US 8762389

I'll search for authoritative information on this patent number, including any litigation activity.

The initial searches returned nothing. Let me try alternate queries.

Let me get the actual claim text and check for any litigation/docket activity.

US Patent 8,762,389 — Analyst Summary

Caveat on sourcing up front: The full patent text I was given (fetched from Google Patents on 2026‑09‑24) is truncated — it cuts off inside the Detailed Description ("Popularity scores and ranking information corresponding to a collection of sea…"), just before the claims and drawings text. So my claim overview below is grounded in (a) the Summary of the Invention section, (b) the abstract, and (c) a verified Google Patents/Justia snippet of claim 8. I do not have the verbatim text of the independent claims and will not fabricate it.

Bibliographic data (verified)

Sources: https://patents.google.com/patent/[US8762389B1](/patent/US8762389B1)/en ; https://patents.justia.com/patent/8762389

Abstract (as published)

"Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for obtaining one or more first search results that were generated for a search query; determining a score associated with the first search results; revising the search query using a query revision rule; obtaining one or more second search results that were generated for the revised search query; determining a score associated with the second search results; and evaluating the query revision rule by comparing the score associated with the first search results with the score associated with the second search results."

Plain-language overview

The patent addresses how a search engine can decide, automatically, whether adding or substituting a candidate synonym in a query actually improves the results — without human raters or relevance judgments.

The core mechanism:

1. Obtain the results for the original query. For each result, obtain a "popularity score" (a query-dependent quality metric derived from user selections such as clicks/rollovers/dwell) and a rank/position.
2. Compute an aggregate "search results score" by generating a position score per result (ranking position × popularity score) and combining them. Results with no available popularity score contribute no position score. Variants include transforming ranks (e.g., raising rank to a power between 0 and 1, or square root) and capping/transforming popularity scores at a threshold.
3. Revise the query (add a candidate synonym; substitute a term; add quotations or exclusionary terms).
4. Repeat the scoring on the revised query's results.
5. Compare the two aggregate scores (e.g., take the difference) and test it against a specified threshold. If satisfied, the candidate synonym is designated a synonym (or the revision rule is judged a "good revision"); if not, it is flagged a non‑synonym.

The specification also covers synonym rules with query contexts (e.g., "pet" is a synonym for "dog" only when "dog" is followed by "food"; "felines" is a synonym for "cats" alone but not for "cats musical"), and notes the evaluation can be driven from queries stored in a query log.

Independent claims — plain-language overview

Claim 1 (independent, method): A method comprising: (1) obtaining one or more first search results generated for a search query that includes a first query term; (2) determining a score associated with the first search results; (3) revising the search query to include a second query term; (4) obtaining one or more second search results generated for the revised query; (5) determining a score associated with the second search results; and (6) comparing the two scores and, in response, determining whether the second query term is a synonym for the first query term. Grounded in the Summary section ("In general, one aspect… can be embodied in methods that include the actions of…").

Claim 8 (dependent on claim 1), verified verbatim snippet: "The method of claim 1, further comprising: storing the second query term as a synonym for the first query term." (Justia, https://patents.justia.com/patent/8762389)

Second independent aspect (the "query revision rule" claim — exact claim number not verified): A method comprising obtaining first search results for a search query; determining a score associated with the first search results; revising the search query using a query revision rule; obtaining second search results for the revised query; determining a score associated with the second search results; and evaluating the query revision rule by comparing the two scores. The specifications also recite corresponding system, apparatus, and computer-program/computer-storage-medium claims ("Other embodiments of this aspect include corresponding systems, apparatus, and computer programs, configured to perform the actions of the methods, encoded on computer storage devices"), so there is almost certainly at least one system and/or non‑transitory CRM independent claim. I could not verify the exact number or wording of that claim from the sources retrieved.

Claim features shared across aspects (all described as optional features): popularity scores reflecting query-context popularity; ranks indicating display order; queries sourced from a query log; position scores = popularity × rank; summing position scores; and calculating a difference and testing against a specified threshold.

Litigation / CAFC 2026 docket status

• No litigation or judicial activity was found. My searches — including a targeted query for "8762389 / 8,762,389 CAFC 2026 docket litigation" and district-court queries — returned no results. I found no CAFC 2026 docket, no district-court case, no PTAB proceeding, and no assignment security interest for this patent.
• Because the search tool returned empty result sets rather than confirming "no records," I cannot state affirmatively that no such proceeding exists. This should be treated as "not found," not as "verified none." My tools did not include direct pacer/CAFC docket access.
• Contextual note (not evidence of litigation): the patent is a Google‑assigned, search-infrastructure patent with a 2012 priority date, and its family members (US 8,812,518; US 9,507,853) and its citations appear in later Google/Verizon/Oath patent references. Nothing indicates it has been asserted.

Uncertainties and oddities I will not paper over

1. Truncated source. I do not have the verbatim independent claim language, the exact number of claims, or the drawing content beyond what the Description references (FIGs. 1–5).
2. Internal inconsistency in the fetched specification. The pseudo‑code discussion of FIG. 2 states the impact definitions inconsistently: it says a synonym has a negative impact if the original score is less than the revised score, a positive impact if the original score is greater than the revised score, and also a "no impact" where the original is greater than the revised — i.e., the "positive" and "no impact" conditions are stated identically, and the directionality appears reversed relative to the higher‑is‑better score math used elsewhere. The FIG. 4 worked example compounds this (original 2.9 vs. revised 3.5 is computed as "−0.6" and described as worse results). I flag this as a drafting/transcription ambiguity in the document as fetched; I am not asserting the granted claims themselves are inconsistent, and the difference could be an artifact of the rendering.
3. Score-reference numbering. In the FIG. 2 description, the text compares "search results score 224" to "search results score 226," but 224/226 are elsewhere the search‑result lists while 234/236 are the scores. Apparent labeling inconsistency in the fetched text.

