1 line
16 KiB
JSON
1 line
16 KiB
JSON
{"repo": "bigdataia-spring26-team-2/case-study-5", "files": {"config/board_governance.yml": {"1ead83e3102eb073bd4161e1586ef1eabb2acf18": {"t": 1785746334.988417, "n": 7746, "h": "e5e785ba56fd090e", "c": "# Board Governance Signal Configuration\n# PE Org-AI-R Platform\n\nscoring:\n base_score: 20\n tech_committee_points: 15\n tech_expertise_points: 20\n data_officer_points: 15\n independence_points: 10\n independence_threshold: 0.5\n max_score: 100\n confidence_threshold: 0.75\n\nextraction:\n chunk_size: 4000\n fuzzy_match_threshold: 85\n\nllm_pool:\n providers:\n - service: deepseek\n prefix: \"sk-\"\n url: \"https://api.deepseek.com/v1\"\n model: deepseek-chat\n temperature: 0\n - service: cerebras\n prefix: \"csk-\"\n url: \"https://api.cerebras.ai/v1\"\n model: gpt-oss-120b\n temperature: 0.05\n - service: groq\n prefix: \"gsk_\"\n url: \"https://api.groq.com/openai/v1\"\n model: llama-3.3-70b-versatile\n temperature: 0.05\n max_keys_per_provider: 20\n max_attempts_per_call: 15\n max_consecutive_failures: 5\n timeout: 90\n max_tokens: 4000\n\nprompts:\n extract: |\n Extract board governance data from this SEC DEF 14A proxy statement chunk.\n\n BOARD MEMBERS (directors only, not officers unless also on board):\n - name: Full legal name\n - is_independent: true/false/null\n - background: Complete career history -- ALL companies, ALL titles, education. 5-6 sentences minimum. Do NOT abbreviate.\n\n COMMITTEES:\n - name: Exact name as written\n - description: Purpose/responsibilities if mentioned\n\n OFFICERS (ALL C-level and VP positions):\n - title: Exact title\n - name: Person's name\n\n Extract everything. Duplicates handled downstream.\n\n Return ONLY valid JSON:\n {{\"board_members\": [{{\"name\": \"\", \"is_independent\": null, \"background\": \"\"}}], \"committees\": [{{\"name\": \"\", \"description\": \"\"}}], \"officers\": [{{\"title\": \"\", \"name\": \"\"}}]}}\n\n Text:\n {text}\n\n cleanup: |\n Remove duplicates from board governance data extracted across multiple chunks of a DEF 14A.\n\n BOARD MEMBERS:\n {members_json}\n\n COMMITTEES:\n {committees_json}\n\n OFFICERS:\n {officers_json}\n\n Remove:\n - Abbreviated names when full name exists (e.g. \"Dr. Shah\" if \"Aarti Shah\" exists)\n - Staff listed as directors (Secretary, Treasurer, Corporate Secretary)\n - Committee name variants (keep full name only)\n - Retired or former directors who are no longer serving on the board\n\n Officer title deduplication rules:\n - All of these refer to the SAME role, keep only ONE (the most complete version):\n \"CEO\", \"Chief Executive Officer\", \"President and CEO\", \"President & CEO\", \"President & Chief Executive Officer\"\n - Same for CFO variants: \"CFO\", \"Chief Financial Officer\", \"EVP and CFO\", \"EVP & CFO\"\n - Same for GC variants: \"General Counsel\", \"EVP, General Counsel and Secretary\", \"EVP, General Counsel & Secretary\"\n - When in doubt, keep the most specific/complete title and drop all shorter variants\n - If two titles describe the same person in different formats, keep one\n\n CRITICAL: Return titles and names EXACTLY as they appear in the input lists above.\n\n Return ONLY valid JSON:\n {{\"final_board_members\": [\"Full Name 1\"], \"final_committees\": [\"Committee 1\"], \"final_officers\": [\"Exact Title 1\"]}}\n\n classify_all: |\n You are classifying board governance entities for a PE AI-readiness assessment.\n Classify all three entity types below. Use context from ALL sections to inform your decisions.\n\n === BOARD MEMBERS (classify technology expertise) ===\n {members_json}\n\n HAS TECH EXPERTISE (true):\n - Executive at a technology/software/semiconductor/AI company\n - CTO, CIO, CDO, CAIO at any company\n - Career in software engineering, CS, tech consulting\n - Board member at tech companies (check background carefully)\n - Technology-focused VC/PE investor\n - Director/leader of major technology research institutions or labs\n\n NOT TECH EXPERTISE (false):\n - General business, finance, legal, operations at non-tech companies\n - Retail, manufacturing, healthcare without tech focus\n\n Empty or ambiguous background -> confidence below 0.5.