achieve a predictable improvement (increased accuracy of interpretation and ad selection).