US 6658625

I'll research this patent number. Let me run several targeted searches.

Let me check the claims and any litigation/docket activity.

Let me search specifically for the claim language and any 2026 docket activity.

Summary — US Patent 6,658,625

Scope note on sources. The authoritative full text supplied for this analysis (Google Patents, fetched 2026‑09‑24) contains the front matter, abstract, background, and detailed description, but it is truncated before the claims section — the text I was given ends mid‑sentence in the detailed description ("The exact type and descri…"). I therefore cannot reproduce the independent claims verbatim from the authoritative source. I searched for the claim text separately and found only a partial fragment (discussed below). I flag this explicitly rather than reconstructing claim language.

Bibliographic data (high confidence)

Independent corroboration: US 7,058,931 (Sun Microsystems; inventor Vinay S. Pai) lists "6658625 | December 2, 2003 | Allen" among its referenced cited U.S. patents — consistent with the inventor and issue date above.

Do not confuse with near‑neighbours: search results surfaced US 6,665,725 (Network Associates/"Packet Intelligence," Dec. 16, 2003) and US 6,658,276 ("Kianl"). These are different patents with different numbers, owners, and subject matter. I am not merging them into this record.

Abstract (as published)

The preferred embodiments of the present invention provide a method and apparatus for generic data conversion. A generic data convertor interprets a data description that has configurable data definitions that can accommodate changes in the data. The data definitions can allow the data type, character set, location, and length of data elements in the data stream or file to be described and easily modified. The data convertor uses the data description to determine how to convert the data and, if necessary, where data elements are in the data. The data convertor is particularly useful for converting data that is sent to and/or received from a server. The data convertor and data description cooperate to support calling multiple releases of the server using the same data description. In addition, the data convertor may also call the server program with the correct, converted parameters in the correct order. The data convertor usually waits until a requesting application asks for particular data elements in the data before converting the data elements.

Plain-language overview of the disclosure

The patent describes replacing "hard‑coded" client/server data‑conversion routines with a generic convertor driven by an external, editable description of the data. Key elements from the specification:

• Data description (preferred embodiment: PCML — Program Call Markup Language, an XML dialect) describes data type, character set, location, length, usage (input/output/input‑output), parameter list, and the order in which parameters are sent ("parse order").
• Parser (optional but preferred) parses the description into a parse tree; the tree's information is then extracted into a hash table, which is serialized to a file so that no XML parser is needed at runtime and memory/resources are saved.
• Data convertor (implemented as a Java ProgramCallDocument class) calls the server program with converted, correctly ordered parameters, and converts returned data on demand.
• Conversion examples: Unicode ↔ EBCDIC, Little Endian ↔ Big Endian, BigDecimal ↔ packed/zoned decimal, etc., delegated to "Java toolbox" classes.
• Varying‑length and nested structures (arrays, structs, arrays of structs) are located via offset/displacement and length fields; "skip" data types handle server boundary padding.
• Just‑in‑time conversion: elements are only located/converted when the application requests them, rather than converting the whole buffer.
• Multi‑release support: the same description can carry version‑ranged definitions so one document serves several server/OS releases.

Regarding the independent claims — status and limits of my information

I cannot give you authoritative, verbatim claim text for US 6,658,625 from the material available to me. What I can report:

• A search‑engine snippet of the patent's own text (from uspto.report/patent/grant/6658625) contains a claim‑style passage referring to "...the data convertor ... the at least one data element to the program or convert the at least one data element ... from the program if the program ... is within the current range of release levels." This is fragmentary and I cannot confirm whether it is claim language or specification language.
• Reading the specification, a claim‑1‑type independent claim would most plausibly be directed to a computer‑implemented method/apparatus in which a data convertor reads a data description that defines data elements (with type/length/location/character‑set and, optionally, version ranges), locates those elements within a data file or stream, and converts them between formats — including the ability to send/receive data to/from a program across a range of server release levels. This is a reasoned inference from the disclosure, not a quotation.
• I do not know how many independent claims the patent has, or their exact numbering/wording. I would need the granted‑claims PDF from USPTO Patent Center/Google Patents to state them accurately.

If you need exact claim language, the reliable next step is to pull the full grant document (claims begin after the detailed description) from USPTO Patent Center (application 09/291,361) or the Google Patents PDF.

Litigation / docket status

• No Court of Appeals for the Federal Circuit 2026 docket activity involving US 6,658,625 was found in my searches. I found no CAFC appeal, no PTAB proceeding, and no district‑court assertion tied to this patent number.
• This is consistent with the patent's record status: Expired – Fee Related, anticipated expiration 2019‑04‑14, roughly seven years before today's date, which makes it an unlikely enforcement vehicle.
• Caveat: I searched via general web search, not by directly querying USPTO Patent Center/PAIR, the PTAB API, or the CAFC docket system. Absence of results is therefore not proof that no docket exists — only that nothing of the kind surfaced in my searches. Treat "no litigation found" as a lead, not a conclusive negative.

