Reconstructed GDELT news records match original data with varying accuracy
Agreement Between Reconstructed and Native GDELT GKG Records: A Field-Level Baseline
Computation and LanguageInformation Retrieval
Summary
The GDELT database summarizes news articles with detailed information like entities and tone but its record creation software is not available. The authors created an open-source tool to recreate these records and tested how closely their results match the original GDELT data. They found that accuracy varies a lot by data type—some fields like locations and tone match well, while others like quotes and date details do not. This means reconstructed data should be checked carefully before use, as it's not fully interchangeable with the original.
What this means in practice
- •For media analysis teams: Use the open-source reconstruction to generate compatible news data records when original GDELT software is unavailable, understanding strengths and weaknesses by data field.
- •For multilingual news monitoring teams: Assess tone and entities in translated non-English news with caution, as tone accuracy is lower than for English articles in the reconstructed data.
Authors
Nikita A. Klimin
Abstract
The Global Knowledge Graph (GKG) of the Global Database of Events, Language, and Tone (GDELT) characterizes news articles using named entities, topics, quantitative and temporal features, quotations, tone, and dictionary-based content measures. Its codebook defines the record format, but the software that produces the records is not public, so researchers cannot generate compatible records for articles they collect themselves. We built an open-source reconstruction, gdelt-regkg, and measured how closely it agrees with GDELT's archived (native) records for 193,550 English articles from 8,729 publishers (2020-2025) and 2,500 articles in five other languages. Agreement depends strongly on the field. English tone, the percentage of positive minus negative words, has a mean absolute error (MAE) of 1.79 points and a concordance correlation coefficient of 0.75. Identity agreement (F1) is 0.65 for persons, 0.42 for organizations, and 0.65 for locations. It is lower for date components (0.42), complete counts with their objects and locations (0.07), and quotations (0.07), while GCAM category presence reaches micro F1 0.84 within the implemented dictionaries. In subsets where retrieved and native titles and lengths match, tone error falls to 1.47 and organization F1 rises to 0.48, yet most disagreement remains. A larger gazetteer raises coverage of native places from 89.5% to 97.1% with little change in location F1. Tone computed on English translations of non-English articles is less accurate (MAE 2.10-2.64) than for English articles. The results are conditional on article pages that could still be retrieved, and the correspondence subsets do not isolate the effect of input differences. Reconstructed records should therefore be assessed field by field and not treated as completely interchangeable with native ones.