US 10320981

Let me search more specifically for litigation and CAFC activity tied to this exact patent number.

Let me check for any Federal Circuit appeal arising from the Delaware case and the current status of that litigation.

US Patent 10,320,981 (US10320981B2) — Summary

Core bibliographic data

• Title: Personal voice-based information retrieval system
• Patent number: US10320981B2 (8-digit form: 10,320,981)
• Assignee: Parus Holdings, Inc. (Bannockburn, IL)
• Inventor: Alexander Kurganov (Buffalo Grove, IL)
• Application: No. 15/707,951, filed 2017-09-18 (continuation)
• Issue date: 2019-06-11
• Priority date: 2000-02-04 (Provisional App. 60/180,343)
• Legal status (per Google Patents/USPTO record): Expired – Fee Related; anticipated expiration listed as 2021-02-06
• Continuity chain: 09/777,406 → US7516190B2 → 11/711,773 (abandoned) → 12/787,801 → US9377992B2 → 15/193,517 → US9769314B2 → 15/707,951 → US10320981B2; child application 16/436,764 (US20190297189A1, abandoned). Source: Google Patents record for US10320981B2.

Abstract

"The present invention relates to a system for retrieving information from a network such as the Internet. A user creates a user-defined record in a database that identifies an information source, such as a web site, containing information of interest to the user. This record identifies the location of the information source and also contains a recognition grammar based upon a speech command assigned by the user. Upon receiving the speech command from the user that is described within the recognition grammar, a network interface system accesses the information source and retrieves the information requested by the user."

Plain-language overview of the independent claims

The patent has 28 claims; the independent claims are 1, 5, 10, 19, and 24.

• Claim 1 (method): A speech-recognition engine receives a user's spoken command over a network via an IVR application offering a "user-defined search." The engine selects a user-assigned recognition grammar tied to a web request and a website URL. A media server retrieves a stored information-source-retrieval instruction for that grammar; a web-browsing server then accesses the site, executes a content extractor that uses a content-descriptor file to locate and separate only the relevant portion of a web page (using a named object), retrieves only that portion, converts it to audio via a speech-synthesis engine, and transmits the audio message to the user's voice-enabled device.

• Claim 5 (voice-browsing system): The system-side counterpart of claim 1 — a speech-recognition engine coupled to a media server (which initiates a voice-response application on connection), a database storing retrieval instructions keyed to the established grammar, a web-browsing server that accesses only the relevant portion of a web page using a content extractor and content-descriptor file with a named object, and a speech-synthesis engine that converts the retrieved information to audio for the voice-enabled device.

• Claim 10 (method of selective retrieval): Identifies one of a plurality of speech commands in a speech-recognition lexicon from audio of the user's spoken words; uses the identified command to access a corresponding descriptor file that identifies (i) a web-accessible information source and (ii) request information; fetches response data containing a named object with content; extracts the content using the named object; generates audio response data with a message responsive to the command and based on the extracted content; and directs playback of that audio on the user's electronic-communication device.

• Claim 19 (apparatus): A system with a transceiver, a database of descriptor files (each identifying a web-accessible source and request info), a speech-recognition engine that identifies a speech command from audio, a media server that accesses the descriptor file, a web-browsing server that retrieves responsive data and extracts content by the name of a named object, and a synthesizer that generates/transmits audio response data — programmed to direct playback on the user's device.

• Claim 24 (apparatus, user-device side): An electronic-communication device having a microphone and speaker, in communication via a network with a remote computer system (comprising a speech-recognition engine, media server, web-browsing server using a named object to extract content, and speech synthesizer) that is programmed to direct playback of the audio response on the device's speaker.

Litigation / CAFC status (search results)

• The patent is asserted in Parus Holdings Inc. v. Amazon.com, Inc., 1:23-cv-00190 (D. Del., filed Feb. 17, 2023) (Judge Gregory B. Williams), where the complaint reports U.S. Patent Nos. 7,516,190 B2; 9,377,992 B2; and 10,320,981 B2 — the last being this patent. The infringement allegations target Amazon's Alexa/Alexa Voice Service ("skills" such as Tide Pool retrieving portions of a website and converting content to speech). Sources: Justia docket 1:2023cv00190; Stanford NPE Litigation Database (patent 10320981); docket Filing 3 (Report to Commissioner).
• CAFC 2026: I found no Federal Circuit appeal specifically directed to patent 10320981 in my searches. The CAFC 2026 matter that surfaced — Google LLC v. Parus Holdings, Inc., Appeal No. 2024-2381 (oral argument audio posted; CAFC vacated a PTAB decision per IPWatchdog, July 2026) — concerns a different Parus patent family (U.S. 6,721,705 and descendants 7,881,941 and 8,185,402), not US10320981.

Uncertainty notes

• I could not independently confirm a CAFC 2026 docket entry naming US10320981; none was located in the searches performed. The most recent district-court docket data I retrieved for 1:23-cv-00190 dates to March 2023; the current disposition of the Delaware case (e.g., any later judgment, settlement, or appeal) was not confirmed.
• The "Expired – Fee Related" status and 2021-02-06 anticipated expiration come from the Google Patents record and may reflect non-payment of maintenance fees/terminal-disclaimer effects; verify against USPTO PAIR/PatentCenter for current maintenance-fee status.

Based on my searches, here is what is known about litigation involving US Patent 10,320,981 B2 ("Personal voice-based information retrieval system," assigned to Parus Holdings, Inc.):

Known litigation

1. Parus Holdings Inc. v. Amazon.com, Inc.

• Plaintiff: Parus Holdings Inc.
• Defendant: Amazon.com, Inc.
• Jurisdiction: U.S. District Court for the District of Delaware (D. Del.)
• Case number: 1:23-cv-00190 (also cited as 1:2023cv00190)
• Filing date: February 17, 2023
• Patents asserted: The complaint (D.I. 1, filed 2/17/2023) expressly identified U.S. Patent Nos. 7,516,190 B2; 9,377,992 B2; and 10,320,981 B2 — so US 10,320,981 was one of three patents asserted in this case. (Sources: Justia docket for 1:2023cv00190; Ex Parte case database; Stanford NPE Litigation Database.)
• Assignments/judges: Assigned to Judge Gregory B. Williams; Magistrate Judge Christopher J. Burke also listed on the docket.
• Status: Closed. Both the Stanford NPE Litigation Database and the Ex Parte case database list the case status as "Closed." I was not able to confirm the precise terminal disposition (e.g., stipulated dismissal, settlement, summary judgment, or trial outcome) from the sources retrieved, so I cannot state with confidence how it resolved beyond the closed status. A docket search on PACER or the Delaware District Court would be needed for the final order.

