US 5855020

I'll search for authoritative details on US 5855020, including any litigation or docket references.

Let me check for litigation/docket references and the related continuation patent.

Let me make a final check for any litigation docket specifically naming this patent.

US Patent 5,855,020 — "Web scan process"

Note on scope of verification: I located the patent's own full text (Google Patents, fetched 2026‑09‑24) plus a corroborating bibliographic record (uspto.report, USPTO patent grant data). I could not find any CAFC 2026 docket, PTAB proceeding, or district-court case specifically naming US 5,855,020. My searches returned no hits for this number in litigation databases. I therefore report the litigation question as unverified—no evidence found, not as a confirmed "no litigation."

Bibliographic data

Assignee chain of record (per Google Patents legal-events data): Infoseek Corp. → UltraSeek Corp. (2002‑11‑04) → Verity Inc. (2003‑01‑21) → Hewlett‑Packard Company (2015‑07‑06, merger) → Hewlett‑Packard Development Co., L.P. (2015‑10‑01) → Hewlett Packard Enterprise Development LP (2015‑11‑09).

Continuation: US 09/153,519, filed 1998‑09‑15, issued as US 6,269,370 B1 ("Web scan process") with the same 1996‑02‑21 priority date. Related concurrent Kirsch applications cited in the specification: US 5,751,956 (redirection of server external hyper-link references) and application Ser. No. 08/604,506 (trans-Internet purchase transactions).

Discrepancy flagged: One aggregator (Unified Patents) listed the priority date as 1996‑02‑20. The patent's own front matter and the authoritative full text state 1996‑02‑21 (filing and priority). I treat 1996‑02‑21 as correct and the 02‑20 listing as an aggregator artifact.

Abstract (as issued)

An information locator system providing for the expedient acquisition, validation, and updating of information locators in a heterogeneous network protocol environment. The system includes an information location discrimination engine coupleable to a network, a validation engine coupled to the discrimination engine to receive information locators, and a database for storing information locators as discrete searchable resource locators. The validation engine autonomously interrogates the network to validate a predetermined information locator as a corresponding resource locator that is unique among those then stored in the database; where a valid, inferred-unique locator is found, it is stored for searchable later retrieval.

Plain-language overview of the two independent claims

Claim 1 (system claim) — covers a system that autonomously builds and maintains a searchable database of Internet information, comprising:

• (a) a discrimination system connected to the Internet that receives messages including electronic mail messages and network news messages and processes them to identify embedded URLs;
• (b) a validation system connected to the Internet, receiving a given embedded URL from the discrimination system, that accesses the Internet to retrieve the Web page information associated with that URL; and
• (c) a database that searchably stores the URL in association with the retrieved Web page information.

In plain terms: mine e‑mail and Usenet/news feeds for URLs, then go fetch the actual page at each URL and index the URL together with a copy/sample of the page content.

Claim 6 (method claim) — covers a method of keeping a URL listing current so it supports searching at least part of the Internet, comprising:

• (a) associating predetermined volatility data with a given URL in the list;
• (b) associating predetermined contextual data with that URL; and
• (c) periodically validating currentness of the URL by:
  • (i) determining whether the volatility data meets predetermined validation criteria (i.e., is it time to re-check?);
  • (ii) determining whether the URL is valid;
  • (iii) if valid, determining whether the stored contextual data is current;
  • (iv) if not current, accessing the Internet to update the contextual data; and
  • (iv) updating the volatility data to reflect validity, currentness, and how often the contextual data is updated.

In plain terms: tag each URL with a "how likely is this to change?" value, re-check URLs on a schedule driven by that value, detect whether the page still works and whether its content changed, refresh the stored content when it has changed, and feed those results back into the schedule. (Drafting note: the issued claim text literally labels two consecutive steps "iv)" — an apparent typographical defect in the granted claim; it does not create a third novel requirement.)

Dependent claims: Claims 2–5 (from claim 1) add: comparing newly retrieved Web page information against stored information and updating on mismatch (2); storing "change data" reflecting frequency of validation match failures (3); selecting which URL to validate based on that change data (4); and purging a URL plus its associated page information when repeatedly found invalid within a predetermined period (5). Claims 7–10 (from claim 6) add: validation criteria that become increasingly sensitive as change frequency rises (7); re-validation frequency approximately equal to or greater than the reflected update frequency (8); adding new URLs with a default initial update frequency (9); and identifying/validating new URLs from a plurality of heterogeneous data resources (10).

Key supporting details from the specification (context for the claims)

• The discrimination engine uses a regular-expression parser on a full Net News (NNTP) feed (up to ~1 GB/day at the time) and on ListServ/list-server mail streams, and also recognizes incompletely formed locators such as www.sub-domain.top-level-domain (implied HTTP) and ftp.sub-domain.top-level-domain (implied FTP).
• New URLs not already in the database are validated by issuing a client request over the Internet; barred-access, "no longer exists," or no-response outcomes are treated as invalid.
• Purge threshold: preferably failure of five consecutive validation attempts within a ten‑day period marks a URL "dead" and purges its context.
• The database is initially seeded by a conventional web crawler limited to a recursive crawl depth of five; repeated dynamic references increment a repeated-reference count field as a proxy for activity/importance.
• New URLs get a default re-validation period of about one week, with forced re-validation (and crawling of deeper linked URLs) the first time a newly added URL is considered.

Litigation / docket status

• CAFC 2026 dockets: I found no CAFC docket, opinion, or briefing naming US 5,855,020. Search queries for the patent number combined with "Court of Appeals for the Federal Circuit," "lawsuit," and "docket" returned no substantive hits.
• PTAB / district court: No IPR, PGR, or district-court case specifically asserting 5,855,020 surfaced. IPR2022‑00096 and the Intellectual Ventures I LLC v. Hewlett Packard Enterprise Co., No. 6:21‑cv‑00226 (W.D. Tex.) matters that appeared in results concern U.S. Pat. No. RE44,818, a different patent — not 5,855,020. I did not conflate them.
• Confidence: Low-to-moderate on the negative litigation finding. Free-web searches are not a complete substitute for a full Lexis/Westlaw or Docket Navigator docket sweep, and I was unable to run exhaustive docket queries. Given the patent expired on 2016‑02‑21, active enforcement in 2026 would be unusual, but I cannot rule it out (e.g., past litigation, or damages/back-royalty proceedings) with certainty.

