Prior art

Analysis of Prior Art for U.S. Patent 6,314,420

As a senior patent analyst, I have reviewed the citations for US Patent 6,314,420 ("the '420 patent"). The patent's core invention, as defined by its independent claims 1 and 9, is a search engine system that can perform two types of searches: a standard, immediate "demand" search, and a continuous, long-term "wire" search. For at least the "wire" searches, the system combines content-based filtering with collaborative feedback from other users to improve and rank results over time.

While the claims of the '420 patent were ultimately invalidated for obviousness under 35 U.S.C. § 103 (meaning the invention would have been an obvious combination of existing technologies to a person of ordinary skill), this analysis focuses on potential anticipation under 35 U.S.C. § 102, which requires a single prior art reference to disclose every element of a claim.

The most relevant prior art references cited in the '420 patent are detailed below.

---

Key Prior Art References

1. U.S. Patent 5,867,799 A (The Parent Patent)

• Full Citation: US Patent 5,867,799 A, "Distributed, collaborative/content-based information filtering system". Filed by the same inventors, Andrew K. Lang and Donald M. Kosak, and assigned to Lycos, Inc.
• Dates: Filing Date: April 4, 1996; Publication Date: February 2, 1999.
• Brief Description: This patent, which is the parent application to the '420 patent and is incorporated by reference, discloses the foundational technology for the "wire" search. It describes in detail an information filtering system that continuously processes a data stream to find "informons" (information entities) relevant to a user. The system uses a combination of adaptive content-based filtering (analyzing the text of the informon) and collaborative filtering (using feedback from communities of users with similar interests) to identify and rank relevant information.
• Potential Anticipation of Claims: This reference discloses the majority of the elements of claims 1 and 9, including the system for continuous searching ("on a continuing basis"), the use of collaborative feedback from other users, and the combination of that feedback with content-profile data to determine relevancy. This system is functionally identical to the "wire" system described in the '420 patent. However, the '799 patent focuses exclusively on this long-term filtering model. It does not appear to explicitly disclose the dual-mode search engine that also performs an immediate "demand response" or a "query processor" that decides between the two modes.
  • Conclusion: This patent does not anticipate claims 1 or 9 under § 102 because it lacks the "demand search" element and the controlling system that selects between a "wire" and "demand" response. It is, however, the most critical reference for an obviousness argument.

2. U.S. Patent 6,014,665 A

• Full Citation: US Patent 6,014,665 A, "Personalized search and retrieval system and method". Inventor: Brad Culliss.
• Dates: Filing Date: May 2, 1997; Publication Date: January 11, 2000.
• Brief Description: This patent describes a system that personalizes search results for a specific user. It creates a user interest profile by monitoring the user's queries and the search results they select. This profile is then used to filter and re-rank future search results to better match the user's interests. The system also discloses a "proactive mode" where it can continuously search for new information related to the user's profile, similar to the "wire" concept.
• Potential Anticipation of Claims: This reference teaches a personalized, content-based search system with both an immediate search-and-rerank function (analogous to a "demand search") and a proactive, continuous search function (analogous to a "wire search"). It uses feedback, albeit from the individual user rather than a collaboration of multiple users.
  • Conclusion: This patent does not anticipate claims 1 or 9 under § 102 because it fails to disclose the element of using collaborative feedback from other users to determine the relevancy of informons. Its feedback mechanism is based on a single user's profile.

3. U.S. Patent 5,740,425 A

• Full Citation: US Patent 5,740,425 A, "Collaborative information filtering system". Inventors: Jerome R. Bellegarda, et al. Assignee: Apple Computer, Inc.
• Dates: Filing Date: September 3, 1996; Publication Date: April 14, 1998.
• Brief Description: This patent details a system that uses collaborative filtering to predict a user's interest in various information items. The system generates a user profile based on that user's ratings and compares it to the profiles of other users to find a "neighborhood" of users with similar tastes. It then recommends items that were highly rated by the users in that neighborhood. The core technology is based on Latent Semantic Indexing to manage the user profile data.
• Potential Anticipation of Claims: This reference provides a strong teaching of a "feedback system" that provides "feedback data from other users," a key element of claims 1 and 9. It explicitly describes using the preferences of a community of users to filter information for an individual.
  • Conclusion: This patent does not anticipate claims 1 or 9 under § 102. While it discloses the collaborative filtering element in detail, it does not describe combining it with a content-based filtering system in a search engine, nor does it teach a dual-mode system for handling both "demand" and "wire" (continuous) queries.