# Entity Match Verification — Task Prompt

**Feed this document as a single user message** after loading the system prompt
from `verify-agent-prompt.md`. The agent receives all 80 entries in one context
window and works through them sequentially, issuing web searches and producing
a verdict per entry.

---

## Background

You are auditing the output of the **WIMS entity mapper** — an automated pipeline
that links raw company name strings from trademark and patent filings to structured
company records in a curated database.

### Why this data exists

Every trademark filing and every patent application names one or more legal
entities: the trademark holder or the patent applicant. These names are free-text
as submitted to the IP office — they may be abbreviated, misspelled, written in a
foreign script, include trailing punctuation, or use a subsidiary/DBA name that
differs from the parent's registered name. The WIMS database holds ~330,000 such
raw names spanning trademarks and patents from dozens of countries.

The entity mapper runs a cascade of matching methods to link each raw name to a
**target company record** — a normalised, enriched company entry with official name,
country, aliases (from Wikidata/GLEIF), and sector tags. Getting these links right
matters: the downstream graph database uses them to attribute IP portfolios to
specific companies, which drives market intelligence and licensing decisions.

### The pipeline methods (in cascade order)

| Method | What it does | Typical precision |
|---|---|---|
| `exact` | Normalised names match character-for-character | ~99% |
| `lei` | Match via GLEIF Legal Entity Identifier code lookup | ~90% |
| `dba` | Source name appears verbatim in target's alias list | ~95% |
| `fuzzy (high)` | Token-overlap similarity ≥ 0.85 | ~90% |
| `fuzzy (low)` | Token-overlap similarity 0.675–0.849 | ~70% estimated |
| `ai_confirmed` | Local LLM (Mistral v3) confirmed, pipeline score ≥ 0.75 | ~85% estimated |
| `ai_rejected` | LLM scored below threshold — pipeline treats as NO match | varies |

The **80 entries below** are a stratified random sample: 10 per method bucket,
selected to give you a representative cross-section for calibration. The goal is
to establish actual precision per bucket and catch systematic failure modes.

### What the confidence score means

- For `exact`: 1.00 = normalised strings identical; 0.90 = one minor difference
- For `lei`: GLEIF confidence in the name–LEI association (0.786–0.920)
- For `dba`: fixed at 0.95 when source name found verbatim in alias list
- For `fuzzy`: raw token-similarity ratio (0.675–0.945)
- For `ai_confirmed`: pipeline score = 0.30 + raw\_model\_conf × 0.55
  - score 0.850 → raw Mistral confidence 0.9–1.0 (near-certain)
  - score 0.823 → raw confidence ~0.95 (high)
  - score 0.795 → raw confidence 0.7–0.8 (uncertain/borderline)
- For `ai_rejected`: score shown is the filtered score (0.685 = raw conf 0.7, below
  MIN\_SCORE=0.75); `rejected` means the model returned `match: false` outright

### Known risk areas to watch for

- **Entry 13** — "New Health Sciences" → "Usana Health Sciences": different brand
  names; LEI match implies a corporate relationship but source may be a subsidiary
- **Entry 19/62/76** — targets named "International Trade Co., Ltd.": extremely
  generic — high risk of being a different unrelated company with the same name
- **Entries 41–50** (fuzzy low): highest expected false-positive rate — treat
  confidently only if web evidence is strong
- **Entries 61–70** (AI borderline 0.795): model was uncertain; some may be
  correct matches the model was too conservative on
- **Entries 71–80** (AI rejected 0.685): model said "uncertain lean-reject" —
  some may actually be correct matches (false negatives)

---

## Your task

Process each of the 80 entries below **in order**. For each one:

1. Use `web_search` to find external evidence (up to 6 queries per entry).
2. Determine whether the source entity and target company are the same real-world
   legal entity (or one is a subsidiary, branch, or DBA of the other).
3. Output a verdict using the JSON schema from the system prompt.

**Efficiency note**: for `exact` entries 1–10, you may apply expedited handling —
if the names differ only by trailing punctuation or legal suffix and the sector
is plausible, one confirmatory search is sufficient. Reserve deep searches for
fuzzy-low, AI-borderline, AI-rejected, and any entry that looks suspicious on
first read.

**Flag anything unusual** even if you can't resolve it: natural persons, dissolved
companies, name collisions between unrelated entities, or suspiciously generic
target company names.