Related matter — not litigation of 10,320,981 (excluded for clarity)

• Google LLC v. Parus Holdings, Inc., No. 24-2381 (Fed. Cir.) — This is a Federal Circuit appeal (filed September 30, 2024; argued June 5, 2026; opinion issued July 13, 2026) of a PTAB IPR decision. However, the CAFC opinion and the claims at issue (polling mechanism, website rank numbers, "unexpected response") concern a different family member — the '705 patent (per the opinion's discussion of the "′705 patent," col. 19–20) — not US 10,320,981. I therefore did not count it as litigation of 10,320,981.

Caveats

• The Google Patents page for 10,320,981 flags the family as having litigation and links to a Delaware District Court case (1:23-cv-00190) and a Darts-ip family record — consistent with the Amazon case above.
• I found no other district-court cases specifically identifying 10,320,981 in the sources searched. The JDJournal article about Mintz's fee suit against Parus references Parus's broader enforcement campaign (including suits against Apple, Google, Samsung, LG), but those cases appear to involve other patents in the family and were not shown to include 10,320,981; I did not verify any specific case numbers for them.
• To confirm the final outcome of the Amazon case and to rule out any additional cases, a PACER query on the Delaware docket 1:23-cv-00190 and a search of the PTAB (IPR/PGR) records for 10,320,981 would be the next steps.

The first search surfaced a Docket Alarm link to IPR2022-00805 (Google LLC v. Parus Holdings) — but that petition cites US 9,769,314 (a sibling in the same family). I need to verify whether any AIA proceeding specifically names US 10,320,981, and map the family-wide PTAB activity. Let me dig deeper.

Excellent — the family-wide picture is emerging. There's a Google IPR (IPR2022-00805) against the sibling '314 patent, an Apple IPR (IPR2022-00948), and Google IPRs where the Board denied institution on Parus' U.S. 7,516,190 (the direct parent of '981). I need to pin down which patents each proceeding targeted, and specifically whether any proceeding names 10,320,981 itself.

I've hit my research-step limit, so I'm answering from the structured ODP data plus the web-verified family record. Here is what the evidence supports.

Proceedings overview

There are zero AIA trial proceedings (IPR / PGR / CBM) on file for US 10,320,981 itself — the USPTO ODP list is empty, and web search surfaced no petition naming the '981 patent. The default stands: no PTAB activity on this patent. That said, the search did surface heavy PTAB activity on the '981 patent's continuation sibling (US 9,769,314, which was fully invalidated, all claims 1–26) and institution denials on its direct parent (US 7,516,190). Bottom line for a defendant: the '981 patent has never been through an IPR and no claim of it has been canceled by the PTAB, but its substantively identical sibling has a complete invalidation roadmap sitting in a public Final Written Decision, and the '981 patent itself expired on 2021-02-06 (per Google Patents legal-status data) — so exposure is damages-only and the "troll's" strongest claims are the very ones a PTAB panel already found unpatentable in the sibling.

Family-adjacent PTAB proceedings (NOT on US 10,320,981 — flagged per instructions)

IPR2022-00805 — Google LLC v. Parus Holdings, Inc. (U.S. 9,769,314 B2 — continuation sibling of '981)

• Type: Inter Partes Review
• Filed: 2022-04-04
• Status: Final Written Decision — all challenged claims (1–26) held unpatentable (DocketAlarm: "Final Written Decision Determining All Challenged Claims Unpatentable 35 U.S.C. § 318(a)"; Law360 headline: "Google Wins PTAB Invalidation On Voice-Command Patent")
• Judge panel: David C. McKone, Patrick M. Boucher, James J. Mayberry — Boucher authoring (per FWD)
• Petition grounds: Claims 1–26 of the '314 patent under 35 U.S.C. § 103(a) — obviousness over Wise + Woods, Wise + Dasan (claims 1–25), and Wise + Dasan + Uppaluru (claim 26); the petition also pressed a broken priority-chain / written-description challenge to whether Woods (Ex. 1012) was prior art (DocketAlarm / AboutBlaw FWD excerpts)
• Institution decision: Instituted on all challenged claims — 2022-11-07 (Patexia)
• Final Written Decision: Issued 2023-11-06 (DocketAlarm FWD PDF; Law360 reported 2023-11-08). Verdict at claim level: claims 1–26 all unpatentable. Panel quote: "Based on the record before us, Petitioner has shown, by a preponderance of the evidence, that claims 1–26 are unpatentable." The FWD also contains a duty-of-candor / show-cause section directed at Parus's counsel (Scott W. Hejny) over priority-chain representations to the Board (Banner Witcoff copy of FWD)
• Settlement / termination: None — full merits FWD
• Appeal: Docketed at the Federal Circuit as Appeal 2024-1329 (per Patexia's case summary). Disposition as of my sources: not confirmed
• Defensive value: This is the single most valuable document for anyone defending the '981 patent. The '314 claims (voice-command retrieval using a speech-recognition grammar, an information-source-retrieval instruction, a web-browsing server accessing "a portion" of a web page, a content extractor, and a content-descriptor file) are substantively the same claim architecture as '981 claims 1, 5, 10, 19, and 24. The Wise/Woods/Dasan/Uppaluru combinations are a ready-made invalidity roadmap. (Note: this proceeding is estoppel-binding only on Google and its privies, not on a new defendant.)

IPR2022-00948 — Apple Inc. v. Parus Holdings, Inc. (U.S. 9,769,314 B2 — same sibling patent)

• Type: Inter Partes Review
• Filed: on or about 2022-05-05 (post-institution fees paid 2022-05-05; DocketAlarm shows Apple filing Google's '314 IPR petition as Ex-1017 on that date)
• Status: Terminated without a Final Written Decision — the Board approved a refund of Apple's post-institution fees on 2023-03-15 (DocketAlarm Notice of Refund), which indicates the proceeding ended before a merits decision (settlement or joinder/termination; the specific basis is not confirmed in my sources)
• Judge panel: Not confirmed in my sources
• Petition grounds: Apple likewise asserted Woods and a priority-chain contention against the '314 patent (referenced in the Google FWD's show-cause discussion: "Apple's IPR2022-00948, which likewise asserted Woods and presented a [priority-chain contention]")
• Institution decision: Not confirmed; the post-institution fee refund implies no FWD followed
• Final Written Decision: None
• Settlement / termination: Terminated by 2023-03-15 with fee refund — terms (if any) not public in my sources
• Appeal: None identified
• Defensive value: Confirms that a second major tech defendant (Apple) independently targeted the same family with the same art — reinforcing that the Wise/Woods/Dasan cluster is the strongest attack path against '981-style claims.