Bottom line: US 5,855,020 is Steven T. Kirsch's 1996 Infoseek "Web scan process" patent — an early web-crawling/indexing system that harvests URLs from e‑mail and news feed messages, validates and fetches each URL, stores the URL with its page content, and then maintains freshness via volatility-driven periodic revalidation and purging. It is expired, now held (per public records) by Hewlett Packard Enterprise Development LP, and I found no current litigation tied to the number.

I'll search for litigation involving US Patent 5855020 across patent litigation databases.

Let me search more targeted litigation databases.

Let me search additional litigation-specific sources.

Litigation Search: U.S. Patent No. 5,855,020

Patent identification (confirmed): U.S. Patent No. 5,855,020, titled "Web scan process," inventor Steven T. Kirsch, filed 1996-02-21, granted 1998-12-29, original assignee Infoseek Corporation. Assignee chain per Google Patents: Infoseek → Ultraseek → Verity → Hewlett-Packard → Hewlett Packard Enterprise Development LP. Status: Expired – Lifetime (anticipated expiration 2016-02-21). It is the parent of continuation U.S. Patent No. 6,269,370 (also "Web scan process").

Result: No litigation confirmed

Based on my searches of publicly indexed patent-litigation and court sources, I could not identify any lawsuit in which U.S. Patent No. 5,855,020 was asserted or otherwise at issue. I found no case in which this specific patent number appears as a patent-in-suit, and therefore cannot provide plaintiff(s), defendant(s), jurisdiction, case number, filing date, or outcome.

What I checked and what I found

Important caution about false positives — patents I did NOT confuse with 5,855,020:

• DataTerm, Inc. v. MicroStrategy Inc. (D. Mass., No. 11-cv-11970, and a related New York action) — the court papers refer to a "’502 patent," but this is a different patent concerning construction of the claim term "class" in a database patent. It is not U.S. 5,855,020. I did not attribute it here.
• Eko Brands v. Adrian Rivera / ARM — involves the "’855 patent" (beverage brewing), unrelated.

Caveats on completeness

• The sources I could reach are the publicly indexed ones named in the task plus general web results. I was not able to directly query PACER or a full Docket Navigator/Unified Patents litigation docket for this number within this session, so this is a "no public record found" conclusion rather than a certified negative.
• Notably, the patent expired on 2016-02-21. Combined with its brief 1996–2016 life and its ownership through Infoseek/Ultraseek/Verity/HP, the absence of indexed litigation is plausible but should be independently verified.

Recommended verification steps for a definitive answer

1. Query PACER (pacer.uscourts.gov) via a full-text/party search for "5,855,020" and for assignee parties (Infoseek, Ultraseek, Verity).
2. Search Unified Patents Portal litigation caselist and the PTAB (for any IPR/CBM/PGR) — note the patent's expiration predates the AIA post-grant era partially, so PTAB activity is unlikely.
3. Check Lex Machina / Docket Navigator / CourtListener RECAP full-text dockets.
4. Review USPTO Patent Center / Assignment records for any litigation-related documents or "Sister" patent cross-references.

Bottom line: On the available record, no litigation involving U.S. Patent No. 5,855,020 is known. Any records found incidentally that use the shorthand "’502 patent" must be checked against the actual patent number before attribution, as at least one well-known "’502" case (DataTerm v. MicroStrategy) concerns a different patent.

I'll verify the structured-data finding with web searches before drafting.

Proceedings overview

Total AIA trial proceedings on file for US 5,855,020: 0 — zero active, zero with claims invalidated, zero with claims sustained, zero settled, and zero institution denials. The structured PTAB block (USPTO Open Data Portal, most recent ingest) is empty, and my independent web searches returned no IPR, PGR, or CBM record for this patent number; nothing in the searches contradicted the structured data. The bottom-line defensive posture is unusual and stronger than "hardened": the patent expired on 1996-02-21 + 20 years = 2016-02-21, it has never been through an AIA trial, its claims 1–10 stand exactly as granted, and as of today (2026-09-29) there is no live proceeding to time, no FWD to cite, and — critically — no enforceable window left in which the claims can be infringed. A demand letter citing US 5,855,020 today has a damages problem, not just a validity problem.

For each proceeding: none exist. Rather than fabricate entries, I've documented the empty record and its consequences below.

Proceedings overview — supporting detail

No proceedings to enumerate

Per the operating rule to use the structured block as canonical, the list is:

• IPR petitions: none
• PGR petitions: none (and as a pre-AIA, 1996-filed patent, PGR is unavailable in any event — PGR applies only to patents with a filing date on or after 2016-03-16)
• CBM petitions: none. Note also that CBM review under AIA § 18 required claims directed to a "financial product or service" (or a covered business method); the '020 claims are directed to URL discrimination, validation, and web-crawl maintenance machinery, so CBM eligibility was never plausible.
• Reexamination (ex parte / inter partes): no reexamination certificate appears associated with the patent in the record I reviewed. I did not exhaustively verify the ex parte reexamination file history, so treat this as "no indication found," not a certified negative.
• Federal Circuit appeals: none tied to a PTAB proceeding, because there is no PTAB proceeding to appeal. The patent's prosecution-era appeal history, if any, is not an AIA trial issue.

Verification trail (so you can reproduce the negative)

Retrieval pages if you want to confirm independently: USPTO PTAB E2E (search Patent Number 5855020), Google Patents US5855020A, USPTO Patent Assignment record, CourtListener search.

Strategic summary

Claim status: everything is UNTESTED and everything is EXPIRED. Claims 1–5 (the system claims: discrimination system → validation system → searchable database) and claims 6–10 (the method claims: volatility data, contextual data, periodic validation, purge on repeated invalidity) were all issued on 1998-12-29 and remain intact in the printed record. No PTAB panel, no district court, and no reexamination has canceled, narrowed, or even construed any of them that I can locate. That sounds like a plaintiff's dream until you get to the date arithmetic: the application was filed 1996-02-21, the Google Patents record carries "Anticipated expiration 2016-02-21" and legal status "Expired – Lifetime," and the current assignee of record is Hewlett Packard Enterprise Development LP (via Ultraseek → Verity → HP → HPE). There are no surviving claims in any commercially meaningful sense, because there is no term left.