If you need the verbatim claim set for a validity/infringement or docket check, I'd recommend pulling the patent's full text and the CAFC/PACER docket directly from USPTO PatentCenter and the Federal Circuit's docket system, since my searches here returned no docket records and the provided patent text was incomplete.

I'll search for litigation involving US patent 8762389 across patent litigation databases.

Let me search patent-specific litigation databases directly.

Let me run a few more targeted checks to be thorough.

Search Results: Litigation Involving U.S. Patent No. 8,762,389

Bottom line: I found no litigation involving U.S. Patent No. 8,762,389.

After searching patent-litigation sources (Unified Patents portal, Docket Alarm, Justia, CourtListener, and general web searches for the exact number), I could not identify any district-court action, ITC investigation, PTAB proceeding (IPR/PGR/CBM), or CAFC appeal in which U.S. Patent No. 8,762,389 was asserted or challenged.

Important caution — do not confuse with other "…389" patents

Search results frequently surfaced unrelated patents whose numbers end in "389." None of these is U.S. Patent No. 8,762,389:

• U.S. Patent No. 8,726,389 — asserted in Webroot, Inc. et al. v. AO Kaspersky Lab / Forcepoint, LLC, No. 6:22-cv-00243-ADA-DTG (W.D. Tex., Waco Div.) (OpenText/Webroot). This is 8,726,389, a different patent.
• U.S. Patent No. 7,873,389 — asserted by Hera Wireless S.A. / Sisvel against Amazon, ARRIS, Belkin, Buffalo, Netgear, Roku, Lenovo, LG (D. Del. 2017); subject of IPR2018-01737 (Unified/Ruckus et al. v. Hera Wireless). Different number.
• U.S. Patent No. 6,233,389 — the TiVo "multimedia time warping" patent in TiVo Inc. v. EchoStar (E.D. Tex., No. 2:04-cv-00001). Different number.
• U.S. Patent No. 8,762,389 should also not be confused with design patent D802,155 (Range of Motion Products v. Armaid) or the OLED "389 patent" (Pictiva v. Samsung, N.D. Tex. No. 2:23-cv-00495).

Context on the patent itself

U.S. Patent No. 8,762,389 ("Synonym identification based on search quality") was filed 2012-02-02, issued 2014-06-24, list inventors Dan Popovici and Robert Spalek, and is assigned to Google LLC (originally Google Inc.). It is listed as active with an adjusted expiration of 2032-02-13. Google is not a known patent-assertion plaintiff, which is consistent with the absence of litigation for this specific patent.

Confidence and limitations

• Moderate-to-high confidence that no litigation exists as of the search date. Google Patents' "litigation" data (which aggregates Docket Alarm/Unified Patents feeds) and the litigation databases I could reach returned no case tied to this exact number.
• Caveat: I was unable to directly query the Unified Patents litigation caselist (portal.unifiedpatents.com/litigation/caselist) or PACER for this specific number, so I cannot rule out a recently filed or very obscure action that those aggregators have not indexed. If you need a definitive negative, the reliable verification steps are:
  1. Unified Patents Portal — https://portal.unifiedpatents.com/litigation/caselist (search "8762389")
  2. Docket Alarm / Lex Machina / Docket Navigator — search the exact patent number
  3. USPTO Patent Center "Litigation" tab and the assignment/reexamination record for the patent
  4. PACER nationwide case search

Because the tools returned no contradicting data and no case for the exact number, my answer is: no known litigation involving U.S. Patent No. 8,762,389 is on record.

I'll verify the structured data against live sources before drafting.

Let me run a few more targeted queries to rule out proceedings the ODP might not have indexed.

Proceedings overview

Zero AIA trial proceedings on file. The structured PTAB block supplied with this task (USPTO Open Data Portal, most recent ingest) returns no IPR, PGR, or CBM proceedings naming US 8,762,389. My independent web checks — Google Patents/Justia family pages, PTAB-related coverage, and docket-level searching for the patent number and for the Popovici/Spalek inventor pair — surfaced no petition, institution decision, FWD, or appeal for this patent. Breakdown by status is therefore: active 0 / claims invalidated 0 / claims sustained 0 / settled 0 / institution denied 0. I found no older or recently-filed proceeding the ODP has simply failed to index, so the instructed default stands: no PTAB activity on file.

Bottom-line defensive posture: there is no IPR-based estoppel or cancellation to lean on, and no PTAB record to cite. The patent (filed 2012-02-02, granted 2014-06-24, adjusted expiration 2032-02-13) reaches the end of its term with every claim untested at the Board. A defendant facing assertion today gets neither help nor hindrance from PTAB history — the validity fight starts from scratch, which cuts both ways: nothing has been canceled, but no claim has been blessed either, and § 315(e)(2) estoppel is a non-issue.

(No per-proceeding headings follow, because there are no proceedings to describe. I will not manufacture proceeding numbers or invent an FWD.)

Strategic summary

Claim status — all UNTESTED. Because no AIA trial was ever instituted, no claim of US 8,762,389 has been canceled, confirmed, or otherwise adjudicated by the PTAB. There is no narrowed, surviving claim set to report. For the record, the patent's claim set as granted appears to run claims 1–24 — a method group (claim 1 et seq.), a non-transitory computer storage medium group (claim 9 et seq.), and a system group (claim 17 et seq.) — with claim 1 reciting the core steps of (i) determining a first search results score from respective ranks and first popularity scores, (ii) revising the query, (iii) determining a second search results score from respective ranks and second popularity scores, (iv) determining that the difference satisfies a specified threshold, and (v) determining that the second query term is a substitute for the first. Claims 8 and 24 add "storing the second query term as a synonym." I state claim numbering as apparent from the published text rather than as a PTAB finding, since the Board has never construed anything here. Anyone relying on this should pull the issued patent from USPTO PatentCenter and confirm.