Google IPRs on the direct parent — US 7,516,190 (and US 7,386,455): institution denied

• Type: Inter Partes Review (petitions by Google LLC)
• Status: Institution denied — per Parus's own counsel: "The current victory follows prior wins in which the Board denied Google's requests to review the claims of Parus' U.S. Patents Nos. 7,386,455 and 7,516,190" (McKool Smith press release, 2024-08-15)
• Proceeding numbers: Not available in my sources — do not guess; the PTAB E2E search for Parus Holdings as patent owner will yield the numbers
• Defensive value: Important caveat — a denial of institution is not a merits holding that the claims are patentable. The '190 parent (from which '981 descends) survived the threshold § 314 screening, but that creates no estoppel for anyone and no PTAB finding of validity. For a defendant, this is a two-sided signal: the Board wasn't persuaded by Google's initial showing on '190, but the identical-in-substance '314 claims were later wiped out on similar art — so the merits question on '981-family claims is genuinely live.

Strategic summary

Claims of '981 — CANCELED vs. SUSTAINED vs. UNTESTED. On the '981 patent's own PTAB record: zero claims canceled, zero claims sustained on the merits, all 28 claims untested (independents 1, 5, 10, 19, 24; dependents 2–4, 6–9, 11–18, 20–23, 25–28). The PTAB has never institutionally reviewed — let alone reached — the '981 claims. That absence is itself meaningful: the '981 patent granted in 2019 and expired (per Google Patents) on 2021-02-06, so the window in which a challenger would bother petitioning for IPR largely closed while the family was being litigated on its siblings instead. The practical exposure is past damages only, and the claims most likely to be asserted mirror the '314 claims that a PTAB panel unanimously found unpatentable.

Estoppel landscape. § 315(e)(2) estoppel runs only to the petitioners and their privies in the proceedings where grounds were raised or reasonably could have been raised. Google (IPR2022-00805) and Apple (IPR2022-00948) are the only parties estopped from re-litigating Wise/Woods/Dasan/Uppaluru against the '314 patent — and neither estoppel extends to a new defendant asserting against the '981 patent. For a defendant being sued on '981 today, all prior-art grounds remain fully available — including exactly the art that felled the sibling: Wise (Ex. 1007), Woods (Ex. 1012), Dasan (Ex. 1013), and Uppaluru, in the combinations and with the claim-element mappings laid out in the IPR2022-00805 FWD. Google's April 2022 stipulation in the W.D. Tex. case (agreeing not to pursue Wise-based obviousness in court if the IPR instituted) is a useful template, not a constraint, on a new defendant.

Pattern signals. This is a coordinated, multi-year, multi-front attack on the Parus voice-browser portfolio by Google and Apple — with Google as the lead aggressor. Across the portfolio the results are mixed: full cancellation of the '314 sibling (IPR2022-00805) and of the related '945,1084 family (CAFC affirmed, appeal 22-1270); institution denied on the '190 parent and '455; and in a different Parus family (the '705 patent), the CAFC just vacated a Parus win and remanded (IPWatchdog, 2026-07-14) — showing the Board and Federal Circuit are actively second-guessing each other on these voice-browser patents. Parus fights hard (McKool Smith, with a sanctions show-cause episode in IPR2022-00805 over priority-chain candor). No defensive aggregator (e.g., Unified Patents) appears in the chain — these were direct competitor-filed petitions. The Amazon litigation in Delaware (Parus Holdings Inc. v. Amazon.com, Inc., 1:23-cv-00190, D. Del.) asserts family patents including 7,516,190 and 9,377,992, and the Stanford NPE Litigation Database also lists '981 as involved — the district court, not the PTAB, is where this patent is being fought.

Recommended next steps

• If you are a defendant: Do not wait for an IPR that will never come — the patent is expired (2021-02-06 per Google Patents' legal-status record; confirm exact expiration/maintenance-fee history via USPTO Patent Center), so only past damages within the § 286 six-year window are at stake, and there is no injunction risk. The highest-value move is district-court invalidity on the IPR2022-00805 record: pull the FWD (DocketAlarm copy: https://www.docketalarm.com/cases/PTAB/IPR2022-00805/Google_LLC_v._Parus_Holdings_Inc/docs/11-06-2023-Board/Final_Written_Decision__original-35-Determining_All_Challenged_Claims_Unpatentable_35_USC_%C2%A7_318a.pdf) and the petition's exhibit set, and map Wise/Woods/Dasan/Uppaluru onto the '981 claims — the claim architecture is the same, and no estoppel bars you.
• If the demand letter cites claims tracking '314 claims 1–26 (speech-recognition grammar → media server selects an information-source-retrieval instruction → web-browsing server accesses "a portion" of a page via a content extractor using a content-descriptor file): quote the FWD's sentence — "Petitioner has shown, by a preponderance of the evidence, that claims 1–26 are unpatentable" — and put Parus to its burden of showing those claims are materially different from the canceled sibling claims; on this record that is a steep climb.
• Monitor the pending threads: Federal Circuit Appeal 2024-1329 (Google's appeal of the '314 FWD) — if the CAFC affirms, the invalidity roadmap is appellate-blessed; if it reverses, the roadmap weakens. Also monitor the Delaware case (1:23-cv-00190) for claim-narrowing and for which '981 claims Parus actually asserts.
• Confirm the procedural record before relying on it: the two items I could not verify from primary sources are (a) the docket numbers of Google's institution-denied IPRs on 7,516,190 / 7,386,455 (search PTAB E2E under "Parus Holdings" as patent owner), and (b) the precise termination basis of IPR2022-00948 (the fee refund indicates termination without a FWD, but the underlying order should be pulled from PTAB E2E). No active PTAB trial-stage milestones exist for '981 itself because no trial was ever instituted on it.

I'll research the assignment history for US10320981B2 through USPTO records and cross-referencing sources.

Let me dig deeper into the USPTO assignment database and the specific recorded transactions.