\n\n === COMMITTEES (classify technology focus) ===\n {committees_json}\n\n TECHNOLOGY-FOCUSED (true): Primary focus on technology, digital transformation, AI, cybersecurity, data, or innovation.\n NOT TECHNOLOGY (false): Audit, Compensation, Nominating, Governance, ESG, general Risk.\n\n === OFFICERS (classify technology/AI/data role) ===\n {officers_json}\n\n TECHNOLOGY ROLES (true): CTO, CIO, CDO, CAIO, Chief Digital Officer, Chief Analytics Officer, VP/SVP/EVP of Technology/Engineering/Data/AI/IT.\n NOT TECHNOLOGY (false): CEO, CFO, COO, General Counsel, CMO, CHRO, EVP Operations, EVP Sales.\n\n CROSS-REFERENCING (CRITICAL):\n When classifying officers, CHECK if they appear as board members above.\n Use their FULL BACKGROUND to determine if their role involves technology, even if the title alone does not say so.\n\n Examples:\n - \"EVP, Operations\" whose board member background shows \"20 years leading hardware engineering teams\" -> true\n - \"President and CEO\" whose background shows \"co-founded a semiconductor company, microprocessor designer\" -> true (dual-role technology leader)\n - \"EVP, Operations\" whose background shows \"supply chain and logistics\" -> false\n\n Classify based on the PERSON and their actual responsibilities, not just the title string.\n\n Return ONLY valid JSON:\n {{\"member_classifications\": [{{\"member_name\": \"\", \"has_tech_expertise\": false, \"confidence\": 0.0, \"reasoning\": \"\"}}], \"committee_classifications\": [{{\"committee_name\": \"\", \"is_technology\": false, \"confidence\": 0.0, \"reasoning\": \"\"}}], \"officer_classifications\": [{{\"officer_title\": \"\", \"is_tech_officer\": false, \"confidence\": 0.0, \"reasoning\": \"\"}}]}}\n\n validate: |\n Fact-check board governance data extracted from {ticker}'s ({company_name}) DEF 14A proxy statement.\n Use your general knowledge of this company to validate.\n\n CRITICAL RULE: The DEF 14A filing is the SOURCE OF TRUTH for board composition.\n If a person is listed as a director in the filing, they ARE a director for that filing period.\n Do NOT remove board members just because your general knowledge differs from the filing.\n Only flag someone as a non-director if they are clearly staff/officers incorrectly extracted as directors (e.g. a Corporate Secretary or Treasurer who is not actually on the board).\n\n EXTRACTED BOARD MEMBERS:\n {members_json}\n\n TECH EXPERTISE CLASSIFICATIONS:\n {tech_classifications_json}\n\n EXTRACTED COMMITTEES:\n {committees_json}\n\n EXTRACTED OFFICERS (with classifications):\n {officers_json}\n\n Validate and return corrections:\n\n 1. member_corrections: For each member where tech classification seems WRONG based on their background, provide correction. Only correct clear errors, not borderline cases.\n\n 2. missing_officers: Known C-level technology officers (CTO, CIO, CDO) at this company missing from the extracted list. Include dual-role holders. Provide title, name, is_tech_officer, and note.\n\n 3. flagged_non_directors: ONLY names that are clearly NOT directors -- staff or officers who were incorrectly extracted as board members (e.g. a CFO or General Counsel who is not on the board). Do NOT flag actual directors just because they may have retired after the filing date.\n\n 4. notes: Brief quality assessment.