Coverage and status
Every M49 country or area has a public page. Regional M49 nodes are used as taxonomy. Empty collections and unavailable values remain explicit; they are never converted to zero.
PIP · Political Intelligence Platform
How CiviOrg collects, resolves, publishes and labels global political data.
Every M49 country or area has a public page. Regional M49 nodes are used as taxonomy. Empty collections and unavailable values remain explicit; they are never converted to zero.
Only sources with free access and commercial reuse rights are stored and redistributed. Material without clear reuse rights may contribute a link or signal, but its content is not copied.
Every endpoint is registered with its licence, reuse policy, host, adapter, language, refresh frequency and expiry window. A registered source is processed only while its endpoint and production adapter are enabled.
Provides the M49 geography, containment taxonomy and localized territory names. Reusable snapshots are stored in full.
Unicode-3.0Provides CC0 identifiers and structured signals for jurisdictions, current leadership, parties and elections. Published assertions retain their provenance.
CC0-1.0Provides a weekly current-leadership source. Only eligible United States government material not separately marked as protected is reused.
US-GOV-PDProvides eligible electoral media signals. PIP stores the GDELT record, URL, date and derived clusters; protected article text is not copied.
GDELT-OPEN-DATAProvides portraits and logos discovered through Wikidata. Every file is checked individually and accepted only under Public Domain, CC0, CC BY or CC BY-SA terms.
PER-FILEParty Facts, Elections Global, Leaders Global, ParlGov and the World Bank Database of Political Institutions are recorded as licence-eligible historical sources. They are not presented as imported while their production adapters remain disabled.
IPU Parline and ElectionGuide are excluded because their non-commercial or reuse restrictions do not satisfy PIP’s commercial-reuse policy. Their content is not copied into the redistributable database.
Sources produce dated assertions. Domain-specific authorities determine the canonical view at a chosen date. Incompatible assertions from the same authority are displayed together as a conflict instead of being guessed away.
Only polls for the same contest, round, population and scenario are combined. Undecided respondents remain explicit and figures are not silently renormalized.
weight = 2^(-age_days / 30) × clamp(√(sample_size / 1000), 0.5, 1.5)
poll_average = Σ(value × weight) / Σ(weight)The first model combines candidacy status, independent media clusters, compatible polling, comparable party vote and incumbency. Every run preserves inputs, model version, uncertainty and dataset hash.
candidacy_score = clamp(base_status
+ min(20, 4 × ln(1 + independent_clusters_30d))
+ min(10, 2 × ln(1 + independent_clusters_31_180d)), 0, 99)
competitiveness = 0.50 × compatible_poll_share
+ 0.25 × last_comparable_party_vote
+ 0.20 × media_cluster_share_30d
+ 0.05 × incumbency_score
raw_win = candidacy_score / 100 × competitivenessMedia is accepted only under Public Domain, CC0, CC BY or CC BY-SA terms. A single content-addressed master is retained; responsive variants are generated on demand, with attribution preserved.