Context Aggregation and Analysis

This is a demo platform aimed to facilitate the verification of UGC video content posted on YouTube and Facebook. In contrast to other approaches, which attempt to analyze the videos themselves for traces of forgery, this platform analyzes the video context: The characteristics of the poster, any relevant user comments, the local weather reports at the time of the event, and other contextual pieces of information are aggregated and presented to the user for analysis. To test the service, simply copy and paste a YouTube or Facebook* URL into the box, then click "Verify"

*Right click on the Facebook video and copy the video URL

Contact: {olgapapa,papadop}@iti.gr

Input Video:

Back to examples

or have a look at some explanatory examples below



Search in Twitter

Loading general...



Name Value
Video ID
Video Title
Video Description
Video Description Mentioned Locations
Video Upload Time
Video View Count
Video Like Count
Video Dislike Count
Video Comment Count
Video Duration
Video Dimension
Video Definition
Video Licensed Content
Video Recording Location Description
Video Recording Time
Name Value
Video Title
Video ID
Video Content Category
Video Content Tags
Video Desctiption
Video Description Mentioned Locations
Video Created Time
Video Updated Time
Video Comments
Video Embeddable
Video Length
Video Picture
Video Privacy


Name Value
Description Mentioned Locations
Created Time
View Count
Comment Count
Subscriber Count
Video Count
Video Per Month
About Page
Name Value
Fan Count
Info! Fetching more comments...
More Comments...

Possible Values:

No details available for input:
Date:Exact time

Possible Values:

All values are in UTC

Loading weather context...

Loading twitter context...

Info! Fetching more tweets...

*Tweets may be added outside viewport. Click 'reset' to reset zooming scale and inspect all tweets.

Timeline of Tweets
Total tweets: 0

Unreliable tweets: 0 (0%)

Reliable tweets: 0 (0%)

Loading results...

This video cannot be found!

Wind speed is caused by air moving from high pressure to low pressure, usually due to changes in temperature. Beaufort: More

Cloud cover values only vary by 0.03 from year to year, whereas the local, day to day variability in cloud amount typically rises to 0.3 over the globe. Most data sets agree on the fact that the land is covered by 0.10-0.15 less cloud than the oceans More

The international definition of fog is a visibility of less than 1 kilometre (3,300 ft); mist is a visibility of between 1 kilometre (0.62 mi) and 2 kilometres (1.2 mi) and haze from 2 kilometres (1.2 mi) to 5 kilometres (3.1 mi). Visibility of less than 100 metres (330 ft) is usually reported as zero. More

General information about the video and the channel it was uploaded, extracted from its YouTube context. For values that are not provided, the corresponding fields are left blank.
NOTE: Video Upload Time is in GMT.

These are all the comments that appear under the video. User name and date/time are provided. Comments are further filtered into Verification-related comments based on the presence of verification- related keywords, such as “fake”, “confirm”, or “location”.

These thumbnails are produced automatically by the video platform (YouTube or Facebook, respectively). There is a fixed number of them, as decided by the platform, extracted by regular time intervals through the video. Click the "Reverse image search" link to run an image-based search through Google.

Weather information, provided by Dark Sky. You can contrast the historical weather data for the time the video was supposedly shot, to the weather displayed in the video. Suggested location names and date-times are extracted from the contextual information, but if these values are not correct, you can manually provide your own. If a specific time is given, the service provides the weather at that time. Otherwise, it provides an overall report for that day”.

Provides information on twitter traffic around the video, and aggregate verification information on any Tweets sharing the video, calculated using CERTH's Tweet Verification Assistant

This is an interactive visualization of all Tweets sharing the video. Click on any one Tweet to see the original post.

Statistics of the credibility scores of Tweets sharing the video.

We have developed a machine learning algorithm for tweet verification, based on linguistic patterns from the text and features from the user profile, and trained on a large number of fake and real tweets. We apply the algorithm to all tweets sharing the video and, depending on the number of such tweets that the algorithm labels as “fake”, we can draw an overall conclusion on the probability that the video itself contains false information.

Indicates whether the video represents licensed content, which means that the content was uploaded to a channel linked to a YouTube content partner and then claimed by that partner.

Pages with a large number of followers can be manually verified by Facebook as having an authentic identity. This field indicates whether the page is verified by this process.