Let me search for the specific assignment records in the USPTO database and the litigation docket details.

I've hit my search budget, so I'll now consolidate the verified findings and flag explicitly where the record could not be independently confirmed.

Ownership Chain Analysis — US Patent 10,320,981 (US10320981B2)

Verification note up front: I could not pull the raw USPTO Assignment Center reel/frame entries for this exact patent in the searches performed. What follows is the reconstructed chain from (i) Google Patents legal-event records for US10320981, (ii) a 37 CFR 3.73(c) chain-of-title statement recorded in a related Parus-family application (12/973,475 → US8,185,402), and (iii) litigation dockets. Where a reel/frame comes from the related-family statement rather than the 10320981 file wrapper, I say so. No reel/frame numbers are fabricated.

Inventors

• Alexander Kurganov — sole named inventor (Buffalo Grove, IL). Employer at time of the underlying invention/filing: Webley Systems, Inc. (the patent's own specification credits "a database designed by Webley Systems Incorporated" and "natural speech recognition grammars … developed by Webley Systems"). Per Kurganov's own declaration in IPR2022-00358 (Ex. 2052, Docket Alarm), he was VP of Engineering at Vail Systems, Inc. from 1994 and "conceived and started the design" of the voice/web platform "in early 1996, while still working for Vail Systems," i.e., the technology behind this family was built at/for Vail/Webley before Parus Holdings existed.
• Unusual pattern? Not the classic "inventors flee the assignee before a fire-sale" pattern. The sole inventor is the founder/principal of the technology and later assigned into the family holding company. The anomaly is the inverse: one individual's platform work migrated through three corporate identities (Vail → Webley → Parus Holdings) and ended up asserted by a "Holdings" entity 23 years after priority.

Original assignee

• Parus Holdings, Inc. is the entity named on the issued patent (also per Google Patents: 2017-12-19 "ASSIGNMENT OF ASSIGNORS INTEREST … Assignors: KURGANOV, ALEXANDER" → PARUS HOLDINGS, INC.).
• Did they ship a product? The predecessor technology did — Webley Systems' network-based integrated voice-and-web platform was "deployed in production and successfully hosted high volume voice and web applications for major companies, such as Allstate, US West" (Kurganov decl., IPR2022-00358 Ex. 2052). Whether the current Parus Holdings, Inc. entity ships anything is not evidenced anywhere I found; its observable activity is assertion.
• Primary line of business / current status: patent holding and assertion. Parus Holdings is the plaintiff in Parus Holdings Inc. v. Amazon.com, Inc., 1:23-cv-00190 (D. Del., filed 2023-02-17) asserting US7,516,190; US9,377,992; and US10,320,981 against Amazon's Alexa/Alexa Voice Service (Justia docket; Stanford NPE Litigation Database; Unified Patents litigation portal). The family is flagged "has litigation" on Google Patents. A related family branch (US6,721,705 / US8,185,402 / US8,843,141) has been through PTAB proceedings with Google (CBM2015-00111; IPR2022-00358; IPR2022-00523) and a CAFC appeal (Google LLC v. Parus Holdings, Inc., Appeal No. 2024-2381, vacatur of a PTAB decision reported July 2026). No bankruptcy, acquisition, or dissolution of Parus was found.

Assignment timeline

Records confirmed or evidenced, in chronological order:

• ~2001 (executed) — recorded Reel 033922 / Frame 0418 (per 3.73(c) chain-of-title statement in related app. 12/973,475, dated 2014-12-01, Docket Alarm)

  • Conveyance: Assignment (inventor → company), covering the original 09/777,406 family
  • Assignor: Alexander Kurganov (with Valery Zhukoff in the related-family record)
  • Assignee: Webley Systems, Inc.
  • Correspondent: not disclosed in my source
  • Context: Initial assignment of the voice-browsing technology into the operating company. Caveat: reel/frame verified only via the related-family 3.73(c) statement, not the 10320981 wrapper.

• ~2001–2004 (executed) — recorded Reel 033922 / Frame 0431 (same 3.73(c) source)

  • Conveyance: Assignment (company → successor)
  • Assignor: Webley Systems, Inc.
  • Assignee: (not legible in my source; the 3.73(c) chain ends at Parus Holdings, Inc.)
  • Correspondent: not disclosed in my source
  • Context: Internal migration of the portfolio from Webley Systems to the Parus Holdings entity.

• 2017-12-19 (recorded) — reel/frame not confirmed in my sources (Google Patents legal event; USPTO Assignment Center should be searched by patent no. 10320981 for the reel/frame)

  • Conveyance: Assignment of Assignor's Interest
  • Assignor: Alexander Kurganov
  • Assignee: PARUS HOLDINGS, INC.
  • Correspondent: not confirmed (prosecution representative of record for US10320981 was Patent Law Works LLP per PatentGuru, and attorney Reena Kuyper appears in family prosecution records — but neither is confirmed as the recorded assignment correspondent)
  • Context: Confirmatory assignment vesting title in Parus Holdings for the 15/707,951 continuation, executed 5+ years before the Amazon suit.

Post-issuance (2019-06-11 → present) assignments: none found. No RPX/AST/LOT/Unified Patents defensive transfers, no LLC chaining, no security-interest filings surfaced. The recorded chain appears to terminate at Parus Holdings, Inc., which is consistent with Parus being the plaintiff asserting the patent in D. Del. (standing would require it to hold title).

Timeline diagram

timeline
    title Ownership of US 10320981
    2000 : Priority provisional filed
    2001 : Original app filed
    2001 : Kurganov assigns to Webley
    2004 : Webley to Parus Holdings
    2009 : Parent patent 7516190 issued
    2017 : Continuation filed
    2017 : Kurganov confirms to Parus
    2019 : US10320981 issued
    2023 : Parus sues Amazon in Delaware

NPE / troll-pattern signals

1. Shell-entity transfer — weakly present / unclear. Parus Holdings, Inc. carries the "Holdings" suffix and shows no current product in commerce in any source I found, but the predecessor Webley Systems genuinely operated a deployed platform (Allstate, US West), so this is not a clean operating-company-to-licensing-LLC transfer. The strongest concrete evidence is Stanford's NPE Litigation Database listing this exact patent with Parus Holdings as the patent asserter (npe.law.stanford.edu/patent/10320981). No registered-agent-address or single-member-LLC evidence was located.