---

## Output format

After processing all 80 entries produce a summary table at the end:

```
## Summary

| # | Source | Target | Pipeline | My Verdict | Conf. | Correct? |
|---|--------|--------|----------|------------|-------|----------|
| 1 | Korea Electronics Technology Institute | Korea Electronics Technology Institute | exact 1.000 | match | 0.98 | true |
...
```

Then a **Per-bucket analysis** section:

```
## Per-bucket analysis

| Bucket | Checked | Match | No Match | Uncertain | Pipeline correct |
|--------|---------|-------|----------|-----------|-----------------|
| exact  | 10      | ...   | ...      | ...       | X/10            |
...
```

And a **Key findings** section (3–8 bullet points) noting any systematic patterns,
suspected false-positive clusters, or cases that warrant further investigation.

---

## Entries

### Entry 1 — Exact Match
Source entity: "Korea Electronics Technology Institute,"  [country: KR]  [source type: patent_applicant]
Target company: "Korea Electronics Technology Institute"
  Original name: 전자부품연구원
  Also known as: electronics institute korea technology
Pipeline verdict: exact @ confidence 1.000

---

### Entry 2 — Exact Match
Source entity: "Maatel"  [country: FR]  [source type: patent_applicant]
Target company: "Maatel SAS"
  Also known as: maatel
Pipeline verdict: exact @ confidence 1.000

---

### Entry 3 — Exact Match
Source entity: "Njoy"  [country: US]  [source type: patent_applicant]
Target company: "NJOY, LLC"
  Also known as: njoy
Pipeline verdict: exact @ confidence 1.000

---

### Entry 4 — Exact Match
Source entity: "Umc Electronics Co"  [country: JP]  [source type: patent_applicant]
Target company: "UMC Electronics Co., Ltd."
  Also known as: electronics umc
Pipeline verdict: exact @ confidence 1.000

---

### Entry 5 — Exact Match
Source entity: "Jft Co Ltd"  [country: KR]  [source type: patent_applicant]
Target company: "JFT Co., Ltd."
  Also known as: jft
Pipeline verdict: exact @ confidence 1.000

---

### Entry 6 — Exact Match
Source entity: "CELSIUS HOLDINGS, INC."  [country: US]  [source type: trademark_holder]
Target company: "Celsius Holdings, Inc."
  Also known as: celsius
Pipeline verdict: exact @ confidence 1.000

---

### Entry 7 — Exact Match
Source entity: "Ignite International, Ltd"  [country: US]  [source type: trademark_holder]
Target company: "Ignite International, Ltd."
  Also known as: ignite, enterprises ignite us vulcan
Pipeline verdict: exact @ confidence 1.000

---

### Entry 8 — Exact Match
Source entity: "Xolo Ltd"  [country: GB]  [source type: patent_applicant]
Target company: "Xolo Ltd."
  Also known as: xolo
Pipeline verdict: exact @ confidence 1.000

---

### Entry 9 — Exact Match
Source entity: "PAX Labs, Inc"  [country: US]  [source type: trademark_holder]
Target company: "PAX Labs, Inc."
  Also known as: labs pax
Pipeline verdict: exact @ confidence 1.000

---

### Entry 10 — Exact Match
Source entity: "Kiva Brands Inc."  [country: US]  [source type: trademark_holder]
Target company: "Kiva Brands, Inc."
  Also known as: brands kiva
Pipeline verdict: exact @ confidence 1.000

---

### Entry 11 — LEI Match
Source entity: "American Tobacco Cy"  [country: GB]  [source type: patent_applicant]
Target company: "British American Tobacco Group"
  Also known as: american british tobacco
Pipeline verdict: lei @ confidence 0.828

---

### Entry 12 — LEI Match
Source entity: "Breathe Technologies,"  [country: DE]  [source type: patent_applicant]
Target company: "Breathe Solutions GmbH"
  Also known as: breathe solutions
Pipeline verdict: lei @ confidence 0.920

---

### Entry 13 — LEI Match
Source entity: "New Health Sciences Inc,"  [country: US]  [source type: patent_applicant]
Target company: "Usana Health Sciences, Inc."
  Also known as: health sciences usana
Pipeline verdict: lei @ confidence 0.920

