Ontology field guide
By NOSIBLE Research
GICS Industry Classification
GICS shows which parts of the economy an event concerns, preserving industry structure that a broad technology label removes.
Industry exposure is not one variable. Semiconductor capacity, cloud-platform investment and data-centre buildout may all involve artificial intelligence, but they transmit through different companies, margins and risks. GICS separates them into sectors, groups, industries and sub-industries. World assigns that hierarchy to reporting so researchers can build cohorts at the required economic level.[1]
GICS supports peer selection, exposure measurement and spillover tests. It does not replace issuer fundamentals or a point-in-time security master. This guide provides the structure, World statistics, an AI-trade study and a falsifiable test of whether industry attention adds information beyond event volume.[2]
- Categories
- 237 categories
- Structure
- 4 levels
- World events labelled
- 23.0%
- Stable codes
- Since v1
- Release data
- CC0
Foundations
GICS classification separates companies through four sector and industry levels globally
GICS assigns companies according to their principal business activity. The hierarchy runs from sector to industry group, industry and sub-industry. Revenue is the primary classification input, with earnings and market perception used where business lines are mixed. The nested structure lets a researcher move from a broad economic exposure to a tighter operating peer group without changing classification systems.[1]
World classifies the economic activity expressed in an event record. It does not assert that every named company belongs to the selected sub-industry, and it does not resolve conglomerate revenue across segments. Event-level GICS labels should be joined to a point-in-time issuer classification when the test concerns security membership. Results should also survive alternative peer definitions because industry schemes differ in how well they explain returns, valuation multiples and growth. Event and issuer classifications answer different questions.[2]
Categories
GICS organises 237 categories into sectors, groups, industries and sub-industries
GICS has four nested levels: 12 sector roots, followed by industry groups, industries and 164 leaf sub-industries. Each World assignment records one path at the available depth. The explorer exposes definitions, stable codes, full paths and World V1.2 counts. Search and bounded expansion keep the same interface usable across the complete hierarchy.[1]
3,516,971 of 15,311,040 World events carry labels from GICS Industry Classification. Select a category to see its count and both relevant shares.
Energy
Energy
Definition
Energy production, drilling and field services.
Potential research use
Oil, gas, drilling and field-service activity
- Code
- Energy
- Events with label
- 137,455
- Share of labelled
- 3.9%
- Share of World
- 0.9%
Events with label is the selected count. Label share divides it by 3,516,971 assigned events; World share divides it by all 15,311,040 events.
Representative World V1.2 events
One strong classified example per available year, with up to ten years shown.
Duke Energy Closes Aging Lake Julian Coal Plant in Asheville
Coverage 68Noble Energy Sells 3% Tamar Field Stake for $369 Million
Coverage 18Hurricane Energy Finds Largest Undeveloped UK Oil Field Off Shetland
Coverage 29Hurricane Energy Achieves First Oil at UK's Lancaster Fractured Field
Coverage 23Aker Energy Revives Ghana Pecan Offshore Oil Field Development Plans
Coverage 21US Releases 50 Million Barrels Oil to Lower Energy Costs
Coverage 1,461Hezbollah Threatens Israel Over Sea Boundary Energy Drilling
Coverage 250Woodside Energy Achieves First Oil at Senegal's Sangomar Offshore Field
Coverage 183Valeura Energy Approves Wassana Field Redevelopment in Thailand
Coverage 64Birchcliff Energy Reports Record 2025 Production and Strong Financial Results
Coverage 132
Browse categories to compare definitions, World V1.2 statistics and representative classified events.
Example usage 1 of 3
Example Usage: AI coverage rises sevenfold and concentrates in fewer industries over time
World contains 13,730 entity-resolved English events that discuss artificial intelligence, carry a GICS sub-industry and name one of 15 compute or platform companies. Normalized coverage rises more than sevenfold after November 2022. The top-three sub-industry share increases from 48.7% to 59.9%, while effective sub-industries fall from 9.9 to 7.0.[3]
Semiconductors rise from 19.8% to 28.7% of the cohort. A compute-infrastructure basket reaches 42.5%, up from 33.3%. Higher attention and narrower industry breadth mean an AI basket can conceal shared economic exposure. ChatGPT's launch sets the boundary; it does not establish causality or portfolio crowding.[4][5]
The intensity line divides the remembered cohort by remembered English World events. Phase comparisons pool all events before December 2022 and all events afterwards. Rising normalized AI attention alongside fewer effective sub-industries indicates concentration in reporting exposure, not portfolio crowding or causality.