Prior-art / estoppel landscape — wide open. With no petition ever filed, § 315(e)(2) estoppel attaches to no one. No defendant, real party in interest, or privy is barred from raising any § 102 or § 103 ground in district court or the ITU. Every printed publication and every prior-art system is fair game. Conversely, no district-court invalidity judgment for this patent appears in my search results either, so there is no collateral estoppel to inherit. Practically: a defendant has a completely unrestricted invalidity case, and picking through the art as a first mover is materially easier than litigating around someone else's failed IPR.

Pattern signals — none of the usual ones. There is no repeat petitioner, no joinder chain, no patent-owner appeal to the Federal Circuit, and no defensive aggregator (Unified Patents or similar) in the chain that I could find. Identifiers to note: the same-day sibling application US 13/365,219 issued as US 8,812,518, and a later filing US 14/300,487 issued as US 9,507,853 — same inventor pair (Dan Popovici, Robert Spalek) and same "synonym identification based on search quality" subject matter. These are separate patents with separate claim sets; an IPR on 8,812,518 or 9,507,853 would not cancel claims in 8,762,389. That family structure matters if you are mapping the portfolio rather than this single patent. The current assignee is Google LLC (originally Google Inc.; assignment re-recorded 2017-10-02). The owner profile — a large operating company holding a 2012-filed, 2014-issued search-quality patent that has never been asserted in the proceedings I could locate — is exactly why this patent has never attracted an IPR. Well-asserted patents eventually draw petitions; this one appears never to have been asserted at all.

One observation flagged as my own analysis, not a PTAB finding: the claimed subject matter is a computational quality-scoring heuristic (rank × click-derived popularity, summed, thresholded). That is the profile of art that routinely draws § 101 eligibility challenges in district court post-Alice. There is no PTAB or Federal Circuit precedent on this patent either way, so this is a theory to develop, not a citation to rely on.

Recommended next steps

1. Do not represent to a court or client that any claim of 8,762,389 has been invalidated. It hasn't. There is no FWD to link to, because none exists. The absence of PTAB activity is a factual statement, and it is the whole of the finding here.
2. Treat institution as still available. For an IPR, the § 315(b) one-year clock runs from service of a complaint alleging infringement of this patent — if you have been served, clock that date immediately. The PGR window (9 months from grant) closed on 2015-03-24 and the CBM transitional program has sunset, so IPR is the only AIA trial route still open for this patent. Statutory trial deadline: FWD due within 1 year of institution (§ 316(a)(11)), extendable up to 6 months for good cause.
3. Run the invalidity search now, from zero. Prior art cited on the face of the patent includes US 7,523,099 (Egnor et al.), US 7,630,972 (Ott et al.), US 7,984,048 (Curtis et al.), US 8,145,623 (Mehta et al.), and US 2009/0144262 (White et al.). Those are the references the examiner saw. The interesting art is the dense 2004–2011 literature on click-based relevance/CTR rank-correlation metrics and query rewriting — where the "sum of rank × click-popularity, thresholded" heuristic was likely well known. Because no estoppel exists, you can raise both the printed publications and the corresponding prior-art systems in the same case.
4. Check claim 1's threshold limitation for a § 112 written-description hook. The specification's own figures cut against a clean reading: in FIG. 4 the panel computes a revised-query score of 3.5 against an original-query score of 2.9 and calls the resulting −0.6 difference a bad synonym, while FIG. 5 treats a +0.6 difference as a good revision — and the pseudo-code discussion at ¶ [0048]-adjacent text describes "no impact" using the same "greater than" relation as "positive impact." The sign convention is at minimum worth probing on the merits. Again: my reading of the specification, not a Board holding.
5. Preserve § 101 and indefiniteness positions. "Specified threshold," "popularity score," and the ordering/comparison direction are all constructions to lock down early. With no prior Board construction to borrow, you have a free hand — and no adverse estoppel.

Confidence note: I have high confidence that no AIA trial proceeding exists on US 8,762,389 as of 2026-10-01, based on both the structured ODP block and multiple independent searches. I do not have high confidence that the patent has never been asserted in district court — my litigation search returned nothing, but that search was not exhaustive, and a complaint would not necessarily generate PTAB activity in any event. If you have a demand letter in hand, verify the assertion history through PACER/Docket Alarm before finalizing your strategy.

I'll research the assignment record and related litigation/ownership signals before writing this up.

Ownership & Assignment Reconstruction — US 8,762,389 B1

Patent: Synonym identification based on search quality — US 8,762,389 B1
Application: 13/365,181 · filed 2012-02-02 · granted 2014-06-24 · adjusted expiration 2032-02-13
Family: same-day sibling app 13/365,219 (→ US 8,812,518 B1) and continuation 14/300,487 filed 2014-06-10 (→ US 9,507,853 B1)

Sourcing caveat up front. The Google Patents legal-events record in the authoritative text supplies only dates and event types — it does not print reel/frame numbers. I was unable to complete a live pull of the per-assignment reel/frame and correspondent fields from USPTO Assignment Center in this session (tool step limit). Everything below that is not verifiable from the supplied record is labelled as such. I have not invented reel/frame numbers or correspondent names.

Inventors

Both were Google employees. The employment assignment of interest to Google Inc. was recorded 2012-02-18, 16 days after the 2012-02-02 filing (Google Patents legal events: "Assigned to GOOGLE INC. — ASSIGNMENT OF ASSIGNORS INTEREST … Assignors: POPOVICI, DAN, SPALEK, ROBERT"). That is a standard new-hire/employment invention-assignment catch-up filing, not a negotiated transfer.