2. Known asserter in the chain — present. Parus Holdings, Inc. appears as the patent asserter for US10320981 in the Stanford NPE Litigation Database (npe.law.stanford.edu/patent/10320981) and in the Unified Patents litigation portal (Delaware District Court case 1:23-cv-00190). Parus is a repeat litigant across this family: Amazon (D. Del. 2023) on 10320981 and its parents, plus Google through CBM2015-00111, IPR2022-00358, IPR2022-00523, and CAFC 2024-2381 on the sibling 8,185,402/8,843,141 lines.

3. Repeat correspondent across the chain — unclear / not determinable. I could not retrieve the recorded correspondent names on the 10320981 assignments. Patent Law Works LLP handled prosecution (per PatentGuru) and Reena Kuyper appears in family prosecution records, but I cannot confirm either as a recurring assignment correspondent — so no finding is made.

4. Cascading transfers — not present. Only two substantive links (inventor → Webley → Parus Holdings) spread over ~16 years; no chained LLCs, no rapid-fire reassignments.

5. Pre-litigation transfer — not present. The last known recorded assignment (2017-12-19, Kurganov → Parus) predates the first infringement suit on this patent (2023-02-17) by more than five years. Caveat: I could not rule out an unrecorded or 2022-2023 assignment that my sources missed.

6. Bankruptcy fire-sale — not present. No bankruptcy proceedings found for Webley Systems or Parus Holdings; the chain is not a Kodak/Nortel-style estate sale.

7. Privateering — not present / unclear. No SEC or press evidence of an operating company funneling this patent to Parus to attack a competitor. Parus appears to assert on its own behalf (its corporate ancestor was the operating company), which is the classic "failed/acquired product company becomes NPE" arc rather than third-party privateering.

8. Defensive aggregator (anti-NPE) — not present. The chain terminates at Parus Holdings, an assertive plaintiff; there is no RPX, AST, LOT, Unified, or OIN involvement.

Verdict

NPE — moderate confidence.

The deciding evidence is one strong signal — Parus Holdings, Inc. is the current assignee (2017-12-19 assignment; plaintiff of record in 1:23-cv-00190) and is classified as the patent asserter in the Stanford NPE Litigation Database for this exact patent — plus weak supporting signals: a "Holdings" entity with no current product in commerce evidenced, and a family-wide repeat-assertion pattern against Amazon and Google. The case would be "high confidence" only with confirmed post-2019 chained transfers or a recurring assignment correspondent, which I could not verify; the historical Webley operating business also prevents a clean shell-entity call. For independent verification, run patent number 10320981 (and family members 7516190, 9377992, 9769314) at the USPTO Assignment Center: https://assignmentcenter.uspto.gov/ (search by patent number; same underlying records at https://assignment.uspto.gov/patent/index.html#/patent/search).

I've hit the search-step limit, so I'll synthesize the answer from the sources already retrieved (Google Patents record for US10320981B2, the parent US9769314B2 PDF, and cross-referenced patent pages). I'll flag confidence levels explicitly.

Prior Art Analysis — US Patent 10,320,981 B2 ("Personal Voice-Based Information Retrieval System")

A. Source & reliability notes

• What I searched: USPTO/Google Patents records for patent number 10320981 (interpreted literally — no similar numbers), the parent US9769314B2 PDF (which carries the family's "References Cited" page, Primary Examiner Lisa Hashem), and cross-referenced "cited-by" pages for individual references.
• Citation count: The face of US10320981B2 lists 393 references ("Citations (393)" on Google Patents). The complete list was truncated in the retrieved record (visible portion ends at US5416834A). I could not retrieve the USPTO PTOL-892 form or file wrapper in this session, so the examiner's per-claim designations (X/Y/A) are not confirmed; the claim-level §102 mapping below is my technical analysis of each reference's disclosure against the independent claims, not the examiner's formal rejection.
• Statutory framework: The '981 patent is a continuation claiming priority to Provisional App. 60/180,343 (Feb. 4, 2000). Because the chain's effective filing date is before March 16, 2013, pre-AIA §102(a)/(b)/(e) governs. A reference is §102 prior art only if it was (i) published or known/used before Feb. 4, 2000 (§102(a)/(b)), or (ii) a U.S. patent/application filed before that date and ultimately published (§102(e)).

B. The independent claims in one line each (for mapping)

• Claim 1 (method): speech command → speech-recognition engine selects user-assigned recognition grammar tied to a URL/web request → media server retrieves stored information-source-retrieval instruction → web-browsing server executes a content extractor using a content-descriptor file and a named object to pull only the relevant portion of the page → speech-synthesis output to the user's device.
• Claim 5 (system): system-side counterpart of claim 1.
• Claim 10 (method): identify speech command from a lexicon → access a descriptor file identifying (i) a web-accessible source and (ii) request info → fetch response data containing a named object with content → extract content using the named object → generate audio response → direct playback.
• Claim 19 (apparatus): transceiver + database of descriptor files + speech-recognition engine + media server + web-browsing server (named-object extraction) + synthesizer.
• Claim 24 (apparatus, device-side): microphone/speaker device in communication with that remote system.

C. Most relevant prior art (ranked)

1. US 5,819,220 A — "Web triggered word set boosting for speech interfaces to the world wide web"

• Full citation: US Patent 5,819,220 A; inventors Sarukkai et al.; assignee Yahoo! Inc.; granted Oct. 6, 1998 (pre-2000).
• Description: A speech interface to the World Wide Web in which the speech-recognition vocabulary is dynamically "boosted" with terms extracted from web pages; a user speaks a natural-language request, the system performs a web retrieval, and results are returned in audible form. This is a voice-browser reference squarely in the same field as the claimed invention.
• Confirmed as cited: Appears in the cited-by list of the '981 patent's family records (it is cited by US9769314B2 and appears in the same family citation record that includes US10320981B2).
• Potential §102 mapping:
  • Claim 10: Strong — steps (a) identify spoken command, (b) access a web-accessible source using request info, (c) fetch response data, (e) generate audio response, (f) direct playback are all arguably present. The weak point is step (d) — extraction "using a named object" — unless the reference's page-parsing/vocabulary-boosting is read to use named/tagged objects.
  • Claims 1 and 5: Moderate — discloses speech→web-retrieval→audio, but does not clearly disclose the user-defined recognition grammar + URL stored in a user-defined record, the content-descriptor file, or the clipping client creation flow. Anticipation of claims 1/5 fails on those elements absent a broad reading.
  • Claims 19 and 24: Moderate — discloses the speech-recognition engine, web server, and synthesizer architecture, but the named-object extraction and descriptor-file database are not clearly present.