---

### Entry 14 — LEI Match
Source entity: "Te Connectivity Poland Ltd"  [country: CH]  [source type: patent_applicant]
Target company: "TE Connectivity Ltd."
  Also known as: connectivity te, electronics tyco
Pipeline verdict: lei @ confidence 0.920

---

### Entry 15 — LEI Match
Source entity: "Polymer Technology Inc us"  [country: ?]  [source type: patent_applicant]
Target company: "Polymer Technology Systems, Inc."
  Also known as: polymer systems technology
Pipeline verdict: lei @ confidence 0.828

---

### Entry 16 — LEI Match
Source entity: "Boehringer Ingelheim Internaitonal Gmbh"  [country: DE]  [source type: patent_applicant]
Target company: "Boehringer Ingelheim International GmbH."
  Also known as: boehringer ingelheim
Pipeline verdict: lei @ confidence 0.920

---

### Entry 17 — LEI Match
Source entity: "Philips North America Llc,"  [country: US]  [source type: patent_applicant]
Target company: "Philips RS North America LLC"
  Also known as: america north philips rs
Pipeline verdict: lei @ confidence 0.920

---

### Entry 18 — LEI Match
Source entity: "Analog Devices Technology,"  [country: US]  [source type: patent_applicant]
Target company: "Analog Devices, Inc."
  Also known as: analog devices
Pipeline verdict: lei @ confidence 0.828

---

### Entry 19 — LEI Match
Source entity: "Wenzhou Tuotian International Trade Co Ltd"  [country: CN]  [source type: patent_applicant]
Target company: "International Trade Co., Ltd."
  Also known as: international trade
Pipeline verdict: lei @ confidence 0.828

---

### Entry 20 — LEI Match
Source entity: "Philip Mrris Products Sa"  [country: CH]  [source type: patent_applicant]
Target company: "Philip Morris Products S.A."
  Also known as: morris philip products, de fabriques reunies tabac
Pipeline verdict: lei @ confidence 0.920

---

### Entry 21 — DBA / Also Known As
Source entity: "บริษัท สหทรัพย์88 จำกัด"  [country: GB]  [source type: trademark_holder]
Target company: "VN Labs Ltd."
  Also known as: labs vn, 88, cigarettes e kik, nation vape
Pipeline verdict: dba @ confidence 0.950

---

### Entry 22 — DBA / Also Known As
Source entity: "Pure Corporation"  [country: US]  [source type: patent_applicant]
Target company: "House of Fun Sales, LLC"
  Also known as: fun house of sales, diffuser pure, pure
Pipeline verdict: dba @ confidence 0.950

---

### Entry 23 — DBA / Also Known As
Source entity: "China Tobacco Guangdong Industrial Corporation"  [country: KH]  [source type: patent_applicant]
Target company: "Viniton Group Co., Ltd. (Asean International Co., Ltd.)"
  Also known as: asean co co group international ltd ltd viniton, china guangdong industrial tobacco
Pipeline verdict: dba @ confidence 0.950

---

### Entry 24 — DBA / Also Known As
Source entity: "ETZ Hayim Holdings LLC"  [country: US]  [source type: trademark_holder]
Target company: "ETZ Hayim Holding, S.P.C."
  Also known as: c etz hayim holding p s, etz hayim
Pipeline verdict: dba @ confidence 0.950

---

### Entry 25 — DBA / Also Known As
Source entity: "Tobacco Industry Co Ltd"  [country: NL]  [source type: patent_applicant]
Target company: "De Dunes Holding B.V."
  Also known as: b de dunes holding v, tobacco
Pipeline verdict: dba @ confidence 0.950

---

### Entry 26 — DBA / Also Known As
Source entity: "Tcl移动通信科技（宁波）有限公司"  [country: CN]  [source type: patent_applicant]
Target company: "TCL Technology"
  Original name: TCL科技集团股份有限公司
  Also known as: tcl technology, tcl
Pipeline verdict: dba @ confidence 0.950

---

### Entry 27 — DBA / Also Known As
Source entity: "Mcc Corp"  [country: US]  [source type: patent_applicant]
Target company: "Micro Commercial Components Corporation"
  Also known as: commercial components micro, mcc
Pipeline verdict: dba @ confidence 0.950

---

### Entry 28 — DBA / Also Known As
Source entity: "บริษัท เบอร์แทรม(1958) จำกัด"  [country: TH]  [source type: trademark_holder]
Target company: "Bertram (1958) Co., Ltd."
  Original name: บริษัท เบอร์แทรม (1958) จํากัด
  Also known as: 1958 bertram, 1958
Pipeline verdict: dba @ confidence 0.950