Unusual-pattern check: No evidence of inventor departure within 12 months of filing. Robert Spalek is credited with a Google patent portfolio on third-party aggregator Patent Leaderboard (4 patents) and is a longstanding Google engineer; no adverse departure signal for Popovici was found either. No fire-sale precursor pattern detected. (Departure dates are not directly evidenced in the sources retrieved — treat as not observed, not as affirmatively disproven.)

Original assignee

Google Inc. (Mountain View, CA) at issuance — the assignee named on the face of the patent and confirmed by Justia's Patent History. Current owner of record: Google LLC, by change of name only.

• Primary line of business: internet search, advertising, and cloud/software services (Delaware corporation, originally incorporated 2002-10-22; converted to a Delaware LLC effective 2017-09-30).
• Product embodying the claims: Yes. The patent claims a method of judging proposed query revisions/synonyms by comparing aggregate "search results scores" (rank × popularity score) between an original and revised query — this is exactly the query-rewriting/synonym-scoring machinery behind Google Search. The specification's popularity scores (click-based user selections, rollover, dwell) are Google's own click-signal ranking infrastructure.
• Current status: Operating. Google Inc. was not acquired or dissolved — it underwent a statutory Delaware entity conversion on 2017-09-30 (Certificate of Conversion, Delaware SR 20176403570, file no. 3582691, filed 2017-09-29, effective 11:59 PM EDT 2017-09-30), subsequently wrapped under Alphabet Inc. Nothing in the chain reflects distress, divestiture, or a portfolio sale.

Assignment timeline

Two recorded events appear in the authoritative legal-events record. I could not retrieve the reel/frame numbers or correspondent-of-record for either from the live Assignment Center index in this session; those fields are stated as unavailable rather than guessed.

• 2012-02-02 (filing) / not a recorded assignment — Priority/doc-filing event

  • Conveyance: n/a (application filed by Google; inventors Popovici and Spalek)
  • Assignor: (inventors)
  • Assignee: Google Inc. (applicant of record)
  • Correspondent: not available
  • Context: initial filing.

• 2012-02-18 (executed; exact execution date not shown separately) / recorded 2012-02-18 — Reel/Frame not retrievable in this session

  • Conveyance: Assignment of assignors' interest (Google Patents: "ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)")
  • Assignor: Popovici, Dan; Spalek, Robert (the inventors)
  • Assignee: Google Inc.
  • Correspondent: not available — this is precisely the field I could not pull. Note that Google's capture filings of this era were made in bulk; the same correspondent firm recurs across thousands of Google assignments, so a single appearance would not by itself be a signal (see Signal 3).
  • Context: inventor-to-employer employment assignment (standard, confirmatory, 16 days post-filing).

• 2017-10-02 (Google Patents event date) / recorded 2017-10-02 per legal events — Reel/Frame not confirmed for this patent

  • Conveyance: Change of Name (Google Patents wording) — the underlying instrument is the Delaware Certificate of Conversion of Google Inc. to Google LLC, effective 2017-09-30
  • Assignor: Google Inc.
  • Assignee: Google LLC
  • Correspondent: not available
  • Context: internal reorganization / entity conversion only — the same legal person continues; no change in beneficial ownership. No consideration paid, no asset sale.
  • ⚠️ Related-record caveat: a TTAB Assignment Abstract of Title for a Google trademark shows "ENTITY CONVERSION," Reel/Frame 6178/0419, executed 2017-09-30, recorded 2017-10-10, assignor Google Inc. → assignee Google LLC (Delaware corp. → Delaware LLC), with the Mountain View 1600 Amphitheatre Parkway address. A separate legacy-assignment filing recorded 2022-04-06 at Reel 059514/Frames 0676–0680 re-files the "GOOGLE INC. TO LLC ENTITY CONVERSION" instrument together with an executed application assignment for an unrelated Google case. These confirm that Google recorded a single corporate-wide conversion instrument series (the 6178/xxxx reel, and later confirmatory re-recordings) rather than patent-by-patent deals. I have not verified that reel 6178/0419 or 059514/06xx covers US 8,762,389 specifically — do not cite it as this patent's reel/frame without pulling the Assignment Center record. Note also the label mismatch: Google Patents says "Change of Name," the recorded instrument is an "Entity Conversion"; those are the same event described two ways.

• No further assignments recorded. No transfer to any third party, no security interest, no license, no release, no correction.

No Assignment Center records beyond the above for this patent. The chain is: inventors → Google Inc. → Google LLC (name/entity change only). Google LLC remains the owner of record.

Timeline diagram

timeline
    title Ownership of US 8762389
    2012 : Filed by Popovici and Spalek
         : Assigned to Google Inc
    2014 : Patent issued as US 8762389
    2017 : Google Inc converts to Google LLC

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT. The only post-filing transfer is Google Inc. → Google LLC, an entity conversion of the same legal person (2017-09-30, Delaware file no. 3582691). No "IP/Patents/Licensing/Holdings/Ventures" suffix entity appears anywhere in the chain; no registered-agent-service address; Google LLC is the operating parent of the Google business.

2. Known asserter in the chain — NOT PRESENT. Neither Google Inc. nor Google LLC appears on any of the listed NPE rosters (Acacia, Marathon, IV, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, Spangenberg entities). No assignee in this chain was surfaced as a high-frequency plaintiff by Unified Patents or RPX in the material reviewed.

3. Repeat correspondent across the chain — UNASSESSABLE (data gap). I could not retrieve the correspondent of record for either recorded event. I therefore decline to name one. This signal should be re-checked directly in Assignment Center; it is the one field in this report that is genuinely missing rather than negative. Do not treat its absence here as a clean bill of health for this signal — it is an admitted hole.