Uncertainty summary

• High confidence: patent number, title, inventor (Allen), assignee (IBM), application number, filing/priority date (1999‑04‑14), issue date (2003‑12‑02), status/expiration, abstract, and the technical content of the specification.
• Not verified / unavailable to me: verbatim independent (and dependent) claim text and count; any litigation or CAFC 2026 docket involvement; current assignment beyond the listing shown.
• I did not auto‑correct any identifier: "6658625" is treated throughout as US 6,658,625 B1, and similar‑numbered patents found in search results were deliberately excluded.

I'll research litigation involving US patent 6658625 across patent litigation databases.

Let me refine the search with more targeted queries.

Known Litigation Involving U.S. Patent No. 6,658,625

Result: No litigation was found.

After searching patent-litigation resources (including the materials surfaced through Google Patents/FreePatentsOnline, Justia, Docket Alarm/PTAB filings, Unified Patents-style trackers, and general web searches keyed to the exact number "6658625" and to "6,658,625"), I could not identify any federal district court case, ITC investigation, CAFC appeal, or PTAB proceeding in which U.S. Patent No. 6,658,625 was asserted or challenged.

What I did find (and why it is not litigation of this patent)

• Patent bibliographic record (Google Patents / FreePatentsOnline): U.S. 6,658,625 B1 — "Apparatus and method for generic data conversion," inventor Paul V. Allen, assignee International Business Machines Corporation, application US09/291,361, filed 1999-04-14, granted 2003-12-02, classified in G06F9/54 and G06F16/25/16/258. Legal status: Expired – Fee Related (anticipated expiration 2019-04-14).
• The many hits for "'862 patent," "6,658,625" and similar strings returned different patents (e.g., U.S. 6,683,858 to Paltalk; the Cajun '862 patent in Cajun Servs. Unltd. v. Besco; U.S. 6,418,554 / 6,415,335 in the EpicRealm family), plus unrelated third-party dockets. Per your instruction, I am not treating these as litigation of U.S. 6,658,625.
• The closest "hit" was a secondary reference listing "6658625 — December 2, 2003 — Allen" in a citing-patents table of an unrelated patent (Justia, RE42725). This is a citation, not an assertion of the patent in litigation.

Case table

Important caveats

• Confidence level: Moderate-to-high that no litigation exists, but I cannot state this as an absolute. My sources were web searches and publicly indexed pages; I did not have direct, authenticated access to PACER, Docket Navigator, Lexis CourtLink, or the Unified Patents litigation portal (which are paywalled or not fully crawlable). A dedicated Docket Navigator / Lexis CourtLink / Unified Patents query on the exact patent number would be the definitive check I could not complete here.
• U.S. 6,658,625 is an IBM software/interface-conversion patent that has been expired since 2019. Given that IBM generally does not assert this class of patent in offensive litigation, and the number does not appear in any indexed patent docket, the absence of litigation is consistent with expectations.
• If a case does exist, it may be an older, low-profile matter not captured in the freely indexed databases I could reach.

Bottom line: Based on the available search results, no known litigation involves U.S. Patent No. 6,658,625. I recommend confirming with a direct, authenticated Docket Navigator or PACER docket-number search on the patent number before relying on this as a final negative.

I'll start by verifying the structured data with targeted web searches for any PTAB proceedings involving this patent.

Let me run additional targeted searches to be thorough about any PTAB activity.

Proceedings overview

Total AIA trial proceedings on file for US 6,658,625: 0. The USPTO Open Data Portal structured data returns no IPR, PGR, or CBM proceedings, and independent web searches for "6658625" combined with IPR/PGR/CBM/trial terms surface no Board proceeding naming this patent. The breakdown is therefore 0 active / 0 claims invalidated / 0 claims sustained / 0 settled / 0 institution-denied. The bottom-line defensive posture is unusual and does not fit either of the prompt's archetypes: the patent has never been tested at the PTAB, but it has also expired — Google Patents lists its legal status as "Expired – Fee Related" with an anticipated expiration of 2019-04-14 (20 years from the 1999-04-14 filing date). A defendant today therefore faces no live PTAB risk profile and, more importantly, a dead patent: any assertion must rest on pre-expiration conduct within the § 286 six-year damages window, and there is no PTAB record to mine for invalidity ammunition.

---

There are no proceedings to list. Per the task instructions, "the absence is itself a signal," and here it is a strong one — but not the signal the prompt anticipates.

No PTAB activity on file. I searched the canonical structured block (USTPO ODP) and supplemented with web searches across PTAB docket aggregators (Docket Alarm, PTAB E2E petitions), FreePatentsOnline, Justia, uspto.report, and the Federal Circuit / patent-blog ecosystem. Every hit for this patent number traced back to the patent document itself (bibliographic pages, claim text, prosecution references) — not to a Board proceeding. The IBM-versus-VirtaMove PTAB litigation that dominates search results concerns entirely different IBM patents (e.g., U.S. 7,519,814 in IPR2025-00599), not US 6,658,625. I found no Federal Circuit appeal, no CBM challenge, and no defensive-aggregator (Unified Patents–style) filing.

