Skip to main content

Overview

The Research Agent is an AI-powered interface that answers questions about global events in natural language. Ask questions in plain English and receive structured results drawn from GDELT’s media event clusters, entity profiles, and news domain data.

How It Works

1

Natural language input

Type your question in the chat interface using everyday language.
2

AI-powered analysis

The Research Agent understands your intent and determines which data to retrieve — media event stories, entity profiles, or domain information.
3

Data retrieval

The agent queries GDELT Cloud’s API to find matching media events, entities, and articles.
4

Results presentation

Data is returned with story summaries, article links, entity context, and sentiment metrics.

Example Questions

The Research Agent can answer a wide variety of questions about global events:

Event & Story Queries

What protests happened in France this week?

Entity Queries

What stories involve Donald Trump this week?

Geographic & Thematic Queries

What are the most significant stories in sub-Saharan Africa?

Features

Conversational Context

The Research Agent maintains conversation context, allowing follow-up questions:
You: What protests happened in France last week?Agent: Returns media event clusters about French protests with article links

Data Visualization

Results are displayed in structured formats:

Story Cards

Media event clusters with headline, category, article count, and sentiment

Article Links

Source articles with titles, domains, and publication dates

Entity Profiles

People and organizations with linked stories and co-occurrences

Sentiment Metrics

Goldstein scale and tone scores indicating conflict/cooperation intensity

Understanding Results

Media Event Stories

Each story is a cluster of related articles. Results include:
FieldDescriptionExample
cluster_labelStory headline”French workers stage nationwide strikes”
categoryTopic categorypolitics_governance
scopeGeographic reachnational
article_countArticles in the cluster47
avg_goldsteinConflict/cooperation (-10 to +10)-5.0
avg_toneSentiment score-3.2

Entity Profiles

Entity results include:
FieldDescriptionExample
display_nameEntity name”Emmanuel Macron”
typePerson or organizationperson
total_articlesArticles mentioning this entity312
linked_storiesMedia event clusters involving the entity28
co_occurrencesOther entities frequently mentioned alongside”Marine Le Pen”, “EU”

Goldstein Scale

The Goldstein scale measures event cooperation/conflict intensity:
  • +10: Yield / major concession
  • +8: Formal agreement
  • +7: Consult / meet
  • +5: Promise / pledge

Tips

Be specific about time

Include time ranges like “today”, “this week”, or “in January” for best results

Use categories

Ask about conflict, politics, economy, health, disaster, crime, society, or tech

Explore entities

Ask about specific people or organizations to see their full story profile

Follow up

Drill into specific stories or entities from initial results