\n\n Return ONLY valid JSON:\n {{\"member_corrections\": [{{\"name\": \"\", \"issue\": \"\", \"corrected_has_tech_expertise\": null, \"reasoning\": \"\"}}], \"missing_officers\": [{{\"title\": \"\", \"name\": \"\", \"is_tech_officer\": true, \"note\": \"\"}}], \"flagged_non_directors\": [\"\"], \"notes\": \"\"}}\n\noutput:\n store_in_snowflake: true\n signal_table: external_signals\n summary_table: company_evidence_summary\n category: governance\n source: sec_def14a"}, "94edfa537225f94d59147c1b8fcc1bdf7235fe51": {"t": 1785762061.4264436, "n": 7746, "h": "e5e785ba56fd090e", "c": "# Board Governance Signal Configuration\n# PE Org-AI-R Platform\n\nscoring:\n base_score: 20\n tech_committee_points: 15\n tech_expertise_points: 20\n data_officer_points: 15\n independence_points: 10\n independence_threshold: 0.5\n max_score: 100\n confidence_threshold: 0.75\n\nextraction:\n chunk_size: 4000\n fuzzy_match_threshold: 85\n\nllm_pool:\n providers:\n - service: deepseek\n prefix: \"sk-\"\n url: \"https://api.deepseek.com/v1\"\n model: deepseek-chat\n temperature: 0\n - service: cerebras\n prefix: \"csk-\"\n url: \"https://api.cerebras.ai/v1\"\n model: gpt-oss-120b\n temperature: 0.05\n - service: groq\n prefix: \"gsk_\"\n url: \"https://api.groq.com/openai/v1\"\n model: llama-3.3-70b-versatile\n temperature: 0.05\n max_keys_per_provider: 20\n max_attempts_per_call: 15\n max_consecutive_failures: 5\n timeout: 90\n max_tokens: 4000\n\nprompts:\n extract: |\n Extract board governance data from this SEC DEF 14A proxy statement chunk.\n\n BOARD MEMBERS (directors only, not officers unless also on board):\n - name: Full legal name\n - is_independent: true/false/null\n - background: Complete career history -- ALL companies, ALL titles, education. 5-6 sentences minimum. Do NOT abbreviate.\n\n COMMITTEES:\n - name: Exact name as written\n - description: Purpose/responsibilities if mentioned\n\n OFFICERS (ALL C-level and VP positions):\n - title: Exact title\n - name: Person's name\n\n Extract everything. Duplicates handled downstream.\n\n Return ONLY valid JSON:\n {{\"board_members\": [{{\"name\": \"\", \"is_independent\": null, \"background\": \"\"}}], \"committees\": [{{\"name\": \"\", \"description\": \"\"}}], \"officers\": [{{\"title\": \"\", \"name\": \"\"}}]}}\n\n Text:\n {text}\n\n cleanup: |\n Remove duplicates from board governance data extracted across multiple chunks of a DEF 14A.\n\n BOARD MEMBERS:\n {members_json}\n\n COMMITTEES:\n {committees_json}\n\n OFFICERS:\n {officers_json}\n\n Remove:\n - Abbreviated names when full name exists (e.g. \"Dr. Shah\" if \"Aarti Shah\" exists)\n - Staff listed as directors (Secretary, Treasurer, Corporate Secretary)\n - Committee name variants (keep full name only)\n - Retired or former directors who are no longer serving on the board\n\n Officer title deduplication rules:\n - All of these refer to the SAME role, keep only ONE (the most complete version):\n \"CEO\", \"Chief Executive Officer\", \"President and CEO\", \"President & CEO\", \"President & Chief Executive Officer\"\n - Same for CFO variants: \"CFO\", \"Chief Financial Officer\", \"EVP and CFO\", \"EVP & CFO\"\n - Same for GC variants: \"General Counsel\", \"EVP, General Counsel and Secretary\", \"EVP, General Counsel & Secretary\"\n - When in doubt, keep the most specific/complete title and drop all shorter variants\n - If two titles describe the same person in different formats, keep one\n\n CRITICAL: Return titles and names EXACTLY as they appear in the input lists above.\n\n Return ONLY valid JSON:\n {{\"final_board_members\": [\"Full Name 1\"], \"final_committees\": [\"Committee 1\"], \"final_officers\": [\"Exact Title 1\"]}}\n\n classify_all: |\n You are classifying board governance entities for a PE AI-readiness assessment.\n Classify all three entity types below. Use context from ALL sections to inform your decisions.\n\n === BOARD MEMBERS (classify technology expertise) ===\n {members_json}\n\n HAS TECH EXPERTISE (true):\n - Executive at a technology/software/semiconductor/AI company\n - CTO, CIO, CDO, CAIO at any company\n - Career in software engineering, CS, tech consulting\n - Board member at tech companies (check background carefully)\n - Technology-focused VC/PE investor\n - Director/leader of major technology research institutions or labs\n\n NOT TECH EXPERTISE (false):\n - General business, finance, legal, operations at non-tech companies\n - Retail, manufacturing, healthcare without tech focus\n\n Empty or ambiguous background -> confidence below 0.5.