---

### Entry 29 — DBA / Also Known As
Source entity: "CAMPBELL SOUP COMPANY"  [country: US]  [source type: trademark_holder]
Target company: "THE CAMPBELL'S COMPANY"
  Also known as: campbell s the, campbells, campbell soup
Pipeline verdict: dba @ confidence 0.950

---

### Entry 30 — DBA / Also Known As
Source entity: "Synapse LLC"  [country: US]  [source type: trademark_holder]
Target company: "Synapse Lifestyle, LLC"
  Also known as: lifestyle synapse, synapse
Pipeline verdict: dba @ confidence 0.950

---

### Entry 31 — Fuzzy — High Confidence (≥0.85)
Source entity: "Hubei China Tobacco Industry Co Ltd"  [country: CN]  [source type: patent_applicant]
Target company: "China Tobacco Hubei Industrial Co., Ltd."
  Original name: 湖北中烟工业有限责任公司
  Also known as: china hubei industrial tobacco
Pipeline verdict: fuzzy (high confidence) @ confidence 0.900

---

### Entry 32 — Fuzzy — High Confidence (≥0.85)
Source entity: "Knoninklijke Philips Electronics N V"  [country: NL]  [source type: patent_applicant]
Target company: "Koninklijke Philips N.V."
  Also known as: koninklijke n philips v
Pipeline verdict: fuzzy (high confidence) @ confidence 0.900

---

### Entry 33 — Fuzzy — High Confidence (≥0.85)
Source entity: "Shenzhen Dingli Electromechanical Automatic Co Ltd"  [country: CN]  [source type: patent_applicant]
Target company: "Shenzhen Dingli Electromechanical Automation Co., Ltd."
  Original name: 深圳市鼎力机电自动化有限公司
  Also known as: automation dingli electromechanical shenzhen
Pipeline verdict: fuzzy (high confidence) @ confidence 0.900

---

### Entry 34 — Fuzzy — High Confidence (≥0.85)
Source entity: "British-American Tobacco (Holdings) Ltd"  [country: GB]  [source type: patent_applicant]
Target company: "British American Tobacco Brands Ltd."
  Also known as: american brands british tobacco
Pipeline verdict: fuzzy (high confidence) @ confidence 0.945

---

### Entry 35 — Fuzzy — High Confidence (≥0.85)
Source entity: "Lite-on Electronics Guangzhou Inc"  [country: TW]  [source type: patent_applicant]
Target company: "LITE-ON Technology Corporation"
  Also known as: lite on technology
Pipeline verdict: fuzzy (high confidence) @ confidence 0.900

---

### Entry 36 — Fuzzy — High Confidence (≥0.85)
Source entity: "Dongguan Jingliangteng New Material Tech Co Ltd"  [country: CN]  [source type: patent_applicant]
Target company: "Dongguan Jingliangteng New Material Technology Co., Ltd."
  Original name: 东莞市晶亮腾新材料科技有限公司
  Also known as: dongguan jingliangteng material new technology
Pipeline verdict: fuzzy (high confidence) @ confidence 0.900

---

### Entry 37 — Fuzzy — High Confidence (≥0.85)
Source entity: "Shenzhen Hong Zhun Tech Co Ltd"  [country: CN]  [source type: patent_applicant]
Target company: "Shenzhen Hong-Zhun Technology Co., Ltd."
  Original name: 深圳市宏准科技有限公司
  Also known as: hong shenzhen technology zhun
Pipeline verdict: fuzzy (high confidence) @ confidence 0.900

---

### Entry 38 — Fuzzy — High Confidence (≥0.85)
Source entity: "British American Tobacco (Investments) Ltd"  [country: GB]  [source type: patent_applicant]
Target company: "British American Tobacco Group"
  Also known as: american british tobacco
Pipeline verdict: fuzzy (high confidence) @ confidence 0.945

---

### Entry 39 — Fuzzy — High Confidence (≥0.85)
Source entity: "Shenzhen Aratek Biometrics Technology Co Ltd"  [country: TW]  [source type: patent_applicant]
Target company: "Aratek Biometrics Technology Co., Ltd."
  Also known as: aratek biometrics technology
Pipeline verdict: fuzzy (high confidence) @ confidence 0.900

