Global Relevance Engine
GREMap’s relevance tools organize source-linked reporting and connect it with geographic, prediction and market context. Connections help you explore; they do not independently verify a report or predict an outcome.
News collection
We attempt 88 direct source channels (publisher RSS feeds and public publisher indexes), nine regional Google News searches, nine additional topic searches, including dedicated Taiwan–China and religious-persecution queries for the general news feed, five publisher/official-source searches, and GDELT. The direct feeds include BBC World and five BBC regional channels, Guardian World, Al Jazeera, DW, France 24, UN News, Africanews, CNA, Focus Taiwan, Korea Times and AllAfrica’s Conflict, Peace and Security reporting. The expanded mix also attempts Sky, CNN, NBC, Fox, NewsNation, CBS, ABC, NPR, the Times of Israel, the Jerusalem Post, CGTN, Xinhua, Tehran Times and IRNA. Latin American coverage also attempts El Universal and La Jornada in Mexico, El Tiempo regional and justice feeds in Colombia, France 24’s Latin America desk, Agência Brasil, Guardian Americas, DW Español and UN News Español. Regional additions include Dabanga and Radio Tamazuj for Sudan/South Sudan, Premium Times and RFI for Africa, the Kyiv Independent, Ukrainska Pravda and the Moscow Times, Myanmar Now, The Irrawaddy, DVB, Dawn, Yonhap, Taipei Times, Rappler and Inquirer. Dedicated UN humanitarian/refugee/peace channels, Defense News, Breaking Defense, Naval News, gCaptain, the Maritime Executive and SecurityWeek expand topic depth. El País América and Infobae broaden Spanish-language coverage. Original Spanish and Portuguese titles are retained. Google News adds Mexico, Spanish-speaking South America and Portuguese-speaking Brazil searches. WION has a publisher link; a working feed has not been confirmed. Source availability and recent-match counts are shown with each feed. Reachable feeds can contain only old or unrelated stories; those do not count as recent matches. Some aggregators are inaccessible from the hosted server.
English, Spanish and Portuguese intake rules select armed-conflict, organized armed violence, military/security, humanitarian, diplomacy and cyber/infrastructure reporting. Relevant reporting can include aid, displacement, sanctions, naval activity and cyber developments. A topic match is distinct from the stricter map rule for the Armed conflict label, which requires an explicit report of an armed attack or fighting. Inclusion is not a legal determination that an incident is an armed conflict. Duplicate reports are reduced and future or undated items are excluded. Displayed reporting is a limited sample within the selected 24-hour, 72-hour, 7-day, 30-day or 60-day window, with representation across publishers, topics and reporting dates. Older stories can fall out of publisher feeds before the window ends. RSS and Atom formats are accepted. Atom feeds exposing only an update time are explicitly labeled “Updated”; their original publication time is not inferred. Undated records are excluded. Completed public feed results can be reused for five minutes across sections/windows, with original report dates and source-check times preserved. Server-rendered collections use a short cache; a request timestamp is not a claim that every story was manually reviewed.
Strategic escalation and evidence
Dedicated searches cover nuclear threats, first-use policy, strategic posture, leaders’ military warnings and strategic facilities. Strategic escalation is a visible label for nuclear posture and threatened or planned strategic military action, with overlapping Military and Diplomacy filters. Explicit reports of an attack occurring retain the Armed conflict label. Threats and hypothetical strikes are not labeled as attacks that have happened. Selection gives attention to strategic escalation while retaining a range of publishers and topics. This is editorial coverage prioritization, not a severity ranking or assessment that a claim is true.
Direct feeds include The War Zone, Anadolu World, White House releases and IAEA news. Official publications record those institutions’ accounts rather than independent confirmation. Reuters, AP, State Department and Pentagon coverage is also sought through separately labeled Google News searches; these are not licensed direct wire feeds. A cross-country official-source search supplements, rather than exhaustively monitors, leaders’ statements. State Department and Pentagon direct pages were inaccessible during setup; search discovery can be delayed or incomplete. No direct Truth Social or X monitoring is configured.