4. Cascading transfers — NOT PRESENT. One substantive transfer (inventor→Google Inc.), plus one name change. Zero chained LLCs; zero transfers within 24 months of each other; total elapsed time from first to last recorded event is 5 years 7 months.

5. Pre-litigation transfer — NOT PRESENT. No infringement action naming US 8,762,389 was located in searches of litigation reporting covering this number; there is no first-suit date to measure against, and no transfer within 6 months of one. Google is a frequent defendant in patent cases but the record here shows no assertion of this patent.

6. Bankruptcy fire-sale — NOT PRESENT. Google Inc./Google LLC never filed Chapter 7 or 11; there was no §363 sale. (Contrast the Kodak/Nortel/Polaroid pattern invoked in the brief — none of that applies.)

7. Privateering — NOT PRESENT. No transfer to a third-party NPE asserting on Google's behalf; the patent never left the Google corporate family. No SEC 8-K/10-K divestiture disclosure is implicated because no divestiture occurred.

8. Defensive aggregator — NOT PRESENT. The chain does not terminate at RPX, AST, LOT Network, Unified Patents, or OIN; it terminates at the operating company. The patent has not been "neutralized" — it remains a live, owned asset of Google LLC with an adjusted expiration of 2032-02-13.

Verdict

Insufficient data — in the sense the brief defines it: the record consists of the original employment assignment plus a name change only.

Justification: the entire recorded chain is the 2012-02-18 inventor→Google Inc. assignment of assignors' interest (Popovici and Spalek) and the 2017-10-02 Google Inc.→Google LLC "Change of Name," which is the Delaware entity conversion effective 2017-09-30 (file no. 3582691) — the same legal person, no change in beneficial ownership and no consideration. No shell-entity transfer, no NPE assignee, no cascading LLC chain, no pre-litigation transfer, no bankruptcy sale, and no defensive-aggregator terminus exist anywhere in the record; Google LLC is the current owner. The genuine open item is Signal 3: the correspondent of record and exact reel/frame numbers were not retrieved, so the repeat-correspondent test should be re-run before this file is closed.

Verify directly: USPTO Assignment Center search — https://assignmentcenter.uspto.gov/ (or the older index at https://assignment.uspto.gov/patent/index.html) → search 8762389; and Google Patents legal events at https://patents.google.com/patent/[US8762389B1](/patent/US8762389B1)/en.

I'll research the prior-art citations for US 8,762,389 directly, since the "References Cited" section was truncated in the patent text provided to me.

I found the cited-references list via Justia. Let me pull the complete citation list and details on each reference.

I've gathered the citation data. Note: I hit my tool-call budget mid-research, so I'll clearly flag which items I verified versus which I could not complete.

Prior Art Analysis — U.S. Patent No. 8,762,389

1. Confirmation of the target document (literal ID check)

From the Google Patents and Justia records (both mirrors of USPTO bibliographic data):

Disambiguation note: this is 8,762,389, not 8,726,389, 7,873,389, or 6,233,389. (I covered that disambiguation in the litigation section; not repeated here.)

Critical family caveat (flagging a potential contradiction in the source record): The Google Patents event timeline for 8,762,389 lists cross-priority to US 13/365,219 → US 8,812,518 B1 (filed the same day, 2012-02-02), and to continuation US 14/300,487 → US 9,507,853 B1 (filed 2014-06-10). These are siblings in the same family, not prior art. This matters because:

• US 8,762,389's own abstract describes "evaluating the query revision rule," while its Summary describes "determining whether the second query term is a synonym for the first query term." The two siblings split those two aspects across different claim sets.
• The public claim text I retrieved in this session (via patents-review.com, URL path a/13365219-...) is for application 13/365,219 (US 8,812,518), not 13/365,181. I could not independently retrieve and verify the verbatim, as-issued claim set of 8,762,389 within my tool budget. Treat any claim mapping below as being to the common specification's disclosure, and verify the exact claim wording at the source before relying on it.

2. References cited on the face of US 8,762,389 (§ 56 listing)

Source: Justia "Referenced Cited / U.S. Patent Documents" for patent 8762389 — https://patents.justia.com/patent/8762389

The Justia record shows no foreign patent documents and no non-patent literature in the "Referenced Cited" block as retrieved — i.e., the § 56 listing appears to be five U.S. references only. I was unable to reach the USPTO Patent Center file wrapper to confirm whether additional art was cited and/or whether these were cited by the examiner versus by the applicant (Google Patents distinguishes these categories; Justia's extraction I retrieved does not). This is the single most important gap in the analysis — see § 5.

3. Per-reference analysis

For each reference I give the citation and dates that are confirmed, and then assess it against the two independent claim-1 aspects that the 8,762,389 specification discloses (per the patent text you supplied): (A) the synonym-identification aspect — obtain first results for a query with a first query term → score them (rank × popularity) → revise the query to add a second term → score the revised results → compare scores → deem the second term a synonym; and (B) the general query-revision-rule aspect — the same score-comparison loop where the revision is an arbitrary revision rule (e.g., adding quotation marks, adding exclusionary terms).