Example usage 2 of 3
Example Usage: Automakers and chipmakers capture 84% of classified semiconductor-shortage coverage across exposed industries
The 2021 semiconductor shortage is not only a chipmaker story. Automobile Manufacturers and Semiconductors each contribute 279 assignments, or 41.9% of the 666-event cohort. Together they account for 83.8% of classified coverage. Monthly normalized attention peaks at 2,198 events per million in October as production constraints persist.[6]
GICS separates suppliers from customers inside one supply-chain shock. Industry composition shows whether reporting concentrates on scarce components, curtailed vehicle production or downstream exposure. The result defines distinct equity cohorts for estimating inventory, margin and delivery effects instead of treating every semiconductor-shortage article as the same sector signal across portfolios in research.[7]
- Automobile Manufacturers
- 41.9%
- Semiconductors
- 41.9%
- Other sub-industries
- 16.2%
The fixed English cohort requires semiconductor and supply-constraint terms. All 666 GICS-assigned events occur in 2021. Monthly counts are divided by English World volume for every month in the fixed year, enabling direct comparison throughout 2021.
Example usage 3 of 3
Example Usage: Commercial-property attention reaches regional banks after office landlords peak during the credit cycle
Office Real Estate Investment Trust attention reaches 238 events per million on 23 January 2024. New York Community Bancorp reports a quarterly loss on 31 January. Mortgage Finance then reaches 158 per million on 22 February, while Regional Banks peaks later at 205 per million on 28 March across the same credit cycle.[8]
The sequence exposes transmission across balance sheets. Office landlords carry vacancy and valuation risk. Mortgage lenders and regional banks carry collateral, funding and credit risk. GICS supplies separate equity cohorts and event clocks for testing whether lender attention precedes revisions, deposits, spreads or relative returns.[9]
- Office REIT peak
- +238 per million
- Regional-bank peak
- +205 per million
- Mortgage-finance peak
- +158 per million
The fixed English cohort covers explicit commercial-property, office-vacancy and commercial-mortgage reporting from October 2022 through June 2024. Each line pools GICS sub-industry counts across the current and prior 55 days, then divides by all English World events in the same window.
Data and sources
Download GICS and References
Downloads
Download categories and counts as CSV, or the complete machine-readable release as JSON.
Release details and citation files
ManifestRelease READMECC0 license and scopeCitation fileChangelog
Need the complete World event schema? Open the World data dictionary.
References
- [1]MSCI and S&P Dow Jones Indices. (2023). Global Industry Classification Standard (GICS) methodology.
- [2]Bhojraj, S., Lee, C. M. C., & Oler, D. K. (2003). What's my line? A comparison of industry classification schemes for capital market research. Journal of Accounting Research, 41(5), 745–774.
- [3]OpenAI. (2022, November 30). Introducing ChatGPT.
- [4]NVIDIA Corporation. (2024). Form 10-K for the fiscal year ended January 28, 2024.
- [5]U.S. Bureau of Industry and Security. (2023, October 17). Commerce strengthens restrictions on advanced computing semiconductors and semiconductor manufacturing equipment.
- [6]U.S. Department of Commerce. (2022). Semiconductor data confirms urgent need to address the chip shortage.
- [7]The White House. (2021). Building resilient supply chains, revitalizing American manufacturing, and fostering broad-based growth: 100-day reviews.
- [8]New York Community Bancorp. (2024). Full-year and fourth-quarter 2023 results.
- [9]Board of Governors of the Federal Reserve System. (2023). Dodd-Frank Act stress-test results.
Continue exploring