2. US 5,884,266 A — "Audio interface for document based information resource navigation and method therefor"

• Full citation: US Patent 5,884,266 A; inventor Waldman et al.; assignee IBM; granted Mar. 16, 1999 (pre-2000).
• Description: An audio interface that converts documents — including web pages — into audio, parsing the document's structure so that only relevant content is read aloud, and accepting voice commands to navigate among documents/links. It discloses content extraction based on document structure (links, headings — arguably "named objects").
• Confirmed as cited: US10320981B2 appears in the cited-by list on the Google Patents page for US5884266A.
• Potential §102 mapping:
  • Claim 10: Strong candidate — identifies speech commands (lexicon), accesses a document/information source, fetches response data, extracts content based on structural elements, generates an audio response, and directs playback.
  • Claims 19 and 24: Strong — the apparatus (transceiver, speech-recognition, web fetching, synthesizer, playback command) maps well; the descriptor-file database limitation is the main gap.
  • Claims 1 and 5: Weaker — no user-assigned recognition grammar/URL record and no content-descriptor-file created by a clipping client.

3. US 6,233,318 B1 — "System for accessing multimedia mailboxes and messages over the internet and via telephone"

• Full citation: US Patent 6,233,318 B1; assignee Bosch Telecom GmbH (later Ipcom GmbH & Co. KG); granted May 15, 2001; filed late-1990s (German priority 1997/1998 — filing date not independently confirmed in this session; treat with caution). If its U.S. filing date is before Feb. 4, 2000, it qualifies under §102(e).
• Description: A system that lets a user retrieve multimedia messages (voice mail, e-mail, fax) over the Internet and via a telephone using speech commands, with text-to-speech playback of retrieved content — a voice-activated network-retrieval system.
• Confirmed as cited: US10320981B2 appears in the cited-by list on the Google Patents page for US6233318B1.
• Potential §102 mapping: General mapping to claims 1, 5, 10 (speech command → retrieve network-stored content → audio playback). It lacks the web-page content-extraction with named objects and the user-defined grammar/URL record specifics, so anticipation of the independent claims in full is unlikely without a very broad reading.

4. US 2001/0054085 A1 — Kurganov, "Personal voice-based information retrieval system" (applicant's own parent publication)

• Full citation: US Patent Application Publication 2001/0054085 A1; inventor Alexander Kurganov; published Dec. 20, 2001; claims priority to Feb. 4, 2000 (parent of this very patent).
• Description: The published application of the '981 patent's own specification (same disclosure).
• Cited-by marker: Appears with a * (examiner citation) in family records.
• Potential §102 mapping: As a same-family publication, it is normally overcome by the priority claim and would only be §102(e)-type "self-collision" prior art against claims not entitled to the benefit of the earlier filing dates. On its face it discloses everything in the '981 specification (identical disclosure) — so if priority were lost, it would be a complete anticipation of all 28 claims.

5. US 5,325,421 A — "Voice directed communications system platform"

• Full citation: US Patent 5,325,421 A; assignee AT&T Bell Laboratories; granted June 28, 1994 (in the visible "Citations" list).
• Description: A telephony platform where callers direct system functions by voice — voice-directed routing and service selection. Relevant to the IVR + speech-command aspects of claims 1, 5 (the "interactive voice response application" and speech-command selection).
• Potential §102 mapping: Anticipates only the IVR/speech-command fragments of claims 1 and 5; does not disclose web retrieval, URL records, grammars tied to web content, or content extraction. Realistically an "A"-type background reference.

6. US 5,054,054 A — "Voice applications generator"

• Full citation: US Patent 5,054,054 A; assignee IBM; granted Oct. 1, 1991 (in the visible "Citations" list).
• Description: A generator for voice applications that lets developers define speech-recognition grammars and voice-driven application flows — relevant to the "recognition grammar" concept in claims 1 and 5.
• Potential §102 mapping: Discloses recognition grammars and voice-driven application generation, but not the user-assigned-grammar-to-URL/web-content linkage or content extraction. Background/"A" reference.

7. US 5,365,574 A — "Telephone network voice recognition and verification using selectively-adjustable signal thresholds"

• Full citation: US Patent 5,365,574 A; assignee VCS Industries, Inc.; granted Nov. 15, 1994 (in the visible "Citations" list).
• Description: Telephone-network speech recognition with adjustable thresholds — relevant to the speech-recognition-engine/telephony interface elements of claims 1, 2, 5, 7.
• Potential §102 mapping: Limited to speech recognition over telephony; no web/content-extraction elements.

8. US 4,930,150 A — "Telephonic interface control system"

• Full citation: US Patent 4,930,150 A; assignee First Data Resources Inc.; granted May 29, 1990 (in the visible "Citations" list).
• Description: A voice-controlled telephonic interface for operator/services selection — early IVR/voice-command art.
• Potential §102 mapping: Background for the voice-command/IVR aspects of claims 1 and 5 only.

9. Other references in the visible "Citations" list (lower relevance)

The bulk of the 393 citations are telephony, voice-mail, IVR, and speech-processing patents from the 1970s–1990s (e.g., US3,728,486 "Voicegram service"; US4,371,752 "Electronic audio communication system"; US4,757,525/4,761,807 (VMX voice-command/voice-authentication); US4,776,016 "Voice control system"; US4,850,170 "Real time computer speech recognition system"; US5,054,054 above; US5,177,958 "Integrated voice messaging/voice response system"; US5,195,086; US5,325,421; US5,365,574). These are best characterized as §102/§103 background for the speech-recognition and IVR limitations, not standalone anticipations of any independent claim, because none discloses the combination of user-defined recognition grammar + URL + content-descriptor-file/named-object web extraction + speech-synthesis output.

10. References in the family's "Families Citing" list — not prior art to '981 (excluded)

Patents such as US6,775,264 B1 (Webley Systems), US6,904,450 B1 (Geodata), US6,937,986 B2 (Comverse), US2001/0017855 A1 (Matsushita) appear in Google Patents' "Families Citing this family" section — i.e., they cite the Parus family, so they are later work, not prior art cited by the '981 patent. (Caveat: US6,775,264 B1, filed Mar. 3, 1997 and granted 2004, could qualify as §102(e) art against the 2000 priority date if it were actually on the '981 reference list; I could not confirm it is among the 393 citations. It is the same Webley/Parus ecosystem and would be a strong §102(e) candidate if present.)