Evidence labels describe headline wording or a configured official source: attributed claim/report, response/denial, analysis/question, official-source statement or publisher reporting. They are automatic attribution aids, not truth scores. The site never automatically labels a story independently corroborated. Multiple outlets can repeat one source. Responses and denials do not settle the original allegation. Open original reporting for full context.
Search filters collected headlines and publisher names in the selected view; it does not initiate a new web search. Source health records the last observed success, failure reason and consecutive failed checks. Internal diagnostics support collection review but are limited; they are not a complete archive of all reporting considered.
Related-report grouping
The Global Relevance Engine groups related reporting to help you compare coverage. A shared place or topic alone does not establish that reports describe the same development. Each group keeps its original publisher links, and you can switch to individual reports. Compare coverage shows dated reports from the current and retained sample, oldest to newest. Report dates may be publication, update or indexing times; they are not confirmed incident times. Grouping can miss matches or combine distinct developments. Source counts use named reporting channels and may include channels from the same publisher. They do not establish independent corroboration. Event pages are rebuilt from available reporting and up to 60 days of retained samples, rather than a permanent event archive.
Relevance and persecution coverage
Reporting is selected for conflict, organized armed violence, strategic security, humanitarian, infrastructure or identity-targeting context, rather than general local crime coverage. Religious persecution includes reported violence or suppression tied to religion or belief, including nonbelief. Mere religious identity or routine religious news does not qualify. Rules can miss context or misclassify reporting; open the original source.
The Religious persecution filter overlaps with the primary armed-violence, humanitarian or other topic. Each map marker still uses one primary topic color. Direct collection includes UN News Human rights and Focus Taiwan politics and cross-strait indexes, alongside the existing regional publishers. Dedicated search attempts do not guarantee availability. Coverage notices identify failed channels; lack of reports is not evidence of calm.
Cyber and infrastructure
The Cyber section attempts specialist publisher feeds from BleepingComputer, The Hacker News and The Record, official CISA and UK NCSC feeds, and topic-matched reports from the wider source pool. Endpoint availability varies. Original article and advisory dates are preserved, with the selected reporting window applied. An official advisory is not evidence that an attack happened. Research and attribution labels describe the publisher wording, not independent confirmation.
Topic-matched cyber reports can also enter the general news/map sample, including spyware, phishing, exploit and vulnerability research. Inclusion does not mean a reported vulnerability has been exploited. AI-agent and botnet reporting requires attack or misuse context; routine AI product stories are excluded. Country mentions may identify victims, alleged attackers or other context. Pins never establish an attack origin. Automation is not presumed autonomous, hostile or deployed merely because an article discusses it.
Geographic matching
Geographic references are identified from publisher headlines and feed excerpts to help readers explore reporting by place. The country lookup is derived from mledoze/countries, under the ODbL license. State and province pins use provided approximate label points from Natural Earth Admin-1 (public domain). Reporting may not support a specific event city, particularly for national policy or equipment-supply stories. City names and approximate centers use the free GeoNames cities15000 gazetteer. Publisher datelines and government statements alone do not establish city locations. When a specific locality cannot be established, a national capital or broader geographic reference may be shown. National capital labels use GeoNames countryInfo. This does not imply an event occurred in the capital. One source-listed capital is used where multiple seats exist; capital labels do not imply a position on disputed recognition. Non-country areas, uninhabited territories without a listed capital, and an unresolved territory retain broader reference points. Solid marker outlines indicate city centers; double outlines indicate states/provinces; dashed outlines indicate national capital references or broader regional fallbacks. A cyan ring marks an institutional reference. City data: GeoNames, under CC BY 4.0. Map markers use approximate geographic centers. They are not incident coordinates, front lines, evacuation routes or safe-zone boundaries.