Estoppel landscape — the short answer is that § 315(e)(2) is irrelevant here. Because no IPR/PGR was ever instituted against the '020, no petitioner (or privy) is estopped from anything, and there is no FWD whose grounds are now foreclosed. That cuts both ways: there is no "safe" ground that a prior petitioner already lost on, but there is also no need to find one. For a defendant being asserted against today, the operative bars are temporal, not estoppel-based: (1) § 286 damages lookback — even the full six years before suit reaches back only to roughly 2020-09-29, which is after the 2016-02-21 expiration, so there is no compensable infringement period left; (2) § 271 — an expired patent cannot be infringed by post-expiration conduct; and (3) if the demand is framed as ongoing royalties, the license-negotiation posture collapses entirely. Any prior-art ground you'd want to raise is still legally available to you (nothing is estopped), but you'd be litigating invalidity of a dead patent, which is a waste of budget. Lead with expiration, not with art.

Pattern signals. There is no petitioner pattern because there is no petitioner: none of Infoseek, Ultraseek, Verity, or HPE ever had the '020 challenged, and no defensive aggregator (Unified Patents or similar) shows up in the chain for this patent number. The family tells a more interesting story than the '020 itself: the '020 shares a 1996-02-21 priority date with Kirsch's sibling redirection patent US 5,751,956 and its continuations, which the Google Patents "Cited By" and family tables show later issuing to Disney Enterprises, Inc. (US 8,117,286; US 7,702,752). The '020 also carries an enormous forward-citation footprint (227 US citations in the listing), meaning the industry treated it as foundational — which is exactly the profile that normally attracts IPRs. The reason it never got one is almost certainly that it was already near or past the end of its term before AIA trials became available in 2012-09-16. The "absence of PTAB activity is itself a signal" heuristic needs an asterisk here: this is not a quietly-hardened patent, it is an expired one.

Recommended next steps

1. If you are a defendant and the plaintiff cites claims 1–5 or 6–10: there is no FWD to quote — none exists — so build the defense on the face of the document instead. Cite the front page: filed 1996-02-21, granted 1998-12-29, status "Expired – Lifetime," anticipated expiration 2016-02-21. A 20-year term from 1996-02-21 ran out over a decade ago; the claims were never canceled because they died of old age, not of prior art.
2. Run the damages math before you run the invalidity math. Compute the § 286 six-year window against the 2016-02-21 expiration. If the complaint or demand letter is dated after 2022-02-21, the entire recoverable window is legally empty. That is a Rule 12 motion or a summary-judgment motion, not an IPR.
3. Do not file an AIA petition. It would be a waste of the filing fee and a volunteer opportunity to fund your opponent's attorney fees; and institution of an IPR on claims you also argue are expired adds nothing you don't already have. (IPRs on expired patents are not categorically barred, but there is no upside here.)
4. Confirm the negative yourself at PTAB E2E using Patent Number 5855020 and check the USPTO assignment record for any post-2016 HPE transfers or reversion that could change who holds the paper. Also pull the file wrapper at Patent Center to confirm no reexamination certificate or certificate of correction issued — I found no indication of one, but I have not certified that.
5. Scope check on the family. Before conceding the whole portfolio is dead, ask whether the demand letter is actually about the '020 or about a sibling (the '956 redirection family now showing Disney Enterprises as assignee). Those siblings have different expiration dates and different status. The '020 analysis above does not automatically transfer to them.

Caveats on this report

• The ODP ingest may lag recent filings in principle, but for this patent the lag explanation is moot: the term expired 2016-02-21, so no active trial could be pending today.
• I did not locate any district court litigation specifically asserting US 5,855,020; if such a case exists, it did not surface in my searches and is not reflected above. Do not treat "no litigation found" as a verified negative.
• One minor data discrepancy worth knowing: a Unified Patents portal listing shows the '020 priority date as 1996-02-20, while Google Patents and the ODP-derived structured data show 1996-02-21. Nothing substantive turns on it (both are 1996 and both predate any AIA regime), but flag it if you rely on the date in a filing.

I'll research the assignment record for US 5,855,020, prioritizing the USPTO Assignment Center and cross-referencing the sources named in the task.

Let me pull the uspto.report assignment page and search for the specific reassignment records with reel/frame data.

Let me look for the actual reel/frame assignment data and correspondent of record.

Assignment & Ownership Analysis — U.S. Patent No. 5,855,020 ("Web scan process")

Data-integrity note up front. I was able to confirm the existence and sequence of recorded assignments (as mirrored in the Google Patents "Legal Events"/reassignment timeline, which is populated from the USPTO assignment record), but I was not able to retrieve the actual reel/frame numbers, execution-vs-recording dates, or the correspondent of record from the USPTO Patent Assignment Center or the assignment.uspto.gov index within this session. I am therefore not inventing reel/frame values or correspondent names. Where the record is silent, I say so. The chain below should be treated as a verified sequence with unverified instrument metadata, and the reel/frame/correspondent cells flagged as [not retrievable this session — verify].

Primary source to verify: USPTO Patent Assignment Search — https://assignmentcenter.uspto.gov/ (search "5855020") and https://assignment.uspto.gov/patent/index.html

Inventors

Unusual-pattern check: The prompt asks whether "all inventors departed the original assignee within 12 months of filing." With a single inventor, there is no multi-inventor attrition pattern to assess. Kirsch is publicly reported to have moved on from Infoseek in the mid-to-late 1990s and subsequently founded other ventures, but I cannot verify an exact departure date within this session — treat "inventor departed within 12 months" as unclear / not established. There is no evidence of a founder-attrition-before-fire-sale dynamic.

Original assignee

Infoseek Corporation — named on the issued patent and recorded as assignee from the original inventor assignment (recorded 1996-02-21, per Google Patents legal events).