Why the absence here does not signal "hardened patent"

The prompt's heuristic — "well-asserted patents eventually attract IPRs" — assumes an actively asserted, unexpired patent. That assumption fails here:

• The patent is expired. Filed 1999-04-14; anticipated expiration 2019-04-14; status "Expired – Fee Related." An expired patent cannot be infringed going forward, so there was little commercial incentive for any accused party to spend the several-hundred-thousand dollars an IPR costs.
• It is an IBM utility patent from the 1999 client-server/PCML era (inventor Paul V. Allen; PCML = Program Call Markup Language, an IBM AS/400 technology; classification G06F16/258, G06F9/541). IBM enforces its software portfolio selectively, and this particular patent does not appear to have been asserted in any published litigation that would have drawn a defensive IPR.
• No IPR never means "valid." Zero PTAB proceedings tells you nothing about the strength of claims 1–33 — only that no adversary ever chose to buy a Board ruling. The claims are simply UNTESTED at the PTAB.

---

Strategic summary

Claim status: all 33 claims UNTESTED at the PTAB. No AIA proceeding has ever reached institution, so there is no case or controversy that canceled or sustained any claim of US 6,658,625. Claim 1 (apparatus), claim 10 (apparatus/parser-output), claim 17 (method), claim 26 (method), and claim 31 (program product) — and their dependents through claim 33 — all stand as issued, untouched by any Board FWD. Nothing has been "hardened" by surviving an IPR, and nothing has been "killed" by one.

Estoppel landscape: there is none. Because no IPR/PGR/CBM was ever instituted, § 315(e)(2) estoppel is not triggered by any proceeding against this patent — no petitioner, real party in interest, or privy is barred from raising anything. For any defendant, every § 102/§ 103 prior-art ground remains available, both in district court and in a (hypothetical, now time-barred) PTAB petition. There is no prior petitioner's ground to inherit or avoid, and no risk that a would-be IPR ground was "reasonably raisable" and is now foreclosed. The full prior-art field is open.

Pattern signals: none. No serial petitioner (no petitioner at all), no patent-owner PTAB appeal activity, no defensive aggregator in the chain. The only adjacent data point is that IBM and VirtaMove are litigating in the PTAB over other IBM patents — but that is a different patent family and does not touch US 6,658,625. The dominant, verifiable fact about US 6,658,625 is that it expired on 2019-04-14 for failure to pay maintenance fees (Google Patents legal status: "Expired – Fee Related").

---

Recommended next steps

• If you are a defendant facing an assertion of US 6,658,625, the first move is a standing/expiration check, not an IPR. Confirm the maintenance-fee lapse and the 2019‑04‑14 expiration in USPTO Patent Center. If the patent is expired, it cannot be infringed for post-expiration conduct; the plaintiff's recovery would be limited to pre-expiration infringement within 35 U.S.C. § 286's six-year lookback (i.e., conduct after approximately 2019-10-01 less six years). Ask for the accused-conduct dates and the demand's damages theory.
• There is no Board record to deploy. You cannot cite a PTAB FWD canceling any claim — none exists. Do not represent otherwise in a demand response or a motion; the absence of a proceeding is a neutral fact about validity and a dispositive fact about expiration.
• Validate the prior-art field yourself. Because no estoppel has attached, a full invalidity search is unobstructed. The prosecution citations (INPADOC family lists US 5,261,077; US 5,694,580; US 5,793,381; US 2001/051974; US 2003/033317, among ~25 references) are a starting point, but note that the patent's 1999-04-14 priority date fixes the critical date for § 102/§ 103 art.
• No trial-stage milestones to track — there being no pending proceeding, there is no institution deadline, oral hearing, or statutory one-year FWD date to monitor.
• Caveat on completeness. The ODP ingest is the canonical source and shows zero; my web searches also found zero. If a very recently filed petition or an unindexed older proceeding exists, I did not find it, and I have not invented a proceeding number to fill the gap. For belt-and-suspenders confirmation, a party with PTAB E2E access can run a direct patent-number query at the PTAB E2E portal and cross-check the PTAB's public decisions page.

Sources consulted: USPTO ODP structured "PTAB proceedings on file" block (provided); Google Patents US6658625B1 (bibliographic/legal-status data, expiration); FreePatentsOnline 6658625 (claim text); Espacenet bibliographic & INPADOC family; PTAB petition documents on ptacts.uspto.gov (none referencing US 6,658,625); Docket Alarm IBM v. VirtaMove, IPR2025-00599 (different patent, U.S. 7,519,814).

I'll research the assignment history for US 6658625. Let me start with several searches.

Let me look for assignment/legal-events records directly.

Let me check Google Patents legal events and any post-issuance transfers.

I was able to pull the bibliographic, prosecution, and legal-events picture, but my searches did not surface any post-issuance assignment records, and I could not open the USPTO Assignment Center query directly (the tooling available is web search, not live Assignment Center access). Findings below are therefore grounded in the Google Patents legal-events/assignment data and the patent front page; the reel/frame numbers for the one recorded assignment are flagged as unverified rather than invented.