D. Bottom-line §102 assessment

Key takeaway: No single pre-2000 reference in the confirmed record appears to disclose the full combination of the independent claims — particularly (i) a user-assigned recognition grammar stored with a URL in a user-defined record, and (ii) content extraction from only the relevant web-page portion using a content-descriptor file and a named object. The two closest anticipatory candidates are US 5,819,220 and US 5,884,266 (both pre-2000 voice-browser systems), with claim 10 the most vulnerable of the independent claims if "named object" extraction is read broadly.

E. Unresolved items (for a complete analysis)

1. Full 393-reference list — the retrieved record was truncated; the complete list (and any post-2000 §102(e) references filed before Feb. 4, 2000) should be pulled from the USPTO PatentCenter file wrapper or the issued PDF.
2. Examiner designations (X/Y/A) on the PTOL-892 — not accessible in this session.
3. Filing date of US 6,233,318 B1 — needed to confirm §102(e) status against the Feb. 4, 2000 priority date.
4. Whether US 6,775,264 B1 (Webley) is among the 393 citations — if yes, it is a significant §102(e)/§103 reference given the shared Webley/Parus lineage.

I have reached the tool-call limit, so I will now construct the § 103 analysis from the prior-art list on the patent page plus the verified disclosures I obtained for the key references (US6775264B1, US6904450B1, US6937986B2, US6934684B2, US20010017855A1). I will flag explicitly where a reference's full disclosure could not be verified.

Obviousness Analysis of US 10,320,981 B2 Under 35 U.S.C. § 103

0. Threshold: which cited references are actually prior art (and the priority-date problem)

The '981 patent is a 2017 continuation claiming priority through a chain to Provisional Application 60/180,343 (Feb. 4, 2000). Because it is an AIA-era filing, prior art is measured against the effective filing date (assumed to be Feb. 4, 2000 if written-description support exists for all claim limitations). This creates a threshold problem for most of the references in the "Prior Art/Citations" section of the Google Patents page:

Consequence: A rigorous § 103 case built solely on the verified pre-2000 references in the list is materially weaker than a case using Geodata, Comverse, Dialsurf, or Matsushita. The post-2000 references only become available if the challenger first defeats the priority claim (e.g., by showing the "content-descriptor file" / "named object" limitations are new matter in the continuation chain — an argument that is plausible but not strong, because the original specification does discuss both content-descriptor files and named objects in hypertext documents). I analyze both scenarios below.

Second caveat (same-inventor/common-assignment): US6775264B1 is by the same inventor (Alexander Kurganov) and the same corporate ecosystem (Webley Systems, Bannockburn, IL — the same address as Parus Holdings and the entity the '981 specification credits for its database). Under AIA 35 U.S.C. § 102(b)(2)(C) and § 103, subject matter in a commonly owned application/patent by the same inventor may be disqualified as prior art if ownership/assignment obligations overlapped at the time of invention. A challenger relying on '264 must therefore establish that Webley and Parus were not the same person/assignee at the relevant time. I note this but proceed on the assumption a challenger could do so, or could use '264 in combination with other references.

Third caveat (unverified references): I could not verify the disclosures of US7653748B2 (Simplexity) and US7228493B2 (Lycos) before the tool limit. I identify them below only by title/assignee as supporting content-extraction/clipping art, and I flag that their specific teachings need confirmation before being relied on.

1. The closest primary reference: US6775264B1 (Webley, priority 1997)

Verified disclosure: a computer/telecommunications network with voice servers providing telephony, speaker-independent speech recognition, text-to-speech, and conferencing, accessible via standard telephone connections and Internet connections; a database server holding subscriber-managed contact lists/administrative data; a file server; and a web server connected to the cluster so the subscriber can access network functions over the Internet. It is, in substance, the server-side voice-browsing/voice-portal infrastructure that the '981 claims presuppose, and it was designed by the same inventor/company that built the '981 system.

What '264 does not verifiably disclose: user-assigned recognition grammars tied to specific web content, selective extraction of only a portion of a web page, content-descriptor files, or named-object-based extraction. Those are the gaps the combinations below fill.

2. Combination A (primary): Webley '264 + Comverse '986 + Geodata '450

This is the strongest combination, though it depends on the priority-date question (Comverse '986 is Dec. 2000; Geodata '450 is Aug. 2000) and thus presumes the priority claim is defeated, or that these references qualify under a different theory (e.g., the claims were amended during prosecution to rely on later-added limitations).

Element-by-element mapping against independent claim 1

Why the combination would have been obvious (motivation): By 2000, a POSITA designing a voice portal faced two well-known, independent problems: (i) speech recognizers were accurate only with constrained, per-application grammars, and (ii) web pages contained far more information than a user wanted to hear read aloud. Comverse '986 was expressly directed to problem (i) — automatically building user-specific grammars from external sources so users can ask "what is the weather in [my city]" — and taught that fetched web text should be cleaned of HTML before synthesis (problem (ii), partially). Geodata '450 was expressly directed to problem (ii) — customizable retrieval so that only selected information from selected sites is obtained. Webley '264 supplied the known telephony + ASR + TTS + web-server platform. Combining a voice portal with a user-grammar generator and a web-content-selection tool is the textbook KSR v. Teleflex situation: each element performs its known function, the combination uses known techniques (grammar augmentation, URL/bookmark storage, HTML parsing and stripping), and the result (voice access to a user-chosen slice of a web page) is predictable. The '981 specification itself concedes that named objects (tables, buttons, forms) in hypertext documents and content-descriptor files are conventional tools, which undercuts any argument that element (c)'s "named object" limitation is a non-obvious inventive step rather than a routine parsing choice.

Weaknesses of Combination A: (i) priority-date problem for '986/'450 (both post-date 2000-02-04); (ii) Geodata '450's director is aimed at administrative/parental control of browsing, not voice retrieval — a challenger must articulate why a voice-portal engineer would adopt it (answer: it is the closest prior teaching of "descriptor file that identifies source + what to retrieve," and its combination with a voice front end is a predictable application); (iii) neither '986 nor '450 verifiably teaches named-object location within a page — that element is likely to be the battleground, and the challenger would need Simplexity '748 / Lycos '849 or general HTML-parsing knowledge to fill it (see § 5).