• Primary line of business: Internet search engine / web portal operator. Infoseek ran one of the major 1990s web search engines and also commercialized the Ultraseek enterprise search product, which is the direct commercial embodiment of the "web scan"/crawl-and-index subject matter claimed here.
• Shipped a product embodying the claims? Yes — Infoseek's crawler/index and the Ultraseek search product are the operative commercial embodiments.
• Current status: No longer independent. Infoseek was absorbed into [The Walt Disney Company](/litigations/by-plaintiff/The%20Walt%20Disney%20Company)'s Go.com internet group (Disney acquired Infoseek in 1999), and the Ultraseek enterprise-search business was subsequently divested — publicly reported as acquired by Inktomi Corporation, which in turn sold the "Enterprise Search Business" to Verity, Inc. (see Asset Purchase Agreement dated 2002-11-13 among Inktomi, Quiver, Inc., Ultraseek Corporation, Quiver Ltd., and Verity, Inc.). Verity was later acquired by Autonomy (2005), then by Hewlett-Packard (2011). Infoseek as a standalone entity is dissolved/absorbed.

Assignment timeline

Chronological reconstruction (sequence per Google Patents legal events / USPTO reassignment records). All reel/frame and correspondent fields could not be retrieved this session.

1. 1996-02-21 (recorded) — Reel [not retrievable this session — verify] [reel NNNNNN/NNNN]

   • Conveyance: Assignment (inventor → company; original/employment assignment)
   • Assignor: Steven T. Kirsch (individual)
   • Assignee: Infoseek Corporation
   • Correspondent: [not retrievable this session — verify]
   • Context: Original assignment of inventor rights to the operating company at filing — standard practice, not an NPE event.

2. 2002-11-04 — Reel [not retrievable this session — verify]

   • Conveyance: Assignment (asset/business transfer)
   • Assignor: Infoseek Corporation
   • Assignee: Ultraseek Corporation
   • Correspondent: [not retrievable this session — verify]
   • Context: Divestiture — carries the patent into the Ultraseek enterprise-search entity that was sold to Verity; ties to the Inktomi/Verity Enterprise Search Business transaction (APA dated 2002-11-13). Third-party, not a shell-LLC hop.

3. 2003-01-21 — Reel [not retrievable this session — verify]

   • Conveyance: Assignment (asset purchase)
   • Assignor: Ultraseek Corporation
   • Assignee: Verity, Inc.
   • Correspondent: [not retrievable this session — verify]
   • Context: Acquisition — Verity's purchase of the Enterprise Search Business (public-company buyer, Verity Inc. DE). Operating-company M&A, not NPE.

4. 2015-07-06 — Reel [not retrievable this session — verify]

   • Conveyance: Merger
   • Assignor: Verity, Inc.
   • Assignee: Hewlett-Packard Company
   • Correspondent: [not retrievable this session — verify]
   • Context: Internal corporate reorg — post-Autonomy/HP merger cleanup; no independent third-party buyer.

5. 2015-10-01 — Reel [not retrievable this session — verify]

   • Conveyance: Assignment (intra-group)
   • Assignor: Hewlett-Packard Company
   • Assignee: Hewlett-Packard Development Company, L.P.
   • Correspondent: [not retrievable this session — verify]
   • Context: Internal corporate reorg/IP-holding consolidation.

6. 2015-11-09 — Reel [not retrievable this session — verify]

   • Conveyance: Assignment (intra-group)
   • Assignor: Hewlett-Packard Development Company, L.P.
   • Assignee: Hewlett Packard Enterprise Development LP
   • Correspondent: [not retrievable this session — verify]
   • Context: Internal corporate reorg — reflects the November 2015 HP split into HP Inc. and Hewlett Packard Enterprise. Terminal assignee is an operating company's IP-holding subsidiary, not a licensing-only NPE.

Flagged gap / potential contradiction (per cross-reference instruction):

• The recorded chain shows Infoseek → Ultraseek directly in 2002 and omits both (a) the Disney/Go.com acquisition of Infoseek (1999) and (b) any recorded Inktomi assignment link, even though public reporting places the Ultraseek business under Inktomi before the Verity sale. Either the Infoseek→Ultraseek recording is a confirmatory instrument covering an earlier transfer, or the intermediate holders' recordings are not reflected in this patent's record. Worth verifying at the reel/frame level.
• This is consistent with the previously generated litigation summary (no litigation identified): there is no "pre-litigation transfer" record to reconcile.

Timeline diagram

timeline
    title Ownership of US 5855020
    1996 : Filed by Infoseek Corporation
         : Inventor Kirsch assigns rights to Infoseek
    2002 : Assigned to Ultraseek Corporation
    2003 : Assigned to Verity Inc
    2015 : Verity merged into Hewlett Packard Company
         : Assigned to HP Development Company LP
         : Assigned to HPE Development LP
    2016 : Patent reaches anticipated expiration

NPE / troll-pattern signals

Verdict

Defensive / non-asserting — with an explicit fit caveat.

Justification (2–3 sentences): The recorded chain is a clean operating-company lineage — inventor Kirsch → Infoseek Corporation (1996-02-21) → Ultraseek Corporation (2002-11-04) → Verity, Inc. (2003-01-21) → Hewlett-Packard Company (2015-07-06, merger) → HP Development Company LP (2015-10-01) → Hewlett Packard Enterprise Development LP (2015-11-09) — with zero of the eight NPE signals present and no identified assertion, and the patent reached its anticipated expiration on 2016-02-21 ("Expired – Lifetime"). Because the terminal assignee is an operating company's IP-holding subsidiary rather than a listed defensive aggregator (RPX/AST/LOT/Unified/OIN), the rubric's "Defensive / non-asserting" trigger is met only in the broad sense of "non-asserting"; strictly, the chain is an operating-company chain, so if your methodology requires a defensive-aggregator terminal assignee, the correct label is no NPE pattern — insufficient signal to classify as NPE rather than a true defensive-aggregation finding. No result supports "NPE — high" or "NPE — moderate."