Names can be ambiguous. A country can be mentioned in a diplomatic story about an event elsewhere. Reports without recognized names remain in the global list. Absence of a marker or headline does not establish absence of violence. Names and boundaries do not imply a position on territorial recognition.
Reports about UN headquarters or deliberations of the UN Security Council and General Assembly can use an approximate New York headquarters reference point, labeled as institutional rather than incident coordinates. The UN headquarters address is published by UN Library Ask DAG. An institutional reference does not place UN field operations at headquarters. Publisher identity alone is not a location.
Figures and economic observations
Headline counts measure this source sample, not the number of active wars. Economic observations use FRED series sourced from EIA, CBOE and the Federal Reserve, and ECB currency data through Frankfurter. Dates, units and original source links are retained. If fresh FRED retrieval fails, a dated saved observation is labeled as saved. These are not intraday quotes; price movements do not establish losses caused by conflict.
Economic exposure and market visuals
The economics workspace connects recent reporting with relevant industry and economic context. General exposure guides describe possible channels and offsetting factors; they do not establish the actual financial effect of a headline. Cards use a sampled selection from the trailing seven days, with reduced duplication and representation across publishers. Economic coverage also includes tariffs, sanctions, export controls and relevant price or supply-chain reporting.
TradingView provides third-party stock and ETF heatmaps and security charts with attribution. Stock data may be delayed; exchanges and symbols can be unavailable. These embedded values are not imported into GREMap or used to calculate event returns. The initial stock view is the S&P 500; ETF coverage is U.S.-listed. The illustrative security list includes international exposure through selected U.S. listings, not comprehensive global exchange coverage. Issuer links provide holdings and disclosures; current fund weights are not ingested. Watchlist selections stay in this browser. EIA, PortWatch, WTO, World Bank and SEC links open original resources; listing them does not imply their full datasets are connected. No automated social sentiment or truth score is shown.
Human impact and humanitarian information
The human-impact block no longer publishes the former hand-selected incident sum. The combined tally is paused because its coverage could not represent all conflicts. The API returns a paused status and a null total, never a zero or an old numeric fallback. Recent casualty articles remain separate source accounts and are not summed.
A consistent monthly dataset is being evaluated. ACLED publishes most event data weekly and supports authenticated API access, but no ACLED data feed is connected here. Access and display must fit its applicable terms. Event-level reported fatalities are estimates subject to revision, not a complete measure of direct and indirect war deaths. A future monthly view would need explicit definitions, geographic coverage, full ingestion checks and dated revisions.
Our World in Data provides a linked UCDP conflict-death chart. Definitions, reporting periods, revisions and direct versus indirect deaths matter. This site does not combine incompatible figures into a live death counter. Official destination advice is guidance, not a guarantee of safety.
Watch: a vertical news edition
Watch presents selected catalog clips and retained headline reports published within the trailing seven days, with regional and publisher variety. Each visit shuffles the order; manual refresh reshuffles, while background retrieval retains the session order. This is a sample, not an objective global importance ranking or a verified count of distinct incidents. Duplicates are reduced; different accounts of the same event may remain.
Headline cards are clearly labeled and do not imply video footage. Curated YouTube clips require an original video ID, publisher, upload date, recent verification, format evidence and embedding eligibility. The current catalog contains 70 Shorts from 18 publisher channels. Public YouTube metadata was last reviewed on 2026-11-14 (UTC). The catalog includes multiple regions and Spanish-language clips, labeled in Watch. Eligibility changes as uploads pass the seven-day limit. Channel IDs, upload timestamps, Shorts eligibility and embedding permission were checked; that does not independently verify the reporting. Old and future-dated entries are excluded; the browser rechecks age while open. Source dates are publication dates, not necessarily event or recording dates.
Only the active video player loads. YouTube branding and controls remain visible; engagement actions sit outside its viewport. Playback is paused when the page is hidden or the player is not sufficiently visible. Publisher restrictions, age or location rules and network failures can prevent embedded playback; the original source link stays available. Weekly sourcing research is separate from publication and is not an automatic video-ingestion service.