---

### Entry 40 — Fuzzy — High Confidence (≥0.85)
Source entity: "Imperial Tobaco Ltd"  [country: GB]  [source type: patent_applicant]
Target company: "Imperial Tobacco Limited"
  Also known as: imperial tobacco
Pipeline verdict: fuzzy (high confidence) @ confidence 0.900

---

### Entry 41 — Fuzzy — Low Confidence (<0.85)
Source entity: "Zoom Information Systems"  [country: CN]  [source type: patent_applicant]
Target company: "Guangzhou Zemu Information Technology Co., Ltd."
  Also known as: guangzhou information technology zemu
Pipeline verdict: fuzzy (low confidence) @ confidence 0.750

---

### Entry 42 — Fuzzy — Low Confidence (<0.85)
Source entity: "Church & Dwight"  [country: US]  [source type: patent_applicant]
Target company: "Church & Dwight Co., Inc."
  Also known as: church dwight
Pipeline verdict: fuzzy (low confidence) @ confidence 0.750

---

### Entry 43 — Fuzzy — Low Confidence (<0.85)
Source entity: "Alpha & Omega Semiconductor Inc"  [country: JP]  [source type: patent_applicant]
Target company: "ROHM Semiconductor Co., Ltd."
  Also known as: rohm semiconductor
Pipeline verdict: fuzzy (low confidence) @ confidence 0.750

---

### Entry 44 — Fuzzy — Low Confidence (<0.85)
Source entity: "Te Connectivity Tech Sip Ltd"  [country: CH]  [source type: patent_applicant]
Target company: "TE Connectivity Ltd."
  Also known as: connectivity te, electronics tyco
Pipeline verdict: fuzzy (low confidence) @ confidence 0.750

---

### Entry 45 — Fuzzy — Low Confidence (<0.85)
Source entity: "Forrest Kevin us"  [country: HU]  [source type: patent_applicant]
Target company: "Kevin Forest, Kft."
  Original name: Kevin Forest Termelő, Szolgáltató és Kereskedelmi Korlátolt Felelősségű Társaság
  Also known as: forest kevin, es felelossegu forest kereskedelmi kevin korlatolt szolgaltato tarsasag termelo
Pipeline verdict: fuzzy (low confidence) @ confidence 0.750

---

### Entry 46 — Fuzzy — Low Confidence (<0.85)
Source entity: "Agrima Scient Corp"  [country: CA]  [source type: patent_applicant]
Target company: "Agrima Scientific Corp."
  Also known as: agrima scientific
Pipeline verdict: fuzzy (low confidence) @ confidence 0.787

---

### Entry 47 — Fuzzy — Low Confidence (<0.85)
Source entity: "Flat Connections Inc"  [country: US]  [source type: patent_applicant]
Target company: "Taste Connections, LLC"
  Also known as: connections taste
Pipeline verdict: fuzzy (low confidence) @ confidence 0.750

---

### Entry 48 — Fuzzy — Low Confidence (<0.85)
Source entity: "Knoll Aktiengesellschaft"  [country: DE]  [source type: patent_applicant]
Target company: "Hanf Aktiengesellschaft"
  Also known as: aktiengesellschaft hanf
Pipeline verdict: fuzzy (low confidence) @ confidence 0.750

---

### Entry 49 — Fuzzy — Low Confidence (<0.85)
Source entity: "Fast Eddies Vape Shop And Lounge Llc us"  [country: US]  [source type: patent_applicant]
Target company: "Fast Eddie's Vape Shop And Lounge, LLC"
  Also known as: and eddie fast lounge s shop vape
Pipeline verdict: fuzzy (low confidence) @ confidence 0.750

---

### Entry 50 — Fuzzy — Low Confidence (<0.85)
Source entity: "Shenzhen Aowei Semiconductor Co Ltd"  [country: US]  [source type: patent_applicant]
Target company: "ON Semiconductor Corporation"
  Also known as: on semiconductor
Pipeline verdict: fuzzy (low confidence) @ confidence 0.750