Verification links:

• USPTO Patent Assignment Center (search by patent number 5855020): https://assignmentcenter.uspto.gov/
• USPTO Patent Assignment Search (indexed mirror): https://assignment.uspto.gov/patent/index.html
• Google Patents legal events (reassignment sequence used above): https://patents.google.com/patent/US5855020/en

Recommended next steps to close the data gaps (do these before relying on the verdict):

1. Pull each record's reel/frame, execution date, recording date, and correspondent from the Assignment Center for patent 5855020 — the 2002/2003 and 2015 entries are the priority.
2. Compare the correspondent of record across all six entries to test signal #3 (repeat correspondent) — a shared attorney across the 2002–2003 Infoseek/Ultraseek/Verity hops would be expected in an M&A cleanup and is not, by itself, an NPE tell.
3. Query PTAB for any IPR/CBM/PGR (unlikely given the timeline) and PACER/CourtListener/Lex Machina for any assertion, to firm up signals #5 and the "non-asserting" call.
4. Verify HPE's LOT Network membership if you intend to argue a neutralization angle.

I'll verify the citation details and search for the specific references before analyzing prior art.

Prior-Art Analysis — U.S. Patent No. 5,855,020 ("Web scan process")

Scope note: Per the task, I analyzed every patent citation listed on the face of US 5,855,020 (the 5 U.S. references in the "Patent Citations" / "Citations" lists of the granted patent) plus the cited non-patent literature. This analysis builds on the earlier bibliographic and claim-summary sections rather than repeating them. Where a reference does not anticipate, I say so explicitly and explain the missing element, because the honest answer here is that none of the cited references anticipates the independent claims.

1. Legal framework applied (important, and frequently misunderstood)

US 5,855,020 has an effective filing/priority date of 1996‑02‑21 and is governed by pre‑AIA 35 U.S.C. § 102 (the AIA first‑to‑file provisions apply only to applications filed on/after 2013‑03‑16). So:

• § 102(b) statutory bar: a printed publication or patent published more than one year before 1996‑02‑21 — i.e., before 1995‑02‑21.
• § 102(a): knowledge/use/publication by others before the applicant's invention date.
• § 102(e): a U.S. patent granted on an application filed before the applicant's invention (in practice, before the applicant's filing date). Critically, a reference patent filed the same day as, or after, 1996‑02‑21 is NOT § 102(e) prior art against this patent.

Anticipation requires a single reference to disclose every limitation of a claim, arranged as claimed. Because claim 1 requires a "discrimination system … to receive messages including electronic mail messages and network news messages … to identify embedded URLs," and claim 6 requires "predetermined volatility data" driving periodic validation of currentness with feed‑back updating of that volatility data, a reference must teach those features to anticipate — not merely generic web/URL/database functionality.

2. The five patent citations on the face of US 5,855,020

Dates cross‑checked against Google Patents and uspto.report bibliographic records (searched 2026‑04‑26).

Reference 1 — US 5,710,918 A (Lagarde & Rogers, IBM)

• Full citation: Lagarde, K.C. & Rogers, R.M., "Method for distributed task fulfillment of web browser requests," U.S. Patent 5,710,918, filed Jun. 7, 1995, granted Jan. 20, 1998 (App. 08/474,572); assignee International Business Machines Corp.
• Brief description: A WWW browser makes requests to a web server that acts as an agent of the browser, organizing distributed "DIS" sub‑agents and "capsule" command objects; on a single HTML request, the server retrieves data from multiple heterogeneous databases (DB2, Oracle, Sybase, Redbrick, etc.) across an intranet/Internet, then calculates, formats, and returns a report to the browser. The disclosure uses URLs generically as Internet addresses.
• § 102 analysis: This reference is about servicing a client browser's request by assembling data from databases. It does not disclose a discrimination system that ingests e‑mail and Net News messages and parses them for embedded locators, nor volatility‑based periodic re‑validation/purge of stored locators.
• Claims potentially anticipated: None. It arguably touches the "access a network to retrieve information" notion of claim 1(b), but fails the claim‑1(a) message‑mining element and all of claim 6's volatility/currentness elements. Best characterized as background art (potential § 103 combination material), not § 102 anticipation.

Reference 2 — US 5,752,246 A (Lagarde et al., IBM)

• Full citation: Lagarde, K.C. et al., "Service agent for fulfilling requests of a web browser," U.S. Patent 5,752,246, filed Jun. 7, 1995, granted May 12, 1998 (App. 08/474,576); assignee IBM.
• Brief description: Same IBM family as Reference 1 — a server‑side "service agent"/control‑program agent plus DIS capsules that fulfill browser requests by retrieving and processing data from disparate sources over intranet/Internet.
• § 102 analysis: Identical deficiencies: no mining of e‑mail/news feeds for embedded URLs; no autonomous validate‑and‑store‑with‑context; no volatility‑driven revalidation.
• Claims potentially anticipated: None.

Reference 3 — US 5,572,643 A (Judson)

• Full citation: Judson, D.H., "Web browser with dynamic display of information objects during linking," U.S. Patent 5,572,643, filed Oct. 19, 1995, granted Nov. 5, 1996 (App. 08/543,876).
• Brief description: A web browser that, while a link to a second hypertext document is being downloaded, retrieves a previously cached "information object" (embedded via an HTML comment tag) and displays it as a "mini" web page — reducing perceived download latency; contemplates advertisements, notices, etc.
• § 102 analysis: Though it was filed before 1996‑02‑21 and is thus a § 102(e) candidate, its disclosure is confined to client‑side browsing/display behavior. It says nothing about a discrimination engine scanning dynamic message streams, a validation engine that issues network requests to verify a locator, or a searchable URL+content database with volatility scheduling.
• Claims potentially anticipated: None. Its nearest overlap is generic recitation of NNTP/Usenet, HTTP, URLs, and SMTP/e‑mail as coexisting Internet services — not the harvesting/validation/currentness pipeline claimed.

Reference 4 — US 5,761,499 A (Sonderegger, Novell)

• Full citation: Sonderegger, K.E., "Method for managing globally distributed software components," U.S. Patent 5,761,499, filed Dec. 21, 1995, granted Jun. 2, 1998 (App. 08/576,647); assignee Novell, Inc. (claims priority to provisionals 60/000,200 filed 1995‑06‑14 and 60/003,615 filed 1995‑09‑12).
• Brief description: Locating and registering COM/OLE software components on a LAN or the Internet. A database associates software‑component identifiers with software‑component locators; unregistered components are found by search (including via an "Internet search engine database"), then transferred and registered. Claim 23 expressly recites accessing an Internet search‑engine database that associates component identifiers with locators.
• § 102 analysis: This is the closest structural analogy among the five — a searchable database mapping identifiers to locators, with network retrieval — which likely explains why the examiner cited it. But the "component" is a software binary object, not a web page; the locator is a component location, not a Web page URL; and there is no scanning of e‑mail/news to discover locators and no volatility‑driven revalidation/currentness maintenance.
• Claims potentially anticipated: None. Relevant only as a § 103‑style secondary reference on the "database associating identifiers with locators" concept.