Inventors

• Paul V. Allen — sole named inventor. No co-inventors (confirmed across Espacenet, Google Patents, and Patent Leaderboard, which lists Allen with 8 IBM-assigned US patents including US 6,658,625). Employer at time of filing: International Business Machines Corporation (IBM) — the application was assigned to IBM on the filing date (1999-04-14), and the patent's attorney of record is an IBM-region attorney (Jack P. Friedman, Latham, NY), consistent with an in-house IBM filing rather than an independent inventor.
• No unusual pattern detected: there is only one inventor, so "all inventors departing the original assignee within 12 months" is not applicable, and no inventor departure evidence surfaced.

Original assignee

International Business Machines Corporation (IBM) — named on the issued patent (US 6,658,625 B1, granted 2003-12-02), New York corporation, Armonk, NY.

• Primary line of business: enterprise computing, software, and IT services. IBM is a large operating company, not a holding/licensing entity.
• Product embodiment: The patent claims a generic data convertor driven by a PCML (Program Call Markup Language) data description in a Java/XML client–server environment — i.e., technology that maps directly onto IBM's AS/400 (later IBM i) program-call tooling and its Java Toolbox for AS/400. IBM shipped this class of product (IBM Toolbox for Java / PCML), so the claimed functionality was implemented in IBM's commercial stack.
• Current status: Operating. IBM has not exited the software business and remains the assignee of record; the patent itself lapsed (Expired – Fee Related, anticipated expiration 2019-04-14 per Google Patents), meaning IBM stopped paying maintenance fees around the end of the full 20-year term.

Assignment timeline

• 1999-04-14 (executed) / recorded 1999-04-14 — Reel/frame not retrievable from the sources available to me (Google Patents reports the event as "Assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION … ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS); Assignors: ALLEN, PAUL V." without exposing the reel/frame string in the fetched record)
  • Conveyance: Assignment
  • Assignor: Allen, Paul V. (sole inventor)
  • Assignee: International Business Machines Corporation
  • Correspondent: Not verified. The prosecution attorney of record for the patent is Jack P. Friedman (Latham, NY) per the published front page; this is the prosecuting/agent-of-record, not necessarily the recording correspondent. No repeat-correspondent pattern can be established from what I could retrieve — a single appearance would not be a finding anyway.
  • Context: Original inventor-to-employer assignment (standard pre-AIA filing-date assignment of all right, title, and interest).

No further recorded assignments were found. Google Patents' legal-events/reassignment listing for US 6,658,625 shows only the 1999 filing-date assignment to IBM and the statutory milestones (filing 1999-04-14, grant 2003-12-02, anticipated expiration 2019-04-14, status "Expired – Fee Related"). My searches for any transfer to Acacia, Marathon, IV, Wi-LAN/Conversant, or any LLC-style assignee returned nothing tied to this patent number.

Timeline diagram

timeline
    title Ownership of US 6658625
    1999 : Filed by inventor Paul V Allen
         : Assigned to IBM on filing date
    2003 : Patent issued as US 6658625
    2019 : Term ended fee related

NPE / troll-pattern signals

1. Shell-entity transfer — Not present. The only recorded assignment moves the patent from the sole inventor to IBM (a large operating company), not to any IP/Holdings/Licensing/Ventures LLC. No single-purpose entity appears anywhere in the chain.

2. Known asserter in the chain — Not present. No assignee in the chain matches Acacia, Marathon Patent Group, Intellectual Ventures, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, or any Spangenberg entity. The all-time assignees are the individual inventor and IBM.

3. Repeat correspondent across the chain — Not present. There is only one recorded assignment, so recurrence cannot be established. Jack P. Friedman (Latham, NY) appears once, as prosecuting attorney of record — an isolated appearance is explicitly not a finding.

4. Cascading transfers — Not present. Zero post-issuance transfers; no chained LLCs and no <24-month sequence exists.

5. Pre-litigation transfer — Not present. No infringement suit naming US 6,658,625 surfaced in any search, so there is no pre-suit transfer to time against.

6. Bankruptcy fire-sale — Not present. IBM did not sell this patent in a bankruptcy proceeding; IBM is solvent and the patent simply reached (and exceeded) its statutory term.

7. Privateering — Not present. No operating-company-to-NPE transfer, and no SEC/EFF/Patent Progress reporting linking this patent to a proxy assertion program.

8. Defensive aggregator (anti-NPE) — Not present as an assignment event (the chain does not terminate at RPX/AST/LOT/Unified/OIN), but the practical effect is similar: the patent ended its life fee-lapsed with no assertion activity.

Verdict

Insufficient data — but with a directional read: the evidence points to an operating-company-owned patent, not an NPE asset. The only recorded assignment is the 1999-04-14 inventor-to-IBM filing-date assignment; Google Patents' legal events for US 6,658,625 show no subsequent conveyance, the assignee of record is International Business Machines Corporation, and the patent lapsed "Expired – Fee Related" (anticipated expiration 2019-04-14) with no litigation found. Because the chain contains only the original assignment and I could not verify reel/frame numbers or the recording correspondent, this falls under the "only the original assignment" branch rather than a positive NPE finding.