---

### Entry 51 — AI Confirmed — High (score ≥0.84)
Source entity: "HI TECH NATURAL PRODUCTS INDIA LTD"  [country: GY]  [source type: trademark_holder]
Target company: "Hi-Tech Natural Products (India) Limited"
  Also known as: hi india natural products tech
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 52 — AI Confirmed — High (score ≥0.84)
Source entity: "JOHNSON & JOHNSON"  [country: TD]  [source type: trademark_holder]
Target company: "Johnson & Johnson, Inc."
  Also known as: johnson johnson
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 53 — AI Confirmed — High (score ≥0.84)
Source entity: "DanCzek Teplice a.s."  [country: DK]  [source type: trademark_holder]
Target company: "DanCzek Teplice A.S."
  Also known as: a danczek s teplice
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 54 — AI Confirmed — High (score ≥0.84)
Source entity: "Societé Des Produits Nestlé S.A"  [country: PR]  [source type: trademark_holder]
Target company: "Société des Produits Nestlé S.A."
  Also known as: des nestle produits societe, nestle products, nestle produkte, nestle prodotti
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 55 — AI Confirmed — High (score ≥0.84)
Source entity: "Shenzhen Hailinda Technology Co., Ltd."  [country: CC]  [source type: trademark_holder]
Target company: "Shenzhen Hailinda Technology Co., Ltd."
  Original name: 深圳市海玲达科技有限公司
  Also known as: hailinda shenzhen technology
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 56 — AI Confirmed — High (score ≥0.84)
Source entity: "ORBITEL INTERNATIONAL CORP."  [country: MF]  [source type: trademark_holder]
Target company: "Orbitel International Corp."
  Also known as: orbitel
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 57 — AI Confirmed — High (score ≥0.84)
Source entity: "Laperva L.L.C."  [country: WS]  [source type: trademark_holder]
Target company: "Laperva L.L.C."
  Also known as: c l l laperva
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 58 — AI Confirmed — High (score ≥0.84)
Source entity: "DANONE"  [country: FI]  [source type: trademark_holder]
Target company: "Danone Inc"
  Also known as: danone
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 59 — AI Confirmed — High (score ≥0.84)
Source entity: "Shenzhen Noto-Tech Electronics Co., Ltd."  [country: CC]  [source type: trademark_holder]
Target company: "Shenzhen Noto-Tech Electronics Co., Ltd."
  Original name: 深圳诺泰科电子有限公司
  Also known as: electronics noto shenzhen tech
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 60 — AI Confirmed — High (score ≥0.84)
Source entity: "Acu-Flow Limited"  [country: SM]  [source type: trademark_holder]
Target company: "Acu-Flow Limited"
  Also known as: acu flow
Pipeline verdict: ai_confirmed (high) @ confidence 0.850

---

### Entry 61 — AI Confirmed — Borderline (score ~0.795)
Source entity: "AU PAYS DU THE"  [country: FI]  [source type: trademark_holder]
Target company: "Au Pays Du The"
  Also known as: au du pays the
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 62 — AI Confirmed — Borderline (score ~0.795)
Source entity: "Qingdao Century International Trade Co., Ltd"  [country: CC]  [source type: trademark_holder]
Target company: "International Trade Co., Ltd."
  Also known as: international trade
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 63 — AI Confirmed — Borderline (score ~0.795)
Source entity: "Beijing XINYOULINGXI Science and, Technology Co., Ltd."  [country: CC]  [source type: trademark_holder]
Target company: "Beijing Xinyoulingxi Technology Co., Ltd."
  Original name: 北京新有灵犀科技有限公司
  Also known as: beijing technology xinyoulingxi
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 64 — AI Confirmed — Borderline (score ~0.795)
Source entity: "M/S. SOM FRAGRANCES PVT LTD"  [country: GY]  [source type: trademark_holder]
Target company: "Som Fragrances Pvt. Ltd."
  Also known as: fragrances som
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 65 — AI Confirmed — Borderline (score ~0.795)
Source entity: "Better Days Wellness Group LLC"  [country: US]  [source type: trademark_holder]
Target company: "Better Way Health, LLC"
  Also known as: better health way
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 66 — AI Confirmed — Borderline (score ~0.795)
Source entity: "British American Tobacco, Limitada"  [country: MG]  [source type: trademark_holder]
Target company: "British American Tobacco Group"
  Also known as: american british tobacco
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 67 — AI Confirmed — Borderline (score ~0.795)
Source entity: "Novadelta - Comércio e Industria de Cafés, S.A."  [country: MA]  [source type: trademark_holder]
Target company: "NOVADELTA COMERCIO E INDUSTRIA DE CAFES, LDA"
  Also known as: cafes comercio de e industria lda novadelta
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 68 — AI Confirmed — Borderline (score ~0.795)
Source entity: "MDM"  [country: FI]  [source type: trademark_holder]
Target company: "Eliquidstop, Inc."
  Also known as: eliquidstop, mdm
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 69 — AI Confirmed — Borderline (score ~0.795)
Source entity: "Vape.io Inc"  [country: US]  [source type: trademark_holder]
Target company: "ARIO VAPE, LLC"
  Also known as: ario vape
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 70 — AI Confirmed — Borderline (score ~0.795)
Source entity: "DGB (PROPRIETARY) LIMITED"  [country: CA]  [source type: trademark_holder]
Target company: "Proprietary Nutritionals, Inc."
  Also known as: nutritionals proprietary
Pipeline verdict: ai_confirmed (borderline) @ confidence 0.795