Reference 5 — US 5,751,956 A (Kirsch, Infoseek) — NOT prior art

• Full citation: Kirsch, S.T., "Method and apparatus for redirection of server external hyper‑link references," U.S. Patent 5,751,956, filed Feb. 21, 1996, granted May 12, 1998 (App. 08/604,584‑era filing; same assignee Infoseek Corp.).
• Brief description: Redirects external hyper‑link references so that off‑site link clicks are routed through a server‑side redirection mechanism (underlying the "ultraseek" redirection concept).
• § 102 analysis — dispositive point: This reference was filed the same day (1996‑02‑21) as US 5,855,020, by the same inventor and the same assignee, and is expressly listed in US 5,855,020's own "Cross‑Reference to Related Applications" section. A same‑day, commonly‑owned, same‑inventor application is not § 102(e) prior art (it was not filed "before the invention by the applicant"), and it could not be combined for § 103 purposes under the pre‑AIA common‑ownership exception of § 103(c). Its appearance in the citation list reflects a related‑application cross‑reference, not an anticipatory teaching.
• Claims potentially anticipated: None (not prior art at all).

3. Cited non‑patent literature (as listed on the patent)

Ten non‑patent citations appear, comprising five unique items (each duplicated in the record):

1. Berners‑Lee, "The World‑Wide Web," Communications of the ACM, v.37, n.8, pp. 76–82, Aug. 1994 — foundational WWW/HTTP/URL architecture article.
2. Cole et al., "Oracle Spins Web Strategy," Network World, v.12, n.3, pp. 1 & 49, Jan. 16, 1995 — industry news on Oracle's web strategy.
3. Davis, J., "EMail World/Internet Expo to Feature Web Solutions," v.18, n.8, p.6, Feb. 19, 1996 — trade‑press notice (published two days before the filing date).
4. Nadile, L., "Adobe Targets Mac Web Development," PC Week, v.12, n.40, p.62(1), Oct. 9, 1995 — industry news.
5. Snell, J., "Webtop Publishing Here at Last," MacUser, v.11, n.12, p.44(2), Dec. 1995 — industry news.

• § 102 analysis: Item 1 (Berners‑Lee, Aug. 1994) predates the 1995‑02‑21 § 102(b) cutoff and is a genuine printed publication, but it describes the general WWW/URL system — it does not disclose mining e‑mail/Net News for embedded URLs, nor validate‑store‑with‑context, nor volatility scheduling. Items 2–5 are general‑interest trade‑press pieces quoting no such system. None anticipates any claim. They function as background showing the state of web/URL technology as of early 1996.

4. Bottom line on § 102 anticipation

• No reference cited on the face of US 5,855,020 anticipates any claim (independent claims 1 or 6, or dependents 2–5 and 7–10), whether taken singly or (as the task is framed for § 102) individually.
• The single most relevant cited reference for structural similarity is US 5,761,499 (Novell) — a searchable database that maps identifiers to network locators — but it concerns software components, not web pages, and lacks the message‑stream discovery and volatility‑based revalidation that define the 5,855,020 invention.
• US 5,751,956 should be excluded from the prior‑art set entirely: same‑day, same‑inventor, same‑assignee, and expressly cross‑referenced in the patent. Flagging this corrects a common misreading of citation lists as "prior art."
• The claims were allowed over these references, which is consistent with the analysis above: none teaches the dynamic‑source URL mining → Internet validation → URL+content storage → volatility‑driven revalidation/purge pipeline.

5. Confidence and caveats

• High confidence on the bibliographic dates and on the § 102(e) date rules (verified via Google Patents and uspto.report on 2026‑04‑26).
• Moderate‑to‑high confidence on the anticipation conclusions; these rest on the references' abstracts, claim sets, and detailed descriptions as retrieved, not on a limitation‑by‑limitation full‑text claim chart. A definitive anticipation opinion would require element‑by‑element charts against each reference's complete disclosure.
• I report the citation set literally as it appears on US 5,855,020 and did not substitute other patents. Note that the patent's own text separately lists a "Family Cites Families" set (e.g., US 5,530,852 Meske; US 5,790,793 Higley; US 5,659,732 Infoseek) and a large "Cited By" list (227 entries per the record) — those are references that cite this patent or its family, not prior art cited against it, and I have excluded them from the anticipation analysis above.
• No litigation was identified for this number (consistent with the earlier litigation section); nothing changed that finding.

I'll verify the reference disclosures I need for the §103 combinations before drafting.

Obviousness Analysis — U.S. Pat. No. 5,855,020 ("Web scan process") under pre‑AIA 35 U.S.C. § 103

Scope note / flag against the earlier sections. This analysis builds on the Prior Art section already generated, which concluded that none of the five examiner‑cited references anticipates any claim. I agree with that conclusion and do not repeat it. § 103, however, asks a different question, and answering it honestly requires me to go beyond the examiner's citation list. Two references I rely on heavily here — Higley, US 5,790,793 and Meske, US 5,530,852 — appear in the patent's "Family Cites Families" / forward‑citation set (which the earlier section expressly excluded from the anticipation analysis), and a third — Harvest (Bowman et al., 1994) — is not in the record at all. I am flagging this expansion explicitly. Tier‑1 combinations below use only the examiner‑cited references; Tier‑2 combinations add the extra art.

1. Legal framework and level of ordinary skill

Common‑ownership caveat carried forward from the earlier sections: US 5,751,956 (Kirsch/Infoseek, filed the same day) is not § 102 prior art and is disqualified for § 103 combination under pre‑AIA § 103(c) (common ownership at the time of invention). It does not appear in any combination below.

2. Claim 1 — element decomposition and mapping

Recall claim 1 requires: (a) a discrimination system coupled to the Internet that receives messages including e‑mail and network‑news messages and processes them to identify embedded URLs; (b) a validation system coupled to the Internet that, for a given embedded URL, accesses the Internet to retrieve Web page information associated with it; and (c) a database that searchably stores the URL in association with that Web page information.