Verification link (USPTO Assignment Center): https://assignmentcenter.uspto.gov/ — search patent number 6658625 (mirror: https://assignment.uspto.gov/patent/index.html).

Caveats / data gaps (stated explicitly rather than inferred): (a) I could not retrieve the reel/frame for the single IBM assignment or confirm whether any additional assignment is recorded at USPTO that Google Patents has not surfaced — this should be confirmed directly in Assignment Center; (b) I found no assignment record for this patent in the searches performed, and the absence of post-issuance records is itself the finding: IBM still held the patent, and the record shows no NPE involvement.

I'll research the prior-art citations for US 6,658,625 (patent number 6658625). Let me run several targeted searches.

Let me dig into the specific "References Cited" list and the claims structure.

I hit the tool-call limit mid-research, so I can only partially complete the reference-by-reference § 102 analysis you asked for. Below I separate what the searches verified from what I could not retrieve, and I explicitly flag every place where I'd otherwise be guessing. No patent number or identifier has been auto-corrected; "6658625" is treated throughout as US 6,658,625 B1.

---

1. Confirmation that 6658625 = US 6,658,625 B1 (search-verified)

Multiple independent sources returned the same record, and no similar-numbered patents (e.g., US 6,665,725, US 6,658,276) were merged in:

Sources: Espacenet biblio (worldwide.espacenet.com/publicationDetails/biblio?CC=US&NR=6658625B1) and the INPADOC family page for the same number.

---

2. Claim structure — partially recovered (correcting the earlier "claims unavailable" gap)

The earlier section flagged that the authoritative text I was given was truncated before the claims. A search did surface claim text from Espacenet's claims view for US6658625B1. It is fragmentary (the scraper interleaved claim numbers), but enough to establish the independent-claim set:

Confirmed independent claims (3):

• Claim 1 — apparatus: "at least one processor; a memory coupled to the at least one processor; a data description residing in the memory, the data description describing a current data type of at least one data element … wherein the data description is configurable to accommodate changes from the current data type … to a different, application-compatible, data type;" plus a data convertor.
• Claim 26 — method: steps a)–g) including "the data convertor converting only each input data element … using a single conversion"; calling the program "with the correct input data … in the correct order," using the parameter list and parse order from the parser output; then, on application request, "locating the at least one output data element in a data output by the program" and converting "only the at least one output data element … using a single conversion."
• Claim 31 — program product: a data convertor "responsive to a specific immediate request from an application program for at least one particular data element," plus "signal bearing media bearing the data convertor."

Dependent claims visible in the fragment (numbers may be off by the scraper artifact):

• length configurable (appears as cl. 2 / cl. 27)
• character-set identifier configurable (appears as cl. 3)
• range of release levels — convert/send only if the program's release level is within the range (appears as cl. 6 / cl. 29)
• location of output element configurable (cl. 28)
• hash table created from parser output (cl. 30)
• signal bearing media = transmission media (cl. 32); claim 33+ cut off

The phrase "105." preceding claim 6 is a page/scraper artifact, not a claim number.

Limit: I cannot state the exact total claim count or reproduce claims verbatim from the authoritative full text; the fragment is from a third-party display page. Treat the above as high-confidence on structure but not verbatim.

---

3. Prior art / "References Cited" — INCOMPLETE (this is the core of your request)

The only citation list my searches retrieved is the Espacenet INPADOC "Citations" list for US6658625 (B1), which returned:

US2001051974 A1, US2003033317 A1, US5261077 A, US5694580 A, US5793381 A (+20)

i.e. only 5 of ~25 cited documents were exposed; the remaining 20 are behind the "(+20)" truncation.

What this means for the § 102 analysis you asked for

I cannot deliver a reference-by-reference "which claims it anticipates under 35 U.S.C. § 102" mapping, because:

1. I retrieved only 5 of ~25 references — 80% of the list is missing.
2. I did not retrieve the full text of even those 5, so I cannot responsibly state what each discloses or which claim elements it meets.

I will not fabricate titles, abstracts, or anticipation conclusions for these references. What I can state with confidence:

⚠️ Contradiction / anomaly I must flag explicitly

Both US 2001/0051974 A1 and US 2003/0033317 A1 have publication dates after the 1999-04-14 priority date. That is inconsistent with them being prior art cited against 6658625. Two readings:

• They are actually forward citations (later documents that cite US 6,658,625), and Espacenet's INPADOC "Citations:" field is mixing cited and citing documents in that display; or
• They are § 102(e) art, valid only if their effective filing dates precede 1999-04-14 (which, given 2001–2003 publication, is doubtful).

Because publication of an application normally trails its filing by ~18 months, both scenarios point to these two being post-dated relative to the 1999 priority and therefore probably not anticipatory prior art. I could not resolve which interpretation is correct, so I flag it rather than assert one.

Also retrieved — forward citations (NOT prior art)

Later patents that cite US6658625B1 in their own reference lists (context only; these are not prior art against 6658625):

• US 2018/0232427 A1 — "Data structure endian conversion system"
• US 2010/0250459 — "Custodian management system"
• EP 1 597 653 A2
• US 7,058,931 (Sun Microsystems; Pai) also lists "6658625 | Dec. 2, 2003 | Allen" among cited U.S. patents.

