GDELT Summary: Online News

Online News Summary

This service allows you to rapidly create a visual dashboard that summarizes global news media coverage of a particular topic. Using the options below you will select the dataset to search, the kind of dashboard you want, the specific displays to include in your dashboard and the search to run.

Step 1: Dataset

GDELT Summary supports searching either global online news coverage monitored by GDELT and its global partners or broadcast television news coverage monitored by the Internet Archive. You can use the dropdown below to switch to global online news coverage.

Dataset

Step 2: Output Type

GDELT Summary supports either a basic visual dashboard that summarizes coverage matching your search, or a comparison visualization that compares up to four searches to make it easier to examine differences in how various topics are being covered in the media.

Output Type

Step 3: Enter Search

Enter your search keywords below. All words/phrases should be in English, since you will be searching the English machine translations of all 65 live translated languages GDELT monitors (allowing you to search across languages). Phrases should be enclosed in quote marks. If you include multiple words/phrases all of them will be required to appear somewhere in a document for it to match. Only the exact keyword(s) entered are searched (searching for "russia" does NOT match "russian" or "russians"). You can perform limited boolean "OR" searches by enclosing a group of OR'd terms inside a set of parantheses - for example, to search for "syria" appearing near a set of russian-related terms you might search for "syria (russia OR russians OR russian OR kremlin OR putin)". You can also put a "-" in front of a given word or phrase to exclude articles that contain it. Remember that machine translation is far from perfect, so try to search for simple keywords and avoid phrases if possible to maximize results from non-English sources.

In addition to searching textual news articles, three special advanced search terms can be used to visually and textually search online news imagery. Each day, GDELT processes up to one million global news images from the articles it monitors through Google's Cloud Vision API deep learning algorithms, visually recognizing up to 10,000 distinct objects and activities, performing a reverse Google Images search to determine its popularity, identifying more than 2 million topics in most of the world's major languages from all captions used to describe the given image across the web, estimating how "happy" or "sad" the majority of the people in the image appear, transcribing background signage and text in 55 languages and extracting all embedded metadata, creating one of the most powerful visual search systems in the world for global news imagery.

To add a visual search term to your search below, type imagetag:"yourkeyword" to search the 10,000 visual topics, type imagetopic:"yourkeyword" to search the 2 million textual caption topics or type imagetext:"yourkeyword" to search the combined raw caption text, embedded metadata, recognized logos and OCR'd background text. You must include the quote marks even when searching for a word rather than a phrase. Both imagetag:"" and imagetopic:"" will autosuggest terms from the predefined list of available tags/topics as you type the first letter after the opening quote mark. Try imagetag:"protest" or imagetopic:"drone" or combine with a textual search like "climate change" imagetag:"protest".

Keyword(s)

TIME PERIOD: By default all worldwide online news coverage monitored by GDELT since 2017 is searched (except for maps which only search the last 7 days and cannot be changed). You can narrow the timeframe to a shorter window to examine a particular time period. (You can also select a "zoomable" volume timeline display in the next section that allows you to interactively narrow your search timeframe.

Time Period

Start Date

End Date

SOURCE COUNTRY: You can limit your search to only consider news published by outlets in a particular country (choose up to 4).

Limit To Country

EXCLUDE COUNTRY: Exclude news published by outlets in a particular country (choose up to 4).

Exclude Country

PUBLICATION LANGUAGE: You can limit your search to only consider news published in a particular language (choose up to 4). Note that you will still search using English keywords, since you will be searching the machine translations of this coverage).

Limit To Language

EXCLUDE LANGUAGE: Exclude coverage originally published in a given language (choose up to 4) (for example to filter out Russian language coverage of Russia).

Exclude Language

TEXTUAL TONE: You can limit your results to just articles that were fairly negative or fairly positive. Tone is calculated at the level of the entire article, not the tone of the sentence(s) mentioning your keywords, so a negative article with a positive mention of your keywords will still be scored negative.

Limit To Tone

IMAGE FACIAL TONE: You can filter by the average "tone" of human faces in the image, matching only those images that contain a majority faces expressing "happiness" or "sadness". If you did not include any image-specific search terms above, then this will filter your returned articles to just those that contain such images.

Facial Tone

IMAGE POPULARITY: You can filter by how many times Google Images has seen the image across the entire web, allowing you to select only common/iconic images or novel/rare imagery. If you did not include any image-specific search terms above, then this will filter your returned articles to just those that contain such images.

Image Popularity

SMOOTHING: By default, timeline visualizations will report exact values, but this can lead to a noisy graph that makes macro-level patterns more difficult to discern. To address this, you can enable smoothing that computes a moving window average to smooth the results and make patterns more apparent. Remember that moving window averages slide trends to the right and under heavy smoothing peaks can appear days to weeks later on the timeline.

Smoothing

Step 4: Displays

Your news summary dashboard is built up by appending together a series of displays and content sections. Use the dropdowns below to select which you'd like to include in your media summary.