Prediction status and personal briefings
Saved prediction countdowns use the provider’s scheduled closing time, which can change and can differ from resolution time. Movement compares the latest observed YES price with the price saved when you first recorded the question, interpreted against the YES or NO lean of your latest estimate. A 50% estimate has no directional lean. Observations older than 24 hours are flagged for refresh. Price movement is not correctness or a return.
Correct or Incorrect labels require an explicit final binary provider result and no custom YES criteria. Proposed, disputed, missing and 50/50 outcomes do not establish correctness. Manual assessments remain separate. Plus briefings can include a timestamped prediction snapshot. Briefings prioritize relevant developments across followed interests, with attention to strategic escalation and reported armed violence. This is editorial triage, not a verified severity rating. Publisher excerpts are attributed, and suggested monitoring steps are general context, not individualized emergency or financial instructions.
Stocks and Robinhood links
The Stocks action uses a limited, checked catalog of companies and instruments. Connection labels distinguish named companies, relevant industry comparisons and broader sector context. Each card explains its relationship and company or fund context. A connection does not establish involvement, financial impact or expected returns.
Broader sector comparisons are shown alongside closer connections and labeled separately. They provide research context, not evidence of direct involvement. Companies outside the checked catalog can remain unmatched.
Company coverage and brokerage links are separate: a checked issuer may appear with issuer information even when no brokerage link is verified. Robinhood links use an explicit symbol allowlist and open public U.S. instrument pages. GREMap does not display live stock prices, connect brokerage accounts or execute trades. Access and instrument availability depend on the broker and jurisdiction. No referral compensation or affiliation is currently configured. Matches do not establish actual exposure, causation or expected returns.
Giving and contextual actions
The relief directory describes 14 organization profiles and 19 focused giving routes, with public accountability evidence. Country and cause matches come first; regional alternatives and broader funds follow. Matching uses explicit country and need terms, not precise location or a claim of best effectiveness. Water, medical, food, shelter, children and protection icons open those profiles. Contextual links do not earmark gifts to the incident beside them. Payments take place on official organization sites; GREMap does not collect donations. Profiles carry a review date and link to financial or oversight sources, rather than an automated trust score.
Story market actions use the Global Relevance Engine to connect reporting with sampled market questions. Connection labels distinguish close matches from related and broader context. They do not establish that a headline meets a contract’s resolution conditions or caused its price. Share actions open the device share menu or copy the original story URL; nothing is posted automatically.
Claims and corroboration
Publisher attribution indicates where a report came from. It does not mean GREMap independently verified it. False-flag allegations remain unassessed; prediction-market activity and social engagement are not factual verification. We do not assign automated truth scores.
Shopping and usage measurement
Optional aggregate counters record section views, action clicks, checklist activity and outbound categories without visitor identifiers or story text. Counts measure interactions, not unique visitors or completed transactions. Preparedness shopping examples link to merchants; paid links require configured approval and are disclosed on the cards. See privacy controls and commercial disclosures. Humanitarian recommendations remain ordered by place and need, independent of commercial links.
Automation and images
Automated filters, collection logic and AI-assisted software and guide drafting were used to create this demo. No fictional reporter profiles or expert credentials are presented. GREMap’s themed page banners and category artwork are AI-generated editorial illustrations, not photographs of reported events or live geographic or economic data. They are decorative; maps, figures, source reporting and product photos remain separate. Watch headline briefs use varied abstract patterns rather than generated depictions of their events. No aid organization endorsed the artwork. Preparation product photos come from the named manufacturer or merchant and are credited on their cards. Map tiles are attributed to their geographic providers.
Corrections and privacy
Read the corrections and feedback process for reporting issues. Preparation-checklist selections remain in your browser. External services have their own privacy practices. We do not require precise device location for map browsing.
Methodology reviewed 2026-10-19.