Reference 1 — US 7,523,099 B1, "Category suggestions relating to a search"

• Full citation: U.S. Patent No. 7,523,099 B1, Category suggestions relating to a search, granted Apr. 21, 2009 (Egnor et al.).
• Dates: Granted 2009-04-21 — prior art under pre-AIA § 102(b) as of the 2012-02-02 filing date (more than one year before filing).
• Description (verified title; content inferred): Assigned in the Google family of query-refinement patents; directed to generating category suggestions for a search (i.e., related-query / related-concept suggestion from a searchable corpus). URL: https://patents.google.com/patent/US7523099
• Anticipation assessment: Weak on its face. Its subject matter is category suggestion rather than scoring a corpus of search results by rank × popularity and comparing before/after aggregate scores. It would need to disclose the aggregate scoring-by-rank-and-popularity comparison to anticipate. My strong expectation, given the examiner's citation, is that it was used in a § 103 combination (query suggestion/refinement context), not as a standalone § 102 reference — but I cannot confirm the ground of rejection without the file wrapper. Flagged as unverified.
• Claims potentially affected: the query-revision/suggestion preamble of claim 1 (aspect B) at most.

Reference 2 — US 7,630,972 B1 (Ott et al.)

• Full citation: U.S. Patent No. 7,630,972 B1, granted Dec. 8, 2009 (Ott et al.).
• Dates: Granted 2009-12-08 → prior art under pre-AIA § 102(b).
• Description: ⚠️ I could not verify the title, assignee, or abstract within this session. I am not going to guess it.
• Anticipation assessment: Cannot be responsibly assessed without the reference's disclosure. Not assessed — verification required.
• Verification link: https://patents.google.com/patent/US7630972

Reference 3 — US 7,984,048 B2 (Curtis et al.)

• Full citation: U.S. Patent No. 7,984,048 B2, granted Jul. 19, 2011 (Curtis et al.).
• Dates: Granted 2011-07-19 → prior art under pre-AIA § 102(a) (after the applicant's invention date cannot be assumed) and, at latest, § 102(b) only if published/issued more than a year before 2012-02-02 — it is not (it issued ~6.5 months before filing), so § 102(a)/(e) is the operative basis, not § 102(b). This distinction matters for a pre-AIA case and is often mis-stated.
• Description: ⚠️ Not verified. My recollection is that the "Curtis" Google-family patent in this area concerns deriving linguistic/concept units from query logs, but I will not assert a title or assignee I could not confirm.
• Anticipation assessment: If it is in fact directed to extracting concept/synonym units from query logs, it would be relevant to the "search query is generated using queries stored in a query log" limitation of the dependent claims and to the synonym-generation concept — but not by itself to the aggregate-score-comparison core of claim 1. Not assessed — verification required.
• Verification link: https://patents.google.com/patent/US7984048

Reference 4 — US 8,145,623 B2 (Mehta et al.)

• Full citation: U.S. Patent No. 8,145,623 B2, granted Mar. 27, 2012 (Mehta et al.).
• Dates: ⚠️ Timing is the issue here. Granted 2012-03-27 — after the 2012-02-02 filing date. As a patent, it therefore cannot be § 102(a)/(b) art; it can only qualify under pre-AIA § 102(e) (a U.S. patent granted on an application filed before the applicant's invention date) — so its filing date, not its grant date, is what would need to be checked. This is a common trap and worth verifying in the file wrapper.
• Description: ⚠️ Not verified. Mehta-authored Google patents in this space typically relate to query-log data mining / synonym derivation.
• Anticipation assessment: Potentially the most substantive of the five if its disclosure covers deriving term relationships from query logs — but I cannot rate it as a § 102 reference without both the disclosure text and its effective filing date. Not assessed — verification required.
• Verification link: https://patents.google.com/patent/US8145623

Reference 5 — US 2009/0144262 A1 (White et al.)

• Full citation: U.S. Patent Application Publication No. 2009/0144262 A1, published Jun. 4, 2009 (White et al.).
• Dates: Published 2009-06-04 → prior art as a printed publication under pre-AIA § 102(b) if the filing date of 2012-02-02 is more than one year after publication — it is (2009-06-04 vs. 2012-02-02). ✅
• Description: ⚠️ Not verified. The "White" authorship in the query-reformulation literature (Ryen White et al., clickthrough/popular-destination work) is suggestive that this publication concerns using implicit user-interaction (click) data to evaluate or generate query refinements — which would map directly onto the patent's "popularity score … determined based on user selections … mouse rollover, a click, a click of at least a certain duration" language. I am flagging this as an inference, not a verified fact.
• Anticipation assessment: Modest-to-moderate, if the inference holds. A clickthrough-based query-refinement publication would supply the popularity-score element, but claim 1 also requires the aggregate position-score = rank × popularity, summed, and compared across original vs. revised queries against a threshold. Unless the publication discloses that comparison loop, it is § 103 material (combining the click-based popularity metric with a revision-evaluation loop), not § 102 anticipation.
• Verification link: https://patents.google.com/patent/US20090144262A1

4. The § 102 vs. § 103 point you should not skip

None of these five references, on the record I could retrieve, is a clean § 102 anticipation of the core claim. Anticipation requires a single reference disclosing every limitation, arranged as in the claim. The distinctive limitations of this patent are:

1. obtaining both an original-query result set and a revised-query result set;
2. computing an aggregate score per result set from rank × query-dependent popularity score, summed;
3. comparing the two aggregate scores and testing the difference against a specified threshold; and
4. automatically outputting a synonym verdict / "good revision" verdict without human feedback or rating (expressly claimed as an advantage).

Elements 1–3 are the natural home of a § 103 combination (a query-suggestion reference + a click-popularity reference + a ranking reference). The examiner's citation of five references, at least two of which (Egnor; the Ott/White lineage) look like query-suggestion/refinement rather than result-set-scoring art, is consistent with a § 103 rejection rather than a § 102 rejection.

Also note: because the application was filed 2012-02-02 — before the AIA first-inventor-to-file date of 2013-03-16 — pre-AIA §§ 102(a)/(b)/(e)/(g) govern, not AIA § 102(a)(1)/(a)(2). Getting this right changes which date (publication vs. filing) matters for each reference.