\n\n === COMMITTEES (classify technology focus) ===\n {committees_json}\n\n TECHNOLOGY-FOCUSED (true): Primary focus on technology, digital transformation, AI, cybersecurity, data, or innovation.\n NOT TECHNOLOGY (false): Audit, Compensation, Nominating, Governance, ESG, general Risk.\n\n === OFFICERS (classify technology/AI/data role) ===\n {officers_json}\n\n TECHNOLOGY ROLES (true): CTO, CIO, CDO, CAIO, Chief Digital Officer, Chief Analytics Officer, VP/SVP/EVP of Technology/Engineering/Data/AI/IT.\n NOT TECHNOLOGY (false): CEO, CFO, COO, General Counsel, CMO, CHRO, EVP Operations, EVP Sales.\n\n CROSS-REFERENCING (CRITICAL):\n When classifying officers, CHECK if they appear as board members above.\n Use their FULL BACKGROUND to determine if their role involves technology, even if the title alone does not say so.\n\n Examples:\n - \"EVP, Operations\" whose board member background shows \"20 years leading hardware engineering teams\" -> true\n - \"President and CEO\" whose background shows \"co-founded a semiconductor company, microprocessor designer\" -> true (dual-role technology leader)\n - \"EVP, Operations\" whose background shows \"supply chain and logistics\" -> false\n\n Classify based on the PERSON and their actual responsibilities, not just the title string.\n\n Return ONLY valid JSON:\n {{\"member_classifications\": [{{\"member_name\": \"\", \"has_tech_expertise\": false, \"confidence\": 0.0, \"reasoning\": \"\"}}], \"committee_classifications\": [{{\"committee_name\": \"\", \"is_technology\": false, \"confidence\": 0.0, \"reasoning\": \"\"}}], \"officer_classifications\": [{{\"officer_title\": \"\", \"is_tech_officer\": false, \"confidence\": 0.0, \"reasoning\": \"\"}}]}}\n\n validate: |\n Fact-check board governance data extracted from {ticker}'s ({company_name}) DEF 14A proxy statement.\n Use your general knowledge of this company to validate.\n\n CRITICAL RULE: The DEF 14A filing is the SOURCE OF TRUTH for board composition.\n If a person is listed as a director in the filing, they ARE a director for that filing period.\n Do NOT remove board members just because your general knowledge differs from the filing.\n Only flag someone as a non-director if they are clearly staff/officers incorrectly extracted as directors (e.g. a Corporate Secretary or Treasurer who is not actually on the board).\n\n EXTRACTED BOARD MEMBERS:\n {members_json}\n\n TECH EXPERTISE CLASSIFICATIONS:\n {tech_classifications_json}\n\n EXTRACTED COMMITTEES:\n {committees_json}\n\n EXTRACTED OFFICERS (with classifications):\n {officers_json}\n\n Validate and return corrections:\n\n 1. member_corrections: For each member where tech classification seems WRONG based on their background, provide correction. Only correct clear errors, not borderline cases.\n\n 2. missing_officers: Known C-level technology officers (CTO, CIO, CDO) at this company missing from the extracted list. Include dual-role holders. Provide title, name, is_tech_officer, and note.\n\n 3. flagged_non_directors: ONLY names that are clearly NOT directors -- staff or officers who were incorrectly extracted as board members (e.g. a CFO or General Counsel who is not on the board). Do NOT flag actual directors just because they may have retired after the filing date.\n\n 4. notes: Brief quality assessment.\n\n Return ONLY valid JSON:\n {{\"member_corrections\": [{{\"name\": \"\", \"issue\": \"\", \"corrected_has_tech_expertise\": null, \"reasoning\": \"\"}}], \"missing_officers\": [{{\"title\": \"\", \"name\": \"\", \"is_tech_officer\": true, \"note\": \"\"}}], \"flagged_non_directors\": [\"\"], \"notes\": \"\"}}\n\noutput:\n store_in_snowflake: true\n signal_table: external_signals\n summary_table: company_evidence_summary\n category: governance\n source: sec_def14a"}}}} |