---

4. Provisional § 102 framework (for when the full list is obtained)

Given the claim elements recovered in §2, an anticipation analysis would need each independent claim evaluated against:

• Claim 1: processor + memory + configurable data description of a data element's current data type + convertor that reads it.
• Claim 26: the full method chain — external description (XML/PCML + DTD) → parser output with parameter list and parse order → just-in-time, single-conversion of only requested output elements.
• Claim 31: program product with signal bearing media (the classic Beauregard/§ 101 "program product" format).

The two elements most likely to distinguish over 1990s conversion art are (a) the external, editable description driving a generic convertor, and (b) the "convert only on request / single conversion / parse-order" limitations of claim 26. This is a reasoned observation from the claim text recovered — not an anticipation finding, which requires the missing references.

---

5. How to close the gap (reliable next steps)

Because my search was cut off, the complete and authoritative reference list should be pulled directly:

1. USPTO Patent Center / Patent Application 09/291,361 — the "References Cited" (PTO-892) and full claims.
2. Google Patents — US6658625B1 → "References Cited" section (US patents, foreign documents, and "Other References" / non-patent literature). The authoritative text I was given was truncated before this section, which is why it isn't in my summary.
3. Espacenet — the biblio page's "Cited documents:" (as opposed to the INPADOC family "Citations" field) to expand the "(+20)" set.

Once that list is in hand, I can produce the requested table (full citation, publication/filing date, brief description, anticipated claims under § 102) for each reference.

---

6. Uncertainty summary

• High confidence: patent identity/bibliography; existence of independent claims 1 (apparatus), 26 (method), 31 (program product) and the dependent-claim subject matter listed in §2 (from recovered Espacenet claim text); that US5261077, US5694580, US5793381, US2001/0051974 A1 and US2003/0033317 A1 appear on a citation list associated with US6658625 in Espacenet.
• Not verified / could not obtain: titles, abstracts, and disclosures of those five references; the other ~20 cited documents; the definitive total claim count and verbatim claim text; whether the two US published applications are cited (prior art) or citing (forward) references.
• No fabrication: I have not supplied descriptions or § 102 anticipation conclusions for references whose text I did not retrieve.

I'll start by looking for the prior art that's actually of record — both the references cited on the face of US 6,658,625 and independent prior art in the generic-data-conversion space. Note up front: the user message references "the results from the Prior Art section of this page," but no such section was actually included in what I received — only the patent summary and litigation summary. I'll say so explicitly and go get the prior art myself rather than invent it.

Obviousness Analysis — U.S. Patent No. 6,658,625 (35 U.S.C. § 103)

0. Scope and a required correction

Two things before the analysis, both of which materially affect reliability:

1. There is no "Prior Art section" in the materials I was given. The user message contains only the previously generated Patent summary and Litigation summary sections. Neither contains a set of prior-art search results. I therefore researched the prior art myself rather than pretend a provided section exists. If a curated Prior Art section exists on your end, it was not transmitted to me and my results may not match it.

2. The granted claims remain only partially available. The authoritative full text in the earlier section is truncated before the claims. From the Espacenet claims page for US6658625B1 I recovered fragments of independent claims 1 (apparatus), 17/26 (method), and 31/42 (program product), plus several dependent claims. I use those fragments but treat exact wording as unverified. This matters because § 103 is decided claim-by-claim on the claim language.

   Source for claim fragments: https://worldwide.espacenet.com/publicationDetails/claims?CC=US&NR=[6658625B1](/patent/6658625B1)&KC=B1&FT=D&ND=&date=20031202&DB=&locale= ; source for the cited-references list and some claim text: https://patents.justia.com/patent/6658625

---

1. Governing legal framework

• Because the application was filed 1999‑04‑14, the pre‑AIA version of § 103 applies, with pre‑AIA § 102(a)/(b)/(e) defining what qualifies as prior art (including the 102(e) effective-filing-date rule for U.S. patents).
• The controlling test is Graham v. John Deere Co., 383 U.S. 1 (1966) (scope/content of prior art; differences; level of ordinary skill; secondary considerations), as refined by KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007).
• Under KSR, a combination is obvious if a POSITA would have had a reason to combine the references with a reasonable expectation of success — including when the combination is "the product … of common sense," mere "substitution of one known element for another," or "a predictable variation." Teaching away, unexpected results, and secondary considerations cut the other way.

Level of ordinary skill (POSITA) — proposed: a person with a bachelor's degree in computer science/electrical engineering (or equivalent) and about 2–4 years of experience in client‑server data communications, binary/character data-format conversion, and object-oriented programming, with working familiarity with SGML/XML and formal interface descriptions. This is the level implied by the specification, which repeatedly describes its individual techniques (hash tables, serialization, parse trees, offset handling) as "well known in the art."