5. What I could not verify (explicit limitations)

These are gaps in my research, not findings:

1. Titles/abstracts/assignees of US 7,630,972, US 7,984,048, US 8,145,623, and US 2009/0144262 — not retrieved before my tool budget ran out. I have deliberately not guessed them.
2. Examiner-cited vs. applicant-cited and the actual ground of rejection (which reference was applied under § 102 and which under § 103, and what claims were rejected) — requires the file wrapper, which I could not open.
3. The verbatim, as-issued claim set of US 8,762,389 — the claim text widely indexed online under the title "Synonym identification based on search quality" belongs to sibling application 13/365,219 (US 8,812,518). Do not conflate.
4. US 8,145,623's effective filing date for a § 102(e) analysis.

Definitive verification steps (recommended):

• USPTO Patent Center file wrapper for 13/365,181: https://patentcenter.uspto.gov (search application 13/365,181) — pull the Examiner's search strategy, Notice of References Cited (PTO-892), and any Examiner's Reasons for Allowance.
• Google Patents "References Cited" with the "Examiner Cited / Applicant Cited" flags: https://patents.google.com/patent/[US8762389B1](/patent/US8762389B1)/en
• Justia: https://patents.justia.com/patent/8762389
• Individual references: https://patents.google.com/patent/US7523099, /US7630972, /US7984048, /US8145623, /US20090144262A1

6. Forward citations are not prior art (do not mix these up)

While retrieving the above I surfaced documents that cite US 8,762,389 (e.g., US 9,116,957; SG 11201406240W A, NEC Corp. "Synonym relation determination device"; the CN 107170447 family). These are later documents (post-2012/2014) and therefore have no § 102 or § 103 effect on 8,762,389. They matter only as evidence of the patent's subject-matter area, not as invalidating art.

Bottom line: The face of US 8,762,389 cites exactly five U.S. references — 7,523,099 (Egnor, confirmed), 7,630,972 (Ott), 7,984,048 (Curtis), 8,145,623 (Mehta), and 2009/0144262 (White) — all pre-dating the 2012-02-02 filing date in some form. Only US 7,523,099's substance is verified here; the other four must be read before any § 102 statement about them can honestly be made. On what I can assess, the art is more naturally a § 103 combination than a § 102 anticipation, because the core aggregate rank×popularity, before/after score-comparison, threshold-tested, no-human-feedback limitation is a specific combination that none of these five appears to disclose in a single reference.

I'll first try to locate the prior-art references cited on this patent's page, since the "Prior Art section" content wasn't included in the material passed to me.

Good — I found the examiner-cited references. Let me pull titles/details for each cited reference and locate the full citation table.

Obviousness Analysis — U.S. Patent No. 8,762,389 (“Synonym identification based on search quality”)

Current date: April 26, 2026 (per task header). Note the header date differs from the fetch date of the patent text (2026‑09‑24); I use the task date as “today.”

0. Flags on the previously generated sections (per instructions)

Two items from the earlier sections are changed or clarified by what I retrieved:

1. The “query revision rule” claim text is not 8,762,389’s. The claim set I could see in full (independent claim reciting “revising the search query using a query revision rule … identifying the query revision rule as a good revision rule,” dependent claim 8 “storing the query revision rule,” claim 9 non‑transitory CRM, claim 17 system) sits at patents-review.com/a/13365219-… — i.e., application 13/365,219, the same‑day sibling that issued as US 8,812,518 B1, not the patent at issue. The Justia record for 8,762,389 shows its claim 8 as “storing the second query term as a synonym for the first query term” (https://patents.justia.com/patent/8762389). So 8,762,389 is the synonym‑validation patent; 8,812,518 is the revision‑rule‑validation patent. This resolves the “exact claim number not verified” caveat in the earlier summary — and means my grounds below must be built on the synonym limitation set.
2. I still cannot read the granted claims of 8,762,389 verbatim, and the patent text furnished to me is still truncated before the claims. Everything below is therefore a claim‑scope‑informed prediction, not a validity opinion.

1. Framework applied

• §103 version: pre‑AIA §103(a) (effective filing 2012‑02‑02; AIA §3 applies only to applications filed on/after 2013‑03‑16).
• Governing law/rationales: Graham v. John Deere (scope/content/differences, PHOSITA, secondary considerations); KSR v. Teleflex (combinations of known elements; “obvious to try” with a finite number of predictable solutions). I apply the MPEP 2143 rationales: (A) combining prior art elements according to known methods; (B) simple substitution of one known element for another; (C) use of known technique to improve similar devices in the same way; (D) applying a known technique to a known device ready for improvement; (E) “obvious to try”; (F) design incentives/parameter optimization (In re Aller — no criticality in a “specified threshold”).
• PHOSITA: an engineer with a B.S./M.S. in CS/EE plus ~2–4 years building web search infrastructure, familiar with ranked retrieval, click logs, query rewriting/synonym expansion, and offline/online relevance evaluation metrics (position‑weighted precision, MAP, NDCG) and A/B testing of ranking changes.
• Claim scope (as best established): independent method claim (obtain first results for query with first term → aggregate score → revise query to add second term → obtain second results → aggregate score → compare and decide synonymy); parallel CRM and system claims; dependents adding rank information, popularity scores, query‑log source, position score = rank × popularity, summing position scores, difference‑vs‑threshold, and storing the synonym.
• Admitted prior art / patent’s own field: the specification states “search engines identify resources… a search engine may, for example, match query terms or synonyms of the query terms.” That is an admission that synonym‑based retrieval itself was known; the asserted novelty is the quality‑based evaluation step.