3. Combination B: Webley '264 + Matsushita '855 (voice-calling of URLs)

Matsushita '855 (filed 2000-02-29, published 2001-08-30 — post-dates the priority date by ~3 weeks) verifiably teaches: a portable telephone capable of calling a URL by voice input, where a spoken name is matched against registered URLs, and the telephone then accesses the URL through a browser. This is the client-side half of the '981 concept: user-assigns a pronounceable name to a URL; speaking the name invokes retrieval of that URL. Webley '264 provides the server-side half: a network-accessible voice server with ASR and TTS that retrieves data from the web and delivers it over the phone.

Motivation: The Matsushita reference itself identifies the usability problem (entering URLs on a phone is cumbersome) and the known solution (voice-to-URL binding). A POSITA combining '855's voice-to-registered-URL mapping with '264's server-side ASR/TTS/web access would be doing no more than relocating a known client-side function into a known voice-portal architecture — an obvious design choice with a predictable result (hands-free, server-side voice retrieval of a user-selected web page). This combination is strong on claim elements (a) (grammar/URL pairing), (b) (stored instruction), and (f) (phone delivery), but it is weak on the content-extraction limitations (c) and (d) — '855 retrieves whole pages; it does not extract only a portion using a content-descriptor file.

4. Combination C (pre-2000 only, if the priority date stands)

If the priority date is upheld and only pre-2000 references are available, the strongest case is:

Webley '264 + US4974254A (interactive data retrieval producing user-selected reports) + US5054054A (IBM voice-applications generator) [+ US5325421A / US5365574A for the ASR/IVR side]

• Webley '264: the voice-portal platform (ASR + TTS + web server + database + telephony).
• US4974254A (1989): verifiably titled "Interactive data retrieval system for producing facsimile reports" — teaches user-selected data retrieval and formatted output of only the selected items, i.e., a pre-2000 analogue of "retrieve only the requested portion" and of a stored retrieval instruction.
• US5054054A (IBM, 1989): "Voice applications generator" — teaches programmatically generating voice applications and their recognition vocabularies, i.e., user/developer-definable spoken commands mapped to application actions.
• US5325421A / US5365574A: voice-directed telecommunications services with speech recognition, useful for the IVR/media-server elements.

Motivation: This combination addresses the same two problems as Combination A but with art that genuinely predates 2000: a POSITA building a voice portal (as in '264) who wanted to let users say a stored name to retrieve a specific slice of an online data source would look to (i) interactive-retrieval systems that already returned only user-selected items ('974,254) and (ii) voice-application generators that already let commands be defined and bound to actions ('054). Combining them is the routine integration of a recognition front end with a retrieval back end.

Weaknesses: The pre-2000 references are older and less on-point for web content extraction than the post-2000 references; '974,254 is facsimile-oriented, and none of the pre-2000 references verifiably teaches named-object HTML extraction or a content-descriptor file per se. This combination is more vulnerable on elements (c)–(d).

5. The "content extractor / named object / content-descriptor file" limitation (claims 1(c), 5(c), 10(d), 19(e), 24)

This is the most distinctive limitation and the most likely point of contention. The best support in the citation list is:

• US7653748B2 (Simplexity, 2000-08-10) — "Systems, methods and computer program products for integrating advertising within web content": by title, teaches parsing web content and manipulating selected portions of a page. Unverified — I could not retrieve its disclosure before the tool limit; its use requires confirmation.
• US7228493B2 (Lycos, 2001-03-09) — "Serving content to a client": by title, teaches server-side selection/delivery of content to a client. Unverified for the same reason.
• Comverse '986 — verified to teach extracting keywords from fetched web pages and stripping HTML before synthesis, which is a limited form of content extraction.
• The '981 specification's own admission that "most hypertext documents include named objects like tables, buttons, and forms" and that an extractor can "simply search[] for the relevant name of [the] desired object" — this frames named-object location as a known, conventional HTML-parsing technique, which a challenger will use to argue that element (c) is an obvious implementation detail.

Even if the named-object limitation is not literally in any single cited reference, the § 103 case would be: user-defined grammar+URL binding (Comverse '986 / Matsushita '855 / Webley '264) + customizable web-content selection (Geodata '450 / Simplexity '748 / Lycos '849 / '974,254) + conventional HTML parsing by named tags (admitted in the '981 spec) = the claimed system. Under KSR, a combination of known prior-art elements each performing its known function, with no unexpected result, is unpatentable.

6. Dependent claims (brief)

• Claim 2/3/7/8 (access via landline/wireless/IP phone; media server as user interface): squarely within Webley '264 (telephone and Internet access; subscriber-facing servers) and Dialsurf '684 (telephone-based speech service).
• Claim 4 and 9 (clipping engine generating the content-descriptor file): Geodata '450's director (user-configurable retrieval definitions) and, if verified, Simplexity '748/Lycos '849 (content selection/serving) provide the natural teaching; the "clipping" client is a conventional browser-extension concept.
• Claims 10–18 and 19–28 (lexicon, descriptor file identifying source + request info, named-object extraction, TTS output, playback on the user device): these track Combination A's mapping; the "descriptor file identifies the web-accessible source and request info with a specific URL format including search parameters" (claim 16) is directly analogous to Geodata '450's URL/director structure and to Comverse '986's URL+keyword records.

7. Bottom line and caveats

1. Strongest combination: Webley '264 + Comverse '986 + Geodata '450 (Combination A) — covers nearly every limitation of claims 1, 5, 10, 19, 24 on a straightforward element-by-element basis, with a classic KSR motivation story. Its principal vulnerability is that '986 and '450 post-date the Feb. 4, 2000 priority date, so it only works if the priority claim is defeated.
2. Pre-2000 fallback: Webley '264 + '974,254 + '054 (Combination C) — defensible but weaker on the web-content-extraction and named-object elements.
3. Uncertainty flags: (i) I could not verify the disclosures of Simplexity '748 and Lycos '849 — do not rely on them without reading them; (ii) Webley '264's availability may be barred by AIA § 102(b)(2)(C) common-ownership/same-inventor rules if Webley and Parus are treated as one assignee; (iii) the priority-date analysis is the single biggest swing factor and should be tested first (does the 2000 disclosure support "named object" extraction and "content-descriptor file" as claimed? If yes, only Combination C's references qualify as § 102 prior art); (iv) I found no PTAB IPR grounds specific to '981 in my searches, so this analysis is constructed from the family's own citation list rather than from a litigated ground.