2.1 Tier‑1: combinations using only the examiner‑cited references

Assessment of Tier‑1: the cited art supplies (b) and (c) cleanly, but element (a) — the "discrimination system … to identify embedded URLs" in e‑mail and news messages — is the crux of the claim, and no cited reference teaches it. A § 103 rejection grounded only on the cited references would therefore rest on an unsupported "inherent incentive" argument for (a), and would be vulnerable. This is fully consistent with the prior section's anticipation finding and with the fact that the claims allowed.

2.2 Tier‑2: adding art already appearing in the patent record plus Harvest

Combination 1 (Tier‑2): Higley '793 + Meske '852 + Sonderegger '499 (+ Lagarde '918/'246 for the retrieval element).

Why a PHOSITA would have combined these (KSR‑supported):

1. Same field of endeavor / analogous art. All four references address Internet information handling — mail, news, HTTP, resource location. Higley, Meske and Sonderegger are all Internet information retrieval art as of 1995.
2. Known technique improving similar subject matter (KSR (C)). Higley teaches detecting a URL in a message and resolving it; the asserted advance is merely resolving and indexing those URLs at scale rather than displaying one to one reader. Substituting an indexing back end for a display front end is a predictable use of a known data‑storage technique — exactly the "known technique improving similar device" rationale.
3. Design incentive / market forces (KSR (F)). The '020 specification itself concedes the field's unmet need: web crawlers cannot keep maps complete or timely, and "the newest sources of information are often the most important." Given that admitted need, using mail and Net News — channels already known to carry URLs (Higley) and already being parsed and indexed (Meske) — as an additional discovery channel is an obvious design choice, not an inventive leap.
4. Predictable result (KSR (A)). Combining a message parser, a network fetch, and a locator database yields nothing more than the sum of the parts: a list of fetched URLs and their content. No new or unexpected behavior is required or claimed.

3. Claim 6 — element decomposition and mapping

Claim 6 requires: (a) associating predetermined volatility data with a URL in a list; (b) associating predetermined contextual data with that URL; and (c) periodically validating currentness, comprising (i) checking the volatility data against validation criteria, (ii) determining URL validity, (iii) if valid, determining whether the contextual data is current, (iv) if not current, accessing the Internet to update the contextual data, and (v) updating the volatility data to reflect validity, currentness, and how often the context was updated.

3.1 Tier‑2 primary reference: Harvest (Bowman et al., 1994)

Citation: C. Mic Bowman, Peter B. Danzig, Darren R. Hardy, Udi Manber, Michael F. Schwartz, "Harvest: A Scalable, Customizable Discovery and Access System," University of Colorado at Boulder Technical Report CU‑CS‑732‑94 (original date August 1994; revised March 1995). A printed publication dated before 1995‑02‑21 → § 102(b) art. (Harvest paper; NZDL copy)

Verified disclosures directly relevant to claim 6:

• Gathering from plural heterogeneous sources, including Net News. Harvest "provides an integrated set of customizable tools for gathering information from diverse repositories … The system interoperates with WWW clients and with HTTP, FTP, Gopher, and NetNews information resources." Its gatherer type table includes dedicated Mail and News summarizers ("Extract certain header fields"). → claim 10's "plurality of heterogeneous data resources."
• "Contextual data" per object. Each object is reduced to a SOIF summary object carrying fields including Update-Time, Last-Modification-Time, MD5 checksum, Title, Keywords, Partial-Text, and URL-References. → maps to claim 6(b) "predetermined contextual data."
• "Volatility data" per object. The same objects carry Refresh-Rate and Time-to-Live ("the number of seconds after Update-Time when the summary object is no longer valid"). → maps to claim 6(a) "predetermined volatility data" and to claim 6(c)(i) "predetermined validation criteria."
• Periodic re‑gathering / revalidating to keep the index fresh, with index replication and caching to reduce load. → maps to claim 6(c)(ii)–(iv).

Combination 2: Harvest + Sonderegger '499 + Higley '793.

• Sonderegger supplies the list of URLs and the association of data with locators (and its own "current location" updating).
• Higley supplies the network retrieval of the object at the locator.
• Harvest supplies the volatility field, the contextual (summary) field, and the refresh/revalidation loop — the heart of claim 6.

Motivation to combine (KSR): Harvest is not merely analogous art; it addresses the same problem in the same way, and its own stated motivation — reduce staleness while containing server/network cost — is the identical incentive recited in the '020 specification ("completeness, timeliness and currentness … is of paramount importance"). Combining Harvest's refresh‑rate-driven gatherer loop with a locator database (Sonderegger) and a fetch mechanism (Higley) is a mere application of Harvest's own scheduling mechanism to URLs — predictable result, known technique, express design incentive.

3.2 Corroborating prior art for the revalidation notion

• Cache-consistency / TTL art (e.g., Gwertzman & Seltzer, "World-Wide Web Cache Consistency," USENIX 1996; Chankhunthod et al., "A Hierarchical Internet Object Cache," USENIX Jan. 1996) — expiration/TTL-based refresh of stored network objects was conventional. Note the dates: these are early-1996 publications, i.e. potentially § 102(a) art only, and I flag them as corroborating, not as primary § 102(b) references.
• Deja News (public since May 1995; full‑text Usenet indexing updated "multiple times a day, to keep data current") shows the concept of a searchable, continuously refreshed index over a dynamic message feed was known and commercialized before the 1996 filing. Date caveat: May 1995 is after the 1995‑02‑21 § 102(b) cutoff, so Deja News is § 102(a) art at best (knowledge/use by others before the invention date), and it is a system/publication, not a patent — use with care. (contemporaneous 1995‑11‑06 Usenet post; Deja News background)

4. Dependent claims — obviousness mapping

5. Consolidated motivation‑to‑combine narrative

A PHOSITA in Feb. 1996, facing the admitted problem of incomplete and stale web indices, had before her:

• art teaching that e‑mail and news messages carry URLs and that those URLs can be detected and resolved (Higley '793);
• art teaching that e‑mail and news/other feeds can be parsed and placed into updated, searchable indices (Meske '852; also Harvest's Mail/News gatherers);
• art teaching a searchable database associating resource identifiers with network locators, including Internet search‑engine databases (Sonderegger '499);
• art teaching server‑side retrieval of information over the Internet via URLs (Lagarde '918/'246); and
• art teaching per‑resource "volatility" scheduling (Refresh‑Rate, Time‑to‑Live) and per‑resource contextual summaries (Update‑Time, MD5), with periodic re‑gathering to combat staleness (Harvest).