---

### Entry 71 — AI Rejected
Source entity: "Shanghai Benefisha Industrial Co., Ltd."  [country: CC]  [source type: trademark_holder]
Target company: "Shanghai Industrial Holdings Ltd."
  Original name: 上海實業控股有限公司
  Also known as: industrial shanghai
Pipeline verdict: ai_rejected @ confidence 0.685

---

### Entry 72 — AI Rejected
Source entity: "CHINA TOBACCO GUANGDONG INDUSTRIAL CO,, LTD."  [country: CC]  [source type: trademark_holder]
Target company: "China Tobacco Hunan Industrial Co., Ltd."
  Original name: 湖南中烟工业有限责任
  Also known as: china hunan industrial tobacco
Pipeline verdict: ai_rejected @ confidence 0.685

---

### Entry 73 — AI Rejected
Source entity: "TOKIO INDUSTRY (M) SDN BHD"  [country: LA]  [source type: trademark_holder]
Target company: "Kardinal Industries (M) Sdn. Bhd."
  Also known as: bhd industries kardinal m sdn, bhd epod global sdn station
Pipeline verdict: ai_rejected @ confidence 0.685

---

### Entry 74 — AI Rejected
Source entity: "BLACK INSOMNIA, COFFEE COMPANY LIMITED"  [country: US]  [source type: trademark_holder]
Target company: "Black Rifle Coffee Company LLC"
  Also known as: black coffee rifle
Pipeline verdict: ai_rejected @ confidence 0.685

---

### Entry 75 — AI Rejected
Source entity: "S-LAB Sp. z o.o."  [country: MS]  [source type: trademark_holder]
Target company: "Elite Lab Sp. z o.o. Sp.k."
  Also known as: elite k lab o o sp sp z
Pipeline verdict: ai_rejected @ confidence 0.685

---

### Entry 76 — AI Rejected
Source entity: "Guangzhou Picky Cat International Trade Co., Ltd."  [country: CC]  [source type: trademark_holder]
Target company: "International Trade Co., Ltd."
  Also known as: international trade
Pipeline verdict: ai_rejected @ confidence 0.685

---

### Entry 77 — AI Rejected
Source entity: "CLOUD NINE MARKETING PVT. LTD."  [country: GY]  [source type: trademark_holder]
Target company: "Cloud Nine International Ltd."
  Also known as: cloud nine
Pipeline verdict: ai_rejected @ confidence 0.685

---

### Entry 78 — AI Rejected
Source entity: "V-NION FOOD INDUSTRIES SDN BHD"  [country: LA]  [source type: trademark_holder]
Target company: "Kardinal Industries (M) Sdn. Bhd."
  Also known as: bhd industries kardinal m sdn, bhd epod global sdn station
Pipeline verdict: ai_rejected @ confidence rejected

---

### Entry 79 — AI Rejected
Source entity: "Clinical Nutra LLC"  [country: CA]  [source type: trademark_holder]
Target company: "Nutra Canada Inc."
  Also known as: canada nutra
Pipeline verdict: ai_rejected @ confidence 0.685

---

### Entry 80 — AI Rejected
Source entity: "PHILIP MORRIS PRODUSTS S.A."  [country: PR]  [source type: trademark_holder]
Target company: "Philip Morris International, Inc."
  Also known as: morris philip
Pipeline verdict: ai_rejected @ confidence 0.685

---

*End of task. Process entries 1–80 sequentially, then output the summary table and per-bucket analysis.*