Frequently Asked Questions
- What is Truthy?
- What is a meme?
- How do you get your data?
- What is a diffusion network?
- What is Astroturfing?
- How can I tell what a meme is about?
- What is the interactive network?
- What is meme co-occurrence?
- What do the movies mean?
- How does Truthy predict political affiliation?
- How can I tell if a meme is truthy?
- Is this site liberal or conservative?
- How do you detect interesting memes?
- What is sentiment analysis?
- What technology do you use?
- Where can I read more?
What is Truthy?
Truthy is a research project that helps you understand how memes spread online. We collect a public sample of tweets from Twitter and analyze them. With our statistics, images, movies, and interactive data, you can explore these dynamic communication networks.
The name Truthy comes from a term coined by Stephen Colbert, truthiness, which describes claims that feel like they ought to be true, but aren't necessarily. Our first application was the study of astroturf campaigns in elections; ideas spread by astroturf techniques are typically truthy.
What is a meme?
A meme is an idea, value or pattern of behavior that is passed from one person to another by imitation. In the Truthy system, a meme can be a #hashtag, @mention/reply, URL, or phrase.
How do you get your data?
We collect a sample of tweets from Twitter via their Streaming API. This is a sample maintained by Twitter, and it is said to be random, i.e. representative. Although the website only shows statistics and visualizations for the past 3 months, we have been collecting this data since September, 2010 and continue to analyze it for research purposes.
What is a diffusion network?
A diffusion network is the graph obtained by connecting user nodes with edges that represent the spread of a meme. Edges can represent retweets (in blue) or mentions (in orange). We run a component analysis algorithm and visualize the largest connected component from the diffusion network for each meme.
What is Astroturfing?
Astroturfing denotes political, advertising, or public relations campaigns that are formally planned by an organization, but are disguised as spontaneous, popular "grassroots" behavior. The term refers to AstroTurf, a brand of synthetic carpeting designed to look like natural grass.
How can I tell what a meme is about?
The detail page for each meme provides a live Twitter stream of tweets for that meme, which can be helpful in understanding the conversations around a meme.
What is the interactive network?
This visualization is calculated every day from the random sample of tweets collected by Truthy over the past 90 days. We compute the users to show by selecting the top 25 users with the most mentions and retweets in. We then collect a small subset of related users using a snowball sample.
Larger circles represent users who have been retweeted more often. A thicker line between two circles represents more interaction between those two users. Circles may be colored according to our prediction of that user's political affiliation.
What is meme co-occurrence?
Co-occurrence is when two or more memes are present in the same tweet. The co-occurrence diagram on the meme detail page shows the top ten memes that co-occur with a given meme. Each line that passes from one meme to another shows how often that meme co-occurs with the others.
What do the movies mean?
They show a meme's activity over time. Truthy can generate three different kinds of movies. Co-occurrence movies show which other memes are co-occurring over the selected time, so you can get a sense about which topics people were discussing together. The two other kinds of movies visualize the retweet and mention networks, so you can see who was retweeting whom (about the particular hashtag), or who was talking to whom.
How does Truthy predict political affiliation?
The process is entirely automated; we have trained a classifier as described in our paper Predicting the Political Alignment of Twitter Users; because the classifier was trained on English-speaking users discussing US politics, we only predict political positions of users speaking English and discussing memes related to US politics.
How can I tell if a meme is truthy?
This site does not provide an explicit assessment of 'truthiness'. The analytics on this site are available as a service to the public. Hopefully the various views, statistics, and visualizations that we provide of the data will help you make up your own mind about particular memes. Whether it is possible to automatically detect misinformation (rumors, astroturf, malicious social bots, and so on) is the subject of ongoing research.
Is this site liberal or conservative?
Truthy may happen to track some political memes as they co-occur with keywords related to themes on which we focus. However, we are a non-partisan research group and there is no attempt to represent or support any political views. The mix of thousands of memes that you can see on the site is imply a reflection of the active chatter in the Twittersphere. In spite of this, in an ironic turn of events, our project has been the target of a disinformation campaign online and on TV. Read our reponse in the truth about Truthy.
How do you detect interesting memes?
The Truthy system focuses on memes that have undergone significant changes in relative volume and those that account for a significant portion of the total activity on Twitter.
What is sentiment analysis?
Sentiment analysis is the practice of using an automated metric to estimate the emotional content of text. Particularly, we use a custom implementation of the OpinionFinder algorithm, which finds emotionally-charged words in the stream of tweets and calculates a ratio of positivity to negativity.
What technology do you use?
We use a variety of tools to bring you the Truthy service. The overall effort is directed using our own custom scripting language called Klatsch. This language uses the Gephi Toolkit for graph layout. We also rely on a number of other publicly available tools, including Boost, Django, Google Chart Tools, ImageMagick, JQuery, d3.js, MPlayer, MySQL, and the Twitter APIs.