The claimed subject matter is the concatenation of these known mechanisms along a known workflow. Under KSR, that is enough: the combination produces no more than the expected sum of its parts (a list of URLs plus the content fetched from them, and a scheduled refresh), the field recognized the design incentive (freshness/completeness), and there was a finite, identified set of solutions. Notably absent is any teaching away: nothing in Higley, Meske, Sonderegger or Harvest warns against indexing message‑derived URLs or against volatility‑driven refresh.

6. Counterarguments, secondary considerations, and honest weaknesses

Arguments the patent owner could raise:

1. No single combination of the examiner‑cited references teaches element (a). Supported by the earlier section, and I agree. This is the strongest defense against a Tier‑1 rejection — and the reason the claims were allowed. It is materially weakened once Higley '793 and Meske '852 are admitted into the § 103 combination (they were not before the examiner during prosecution of the '020 — at least not in the printed record).
2. Teaching away / different purpose. Higley is a point‑to‑point messaging patent ("push" of one URL to one reader); Sonderegger concerns software components, not web pages; Harvest is a standalone indexing toolkit. A patent owner could argue there was no reason to repurpose an e‑mail‑annotation scheme (Higley) as a discovery channel for a search index. Rebuttal: KSR rejects the rigid "same purpose" requirement — the references are analogous art addressing Internet information retrieval, and the motivation arises from the field's recognized need for fresher indexes.
3. Secondary considerations. Possible evidence includes the commercial success of Infoseek/Ultraseek and the patent's large forward‑citation footprint (227 U.S. citations per the Google Patents record). However: (i) forward citations are not a Graham secondary consideration and prove only that later inventors cited it; (ii) commercial success requires a nexus to the claimed invention, not merely to the search‑engine market; and (iii) I found no evidence of unexpected results in the record. Absent a nexus, these do not rebut the prima facie case.

Weak links I will not paper over:

• Claim 1(a) remains the pivotal element. Everything turns on whether Higley/Meske/Harvest render it obvious. Meske teaches parsing e‑mail into an index; Higley teaches detecting URLs in mail and news. Neither, standing alone, teaches an automated url‑harvesting engine over a full news feed — the '020's discrimination engine. The combination argument is strong on rationale but is a combination, not a clean single‑reference teaching. Confidence: moderate‑to‑high.
• Claim 6 / Harvest is the strongest ground in the patent. Harvest's Refresh-Rate, Time-to-Live, Update-Time, MD5, and multi‑protocol (HTTP/FTP/Gopher/NetNews) gathering map onto claim 6 and claims 9–10 with unusual precision. Confidence: high.
• Claim 5 (purge) is the softest dependent — the "five failures in ten days" floor is not disclosed; obviousness rests on it being a predictable threshold choice. A patent owner could argue the specific retry/suppression regime is not taught. Confidence: moderate.
• Date diligence. Harvest is solid § 102(b) art (Aug. 1994). Deja News and the cache‑consistency papers are § 102(a)-only (post‑1995‑02‑21) and should not be presented as § 102(b) bar art. I have labeled them accordingly.

7. Flagged contradictions / discrepancies with the earlier sections

1. Scope expansion (explicit). The Prior Art section analyzed only the five examiner‑cited references + NPL and stated that Higley '793 and Meske '852 (and the other "Family Cites Families" items) were excluded from the prior‑art set. This § 103 analysis re‑admits Higley '793, Meske '852 and adds Harvest, because § 103 permits combination with art beyond the examiner's citation list. Any downstream use of these conclusions must carry that provenance caveat.
2. Higley '793's own post‑grant record. US 5,790,793 has a reexamination certificate and a PTAB trial certificate; in IPR2014‑00500 its claim 1 was found anticipated, and in Reexam 90/013,677 / Appeal 2017‑011415 claims 19, 20 and 41 were affirmed rejected. This affects Higley's validity, not its disclosure — it remains citable for what it describes. Flagging so the two are not conflated.
3. Harvest is absent from the record entirely. It is not in the '020 patent's citation lists, the "Family Cites Families" list, or the 227‑entry "Cited By" list. Its introduction is my addition based on independent verification of the Aug. 1994 technical report. It is the single most consequential addition; if the analysis is used in a filing, the report should be authenticated as a printed publication.
4. Litigation/term caveat carried forward. Per the earlier sections, no litigation or PTAB proceeding names 5,855,020, and the patent expired 2016‑02‑21. A § 103 analysis is therefore relevant only to a retrospective validity challenge (or a past‑damages/licensing posture) — not to prospective enforcement. No § 315(e)(2) estoppel applies anywhere in this analysis.

8. Bottom line

• Anticipation: none (consistent with the earlier section).
• Obviousness: strong, but combination‑dependent.
  • Claim 1 is obvious over Higley '793 + Meske '852 + Sonderegger '499 (with Lagarde '918/'246 for the retrieval element), because the only element not in the examiner‑cited set — message‑stream URL harvesting — is squarely taught once Higley and Meske are admitted. Moderate‑to‑high confidence.
  • Claim 6 and dependents 7–10 are obvious over Harvest (Bowman et al., Aug. 1994) in view of Sonderegger '499 and Higley '793, because Harvest discloses per‑object volatility fields (Refresh‑Rate, Time‑to‑Live), per‑object contextual summaries (Update‑Time, MD5), periodic re‑gathering from plural heterogeneous sources including NetNews, and default field values. High confidence.
  • Claims 2, 3, 4, 9, 10 follow with high confidence; claims 5, 7, 8 with moderate‑to‑high confidence, framed as predictable design variations under KSR.
• Practical caveat: given the 2016‑02‑21 expiration (and the absence of any FWD or litigation, per the earlier sections), the § 103 case is academic for prospective enforcement — it matters only if validity is at issue for the 1996–2016 term.