VOLUME TIMELINE: The volume timeline is the most basic kind of visualization and shows you the percent of all worldwide coverage monitored by GDELT over the selected time period that matched your search or the actual number of articles that matched your search. It offers a quick visual gauge of how much news attention your search is receiving. By default the timeline offers a static display, but you can optionally include a zoomable timeline that allows the user to interactively zoom into a particular period of time to see only coverage from that period (such as to drill in to a particular day where coverage of the topic peaked). Annotated versions of both timelines are also available, in which hovering the mouse over a given day/hour will show several top relevant articles from that day, making it easy to rapidly get a sense of the trajectory of coverage over time.

Volume Timeline

GOOGLE TRENDS TIMELINE: Embeds a Google Trends timeline showing you changes in search interest over time. This allows you to compare trends in media coverage of a topic against search interest in that topic. Note that Google Trends only supports basic keyword and phrase searches, so this display will generate an error for more complex queries, including ones using OR operators.

Google Trends Timeline

TONE TIMELINE: The tone timeline displays the average emotional "tone" (from extremely negative to extremely positive) of all coverage monitored by GDELT by day/hour over the selected time period that matched your search. It offers a quick visual gauge of the emotional tenor / kind of coverage your search is receiving.

Tone Timeline

TONE BARCHART: The tone barchart is a unique display that bins each matching article by its average emotional "tone" into a combined histogram/barchart that ranges from extremely negative on the left to extremely positive on the right. Each bin shows you how much of the matching coverage had that tone. This lets you instantly see the distribution of tone across all matching coverage and you can mouse over each bar to see the top most relevant coverage matching your search that was scored with that tone level and you can click on each link to browse to that article. Note that this visual is based on the overall tone of the entire matching article, not just the part of the article relating to your keyword.

Tone Barchart

SOURCE MAP: The source map aggregates matching news coverage from the past 24 hours by country of origin (the country where the news outlet publishing the article is based). The end result is a map where you can click on each country to see several top relevant articles matching your search that were published in that country's press. While we strive for maximal accuracy, note that you may discover some news outlets that have been mistakenly placed in the wrong country. When using this map for reporting purposes, you should verify the location of a given outlet. Please email us with any corrections. Note that unlike the other displays, map visuals search a fix rolling time window of the last 24 hours that cannot be changed. Note that the source map display does not currently support searches with image-related terms and will not be shown if your search contains them.

Source Map

ASSOCIATED LOCATIONS MAP: The associated locations map shows the top thousand or so worldwide locations that are mentioned most frequently within a few sentences of your keyword(s) over the last 24 hours. The underlying algorithms that identify and disambiguate textual geographic mentions are 100% automated, so you may see errors in this map, but it offers a useful overview of what locations are most associated with your keywords over the last 24 hours. You can choose to map by country (all locations are aggregated to the country level) or by city/precise location. You can click on a country/point to see several top relevant articles that mention your keyword(s) in the context of that location. Note that unlike the other displays, map visuals searched a fix rolling time window of the last 24 hours that cannot be changed. Note that the associated locations map display does not currently support searches with image-related terms and will not be shown if your search contains them.

Associated Locations Map

IMAGE TAGS WORDCLOUD: The image tags wordcloud displays the common visual description tags generated by Google's Cloud Vision API deep learning algorithms for the top 75 most relevant images. Google's neural network algorithms visually examine each image and recognize up to 10,000 objects and activities, labeling with it the associated tags. If your search includes image-specific search options or filters, the 75 most relevant images will be used, otherwise if your search contains only textual search options, the first 75 images from the most relevant articles matching your search will be used. This wordcloud can be used as a powerful way to understand the visual narratives and topical focus of the images used to illustrate coverage of your search.

Image Tags WordCloud

IMAGE TOPICS WORDCLOUD: The image topics wordcloud displays the common textual topic tags generated by Google's Cloud Vision API deep learning algorithms for the top 75 most relevant images. Google's neural network algorithms read the textual caption of each image, along with any captions used to describe that image anywhere else it appears on the web (via a reverse Google Images search) and labels it with more than 2 million topical tags. Unlike the Image Tags Wordcloud above, this wordcloud reflects strictly the topics used to textually describe the images via their captions and thus may contain richer detail and context than visual analysis, but also may contain errors due to incorrect captions or caption identification. If your search includes image-specific search options or filters, the 75 most relevant images will be used, otherwise if your search contains only textual search options, the first 75 images from the most relevant articles matching your search will be used. This wordcloud can be used as a powerful way to understand the visual narratives and topical focus of the images used to illustrate coverage of your search.

Image Topics WordCloud

TOP MATCHING ARTICLES: You can include a short list of top matching articles that were most relevant to your search. The "Text List" mode returns top articles regardless of whether the article contained any images (this returns all online coverage GDELT found for your search), while "Visual List" only returns matching coverage that includes an editorially-selected display image (the news outlet embedded special HTML code in the article declaring an image to show when sharing or linking to the article). Using the Visual List mode means you can miss important coverage, but allows you to rapidly assess the visual portrayal of your search. Note that the Visual List mode can load slowly on slow internet connections due to the number of images that are loaded.