2. Prior‑art inventory (as retrieved)

Source for the examiner‑cited list: https://patents.justia.com/patent/8762389 (“Referenced Cited”: 7523099 Egnor; 7630972 Ott; 7984048 Curtis; 8145623 Mehta; 20090144262 White). Google Patents’ own “Prior art keywords” for this patent are search, search results, score, query, popularity (https://patents.google.com/patent/[US8762389B1](/patent/US8762389B1)/en).

3. Element‑by‑element mapping (independent synonym claim)

4. Proposed §103 grounds

Ground 1 (strongest): US 7,984,048 + US 7,630,972 + US 8,041,730

• 7,984,048 supplies query‑specific, click‑derived popularity with explicit awareness of rank position at selection time (element f).
• 7,630,972 supplies the aggregate set‑level score derived from per‑result relevance scores (elements b, e).
• 8,041,730 supplies scoring of substitute/synonym terms for query expansion (element c) and the notion of gating candidates by score (element h/i).
• Motivation (KSR A, C, D): all three are in the same field (search relevance engineering) and address the same problem the specification frames — deciding which query reformulations produce better results. A POSITA seeking to decide whether adding “plantain” to “banana smoothie” helps has a strong, articulated reason to (i) expand the query with a candidate substitute (8,041,730), (ii) measure the reformulated result set with a set‑level score (7,630,972), and (iii) base that score on click popularity (7,984,048) rather than human judgments, which were expensive and slow.
• Expectation of success: high — combining a relevance score, a set aggregation, and a candidate‑term expansion is mechanical in a field of predictable, numeric operations (MPEP 2143(A)).

Ground 2: US 7,984,048 + US 7,523,099 + US 7,630,972

Same mapping; 7,523,099/Egnor family supplies IR‑score generation and score combining, reinforcing elements (b)/(e). Motivation: routine use of a known score‑aggregation technique to improve a similar evaluation task (KSR rationale C).

Ground 3 (NPL‑anchored): Jones et al. (WWW 2006) “Generating query substitutions” + US 7,984,048 + US 7,630,972

Jones et al. is directly on point for generating and validating query substitutions from query logs and clicks; combined with the click‑popularity ranking of 7,984,048 and the aggregate scoring of 7,630,972 it renders the claim obvious. Motivation: replacing human relevance judgments with behavioral data was an express, known goal (and is the patent’s own stated advantage — an advantage that, notably, is not a claim limitation).

Ground 4 (evaluation‑metric framing): US 7,984,048 alone in view of Manning/Raghavan/Schütze (position‑weighted evaluation metrics)

Element (g) — summing rank × popularity — is the definition of a position‑weighted evaluation metric (closely related to average precision / DCG). Using a metric computed over a system’s own results to compare two candidate queries and pick the better one is standard offline evaluation methodology. KSR rationale D (known technique applied to a known device ready for improvement).

Ground 5: Threshold and transform limitations (dependents / spec variants)

“Difference … satisfies a specified threshold,” “rank raised to a power between 0 and 1,” “square root of rank,” and “popularity capped at a threshold” are parameter selections with no disclosed criticality — obvious to optimize (In re Aller, In re Boesch; KSR). Dependent “storing the synonym/rule” is a data‑persistence step routinely performed once a synonym is validated.

5. Where the obviousness case is vulnerable (defense side)

1. No single reference, on the record I can see, expressly compares two aggregate click‑based scores of the same query before and after a synonym substitution to conclude synonymy. The examiner evidently considered the art insufficient — that gap is the patent’s whole foothold, and a patent owner will argue no reference “suggests the comparison step” and that the prior art (e.g., 7,984,048’s demographic/pageview correlation) is directed to ordinary ranking, not synonym validation. A strong rebuttal is KSR “obvious to try”: one metric, a finite set of candidate substitutions, predictable results.
2. Motivation must be articulated with record evidence. My mappings for 7,523,099, 8,145,623, 2009/0144262, and 8,041,730 are title/bibliographic‑level only in this session. Any real Office action or IPR petition would need full‑text pin cites (column/line) — which I could not obtain here.
3. Secondary considerations: none apparent. Google’s commercial success in search would not, without a nexus, be attributable to this specific synonym‑validation feature.
4. Literal‑scope oddity noted earlier — the FIG. 2 pseudo‑code states the “positive impact” and “no impact” conditions identically and inconsistently with the higher‑is‑better math, and FIG. 4’s worked example computes 2.9 vs. 3.5 as “−0.6 … worse.” This is a disclosure/claim‑support risk for any claim requiring a particular polarity, but it helps an obviousness attack on claims broad enough to read on either polarity, since they reduce to “compare and threshold.”

6. Bottom line

• Likely §103 conclusion: the claims are reasonably likely obvious over US 7,984,048 in view of US 7,630,972 and US 8,041,730 (“Substitute term scoring”), with US 7,523,099 (Egnor) as a secondary scoring reference and the clickthrough/query‑substitution NPL (Joachims 2002; Jones et al. 2006; Agichtein et al. 2006) supplying the “no human raters” motivation. Confidence: moderate. The motivation and predictability arguments are strong; the decisive weakness is that I could not read the full text of any cited reference here, so I cannot supply the column/line pin cites a rejection or petition requires.
• Corrections to prior sections: (i) the “query revision rule” independent claim text belongs to sibling application 13/365,219 / US 8,812,518, not to 8,762,389; (ii) the examiner‑cited five references are confirmed by number, inventor and date, but US 8,145,623 and US 2009/0144262 A1 subject matter remains unverified; (iii) the internal “positive/no‑impact” inconsistency in the spec remains as flagged and is a claim‑support risk rather than an obviousness bar.
• Not established here: any litigation (consistent with the earlier section — none found, still “not found,” not “verified none”), and the granted claim language of 8,762,389, which should be pulled from USPTO PatentCenter before relying on this analysis in any proceeding.