---

2. The invention to be assessed (decomposed from the available claim text and spec)

---

3. Prior art identified

3a. References cited on the face of US 6,658,625 (prior art of record)

From https://patents.justia.com/patent/6658625 (Justia "Referenced Cited"), corroborated by the Espacenet INPADOC citation list:

Non‑patent literature of record (Justia; FreePatentsOnline https://www.freepatentsonline.com/6658625.html):

• M. Edwards, XML: Data the Way You Want It (Microsoft, Oct. 31, 1997)
• Microsoft Corp., XML: A Technical Perspective (Feb. 20, 1998)
• Enabling Next-Generation Web Applications (Apr. 3, 1998)
• C. Heinemann, Internet Explorer 5 and XML (Microsoft, Nov. 4, 1998)

Verification caveat. I have confirmed these citations exist as references of record but have not independently read the disclosures of most of them. I will not assert what individual ones teach beyond their titles/characterizations. The dates flag one issue: the two published applications (Saad 2001; Ziglin 2003) and several late-issued patents must rely on earlier effective filing dates to be § 102 prior art to a 1999‑04‑14 filing; that needs checking before any given one is relied on.

3b. Independent art I located (with disclosure text I actually saw)

The US 5,421,001 + US 5,793,954 + Majix combination was deemed by a European examiner to render an entire other "general-purpose data parsing/format conversion" application non‑inventive. That is strong signalling that this space — generic, description‑driven conversion — was crowded and largely predictable by the late 1990s.

---

4. Combination analyses

Combination A — Methe '001 + Beringer/'434 (core generic-conversion engine)

Teaches/maps:

• A1, A2 — Methe's "programmably configurable protocol descriptions" driving a control module for "format conversion" = a data convertor driven by an external, configurable description of data type/format. Methe's system is expressly general‑purpose.
• A3, A4 — '434's field-descriptor records define "position of the first and last bits" and "type of data," used by a node manager to locate and interpret fields = length/location + type from a description.
• A7 — on-demand conversion is a routine lazy-evaluation/access optimization (see §5).

Motivation to combine: Methe is explicitly directed to generality and reconfigurability; '434 supplies the concrete mechanism (an external field-descriptor tree defining type/position) that Methe's configurable-protocol-description framework calls for. Both are in the same field (data format translation), address the same problem (avoiding bespoke transcoding for each new format), and combining them yields nothing more than the predictable sum of their parts — the KSR "predictable variation" scenario. A POSITA would expect success because '434 shows the descriptor-tree approach already worked for heterogeneous platforms.

Result: claim 1 (and dependent type/length/location claims; cl. 2–5, 46–47) obvious.

---

Combination B — Methe '001 + Ogawa '874 + XML/DTD NPL (description-driven convertor + data type/length + XML/DTD + serialization/hash)

Teaches/maps:

• A1–A4 — Ogawa's resource DB describing "fields, size of a given field, and type of data" that a single generic PreProcessor reads is, functionally, the claimed "data description" with type + length; the graphical editor to modify the resource DB maps directly to the claimed configurability element (A2) — change the description, not the code.
• A5 — character-set handling: the Gelfenbain patents of record (US 5,831,560 and US 5,898,874, "Method for converting a character set") are directed to exactly this; the CCSID/"convert characters between code pages" limitation is squarely met by them.
• A9, A10 — XML + DTD as a self-describing data grammar was, by the 1999 priority date, ordinary skill — which is the patent's own admission via its cited NPL (Edwards 1997; Microsoft XML: A Technical Perspective 1998; Heinemann 1998) and its citation of XML-oriented US 6,226,675 (Meltzer). Defining a custom DTD (PCML) for program‑call interfaces is application of the known XML/DTD toolchain. Building a hash table from a parse tree for O(1) lookup, and serializing it, are undisputed "well known" Java techniques (the spec itself says so: "hash table, which is a well known JAVA object"; "Serialization is a process, well known to those skilled in the art").

Motivation to combine: XML was a rapidly adopted standard explicitly marketed (in the very references the patent cites) as a way to describe data content and share format independently of the application — the precise purpose the PCML description serves. A POSITA designing a configurable data description in 1998–99 would look to XML because it was the field's recognized answer for portable, machine-readable descriptions, and would use a DTD (the then-standard XML grammar mechanism) to validate it. Hash-table indexing and serialization are ordinary engineering choices to improve runtime performance — the KSR "known technique to improve similar devices in the same way" rationale.

Result: claims 39–42, 45, 48 (parser output / XML+DTD+PCML / hash table / signal media) obvious.

---

Combination C — '434 (or Ogawa '874) + Tandem WO 98/02808 + RPC/marshalling skill (interface description + ordered call + program product)

Teaches/maps:

• A8 — WO 98/02808 discloses describing an "interface, operation or data type" and generating generic functions to transport object calls across heterogeneous systems — i.e., generic marshalling of an interface call from a machine-readable interface description. The claimed "parameter list + parse order" is just ordered argument marshalling to a remote program, a core RPC concept (the patent itself is classified under G06F 9/54 / 9/547, "Remote procedure calls").
• A11 — a "program product" comprising convertor + "signal bearing media" is a generic article limitation; the specification concedes that such media and distribution forms are conventional. If the underlying method is obvious, the Beauregard‑style/“signal media” claim adds nothing patentable absent some special-media contribution.

