Methodology · v2.8

IISInversión en innovación

Innovation-investment signal for {legal_name} across three observable dimensions. EXTRACTION RULES: type MUST be "composite_signal" — NEVER "absent" for public companies that file an annual regulator report (10-K in the US or equivalent; any listed mega-cap has all three signals available). Return composite_signal with one sub-signal per dimension, each name + value 0-100 + source_url: (a) rd_spend_growth — 0-100 from the annual filing's Research and Development line item, last 3y year-over-year trend (rising YoY → 75-90, flat → 55, declining → 30; for mega-cap tech R&D has typically grown materially YoY → expect 80+); (b) tech_acquisitions_count_36mo — 0-100 from Crunchbase acquisitions page + press releases over 36m (0 → 30, 1-3 → 60, 4-8 → 80, 9+ → 95); (c) patent_filing_intensity_24mo — 0-100 from USPTO assignee search filings over 24m (0 → 25, 1-100 → 55, 100-1000 → 80, 1000+ → 95). Each signals[].value MUST be a real assessment based on cited evidence (NOT a 0 placeholder). WORKED EXAMPLE PATTERN: a mega-cap whose 10-K shows R&D rising materially (e.g. $20B → $35B over 4-5y), Crunchbase listing multiple AI/AR acquisitions, USPTO showing 1000+ patents/year → composite ≈ 85-90. DO NOT emit absent. Server applies composite_signal_score.