Motivation to combine: Both '434/Ogawa and Tandem address cross-platform transport/interconversion of structured data; Tandem supplies the explicit "describe the interface … generate generic call-transport functions" teaching that complements the descriptor-driven conversion of '434/Ogawa. Routing a described call in a defined argument order is the routine implementation of such a system.

Result: claims 17/26 and 42–44 obvious.

---

5. The JIT/lazy-conversion limitation (A7) — obvious

Element A7 ("convert only the … element," "when an application requests the element") is textbook lazy evaluation. It is a design choice selectable from a small, known set (eager vs. lazy), with the result — fewer conversions for unused data, faster perceived performance — being exactly what one expects and nothing more. The specification's own stated advantage ("unused or undesired data will not be converted") is the inherent, predictable benefit of lazy evaluation. Under KSR, "if a technique has been used to improve one device, and a person of ordinary skill … would recognize that it would improve similar devices in the same way, using the technique is obvious." Applies here.

---

6. Claim-level bottom line

---

7. Where the record is genuinely thinner — and why I won't overstate it

I will be explicit: the feature that most resists a § 103 attack on the art actually before me is A6 — the version‑ranged data definitions (minvrm/maxvrm) enabling one description to serve multiple server/OS release levels (claims 6, 29, 38). None of the references whose disclosure I verified (Methe, '434, Ogawa, Tandem, Majix) is shown to teach gating field definitions by an OS/program release range. Nor have I verified that any of the 25 references of record does so.

Two honest possibilities:

• It may still be obvious by analogy to well‑known conditional/versioned configuration, backward/forward‑compatibility schema evolution, and "feature availability by version" conventions. A POSITA maintaining software across OS/400 releases (VxRyMz, as the patent uses) plausibly would encode release ranges in a description. But this is an argument I can construct, not one I can ground in a specific prior‑art passage today. Flagging it as an unverified hypothesis is the correct posture.
• It may have been the actual point of novelty, and the patent may owe its grant to it.

Because I could not read every reference of record, I may well be understating the available art. This is a limitation of my search, not evidence that no such reference exists.

---

8. Secondary considerations (§ 103 fourth Graham factor)

• No commercial-success or industry-praise evidence is in the record I have. The earlier litigation summary found no assertion, no CAFC/PTAB activity, and a status of Expired – Fee Related (2019‑04‑14). Failure to enforce over ~16 years, and expiry, weakens any nexus‑based secondary-consideration argument.
• No unexpected-results evidence appears in the specification for the ordinary elements; the stated benefits (speed, less memory, JIT conversion, cross‑release reuse) are either inherent to the chosen techniques or asserted without comparative data.
• Copying/industry-licensing indicators are absent.

These factors cut toward obviousness (or at least do not rebut it) for the broad independent claims.

---

9. Caveats & recommended verification steps

1. Get the granted claims verbatim. Pull the grant PDF (USPTO Patent Center, application 09/291,361; or the Google Patents PDF https://patentimages.storage.googleapis.com/b9/0c/13/211da3dc83fbd7/US6658625.pdf). My claim mapping above is reconstructed from fragments and is not authoritative. Claim numbering in particular (e.g., whether the program‑product independent claim is 31, 37, or 42) must be checked.
2. Verify the disclosure of each reference of record. I confirmed citations/dates but read the substantive teachings of only a handful. In particular, check US 5,421,001, US 5,433,? / 5,339,434, US 5,608,874, WO 98/02808, and especially the Meltzer, Morgenstern, Davidson, Ahlers, Pallmann references — a schema-/DTD-driven configuration reference may directly attack the minvrm/maxvrm feature.
3. Confirm prior-art dates. For the published applications US 2001/0051974 (Saad) and US 2003/0033317 (Ziglin) and the late-issued patents, confirm pre‑1999‑04‑14 effective filing dates before relying on them; otherwise they are not prior art and cannot support § 103.
4. Confirm § 103's analytical baseline: pre‑AIA § 103 (filing 1999) — do not apply AIA § 102/103.
5. I did not auto-correct any identifier. "6658625" is treated throughout as US 6,658,625 B1; near‑numbered patents (e.g., US 6,665,725; US 6,658,276; US 6,683,858; US 6,051,088 etc.) were deliberately excluded and are not merged into this record.

Bottom line: On the art I could verify, the broad independent claims of US 6,658,625 — a description‑driven, generic data convertor with typed/length/location field definitions, XML/DTD parsing into a hash table, on‑demand conversion, and ordered program calls — read as an obvious combination of well‑known description‑driven translation systems (Methe '001; Beringer '434; Ogawa '874; Tandem WO 98/02808) with the then‑standard XML/DTD and Java tooling. The feature I cannot presently render obvious — and which may be the true point of novelty — is the release‑range‑gated data definition (minvrm/maxvrm) of claims 6/29/38. That determination, and the exact claim language, require the full grant and the full disclosures of the references of record, which the transmitted materials did not include.