Media Cloud, jointly run by the Berkman Klein Center for Internet & Society at Harvard University and the Center for Civic Media at MIT, is an open source, open data platform offering a suite of tools for media analysis.
Media Cloud is free to use and allows you to analyze media attention and influence for topics you care about, better understand how narratives and ideas spread, and explore media trends. There already is a comprehensive webinar out there, that introduces the platform in ~20mins, so I am not going to explain how to use the tools too much, but rather give you an example of what you can do with it.
Media Cloud Tools
Currently, there are three main tools: Explorer, Topic Mapper, and Source Manager. With Explorer you can get a quick overview of a specific topic, whereas Topic Mapper allows you to explore topics in-depth. Source Manager lets you browse the media sources and collections in the Media Cloud database, and suggest sources to add to the database. I will introduce each tool in a separate post. The following is about Explorer.
After entering your search term(s), date range, and news sources, this tool will give you information on media attention, language, and people and places. You can also enter multiple queries, which is what I did for this post to compare right* (red) and left* (blue) media coverage of Facebook over the last month (1 March 2018 – 16 April 2018).
*based on relative number of tweets of election story urls by Clinton vs. Trump retweeters
The first tab, helps you better understand the media’s attention around your topic. The chart below visualizes the attention paid to Facebook over the last month, and you can see that right media sources consistently covered the topic more than their left counterparts (as measured by sentences/day). Hovering over the line chart tells you the number of sentences per day, and by clicking on one of the dots you can see the headlines for that day. The line chart also tells you that both media’s attention spiked: 20-21 March, 26 March, 28 March, 2 April, 6 April, and 10 April. To put these spikes into context, see the timeline of events below.
Next, Explorer shows the total number of stories, which include at least one sentence matching your query. And as could already be assumed right media sources published more stories on Facebook than left sources.
Explorer also tells you which themes were most prevealent in the media coverage. For the Facebook query, these look pretty similar.
The last section of the Attention tab, presents a list of sample news stories. To make sure this is all within copyright, Media Cloud provides the title, media source, date the story was published, and a link to the piece, but not the full text.
It would be good to know, however, how these stories are selected (same for the sample headlines mentioned further up). Does this happen randomly, or are the most influential stories/headlines shown?
The next tab lets you find out more about the language used in the coverage. It tells you, which words were used most frequently (row 1 below), which words were used together (row 2 below), and if you entered multiple queries, which words were unique to each query and which common to both (row 3 below).
There is not much difference when looking at the top words used by both media source categories, however when looking at the top words unique to each, you can spot a few interesting differences. For example, only right sources frequently referred to “Obama” and “Democrats”, whereas left sources mentioned “Republicans” (and there seem to be Spanish language sources as well).
If you want to find out more about a specific word within the context of your query, simpy click on it and Media Cloud will start a new query searching for “your initial query AND word”. In my case, that was “facebook AND Obama”. Through this new search, I found out that the discrepancy in attention was in fact quite striking. Right media sources published 484 stories mentioning Facebook and Obama, and used an average of 23.2 sentences per day, whereas instances for left media sources came to only 51 and 1.6 respectively. By taking a look at the sample stories, I assume that most stories published by right media sources focused on how the Obama campaign had used Facebook data (unlawfully). Since these two terms did not appear in left media coverage frequently, data for the Language and People & Places tabs is highly limited, which is why I did not continue with the comparison.
This is not a priority, but for better usability, it would be great if new queries (e.g., by clicking on a word) opened in a new tab.
It should be noted that the top words are based on a sample of news stories, so it is unclear how representative the results are of the entire set of stories. Nevertheless, this tool is great way to get a quick overview of how a certain topic is talked about, what potential subtopics are, and where (groups of) media sources differ in their reporting.
It would be good to know how this sample is selected, and if the word space is also based on a sample.
People & Places
The People & Places tab gives you an overview of who is being talked about (i.e., people and organizations/companies) and which countries are mentioned. Again, these look pretty similar in my query, but are still very interesting to look at.
The Bottom Line
Media Cloud Explorer is an awesome tool for getting a quick overview of topics that interest you. It’s also a great starting point if you want to delve deeper into a specific topic and are not quite sure yet where to start. Explorer can give you several signposts, such as media trends, themes, subtopics, and actors. Explorer is super easy to use, and even provides additional guidance on how to interpret the charts. Another cool feature is that you are able to download the visualizations (in some cases) and underlying data (in all cases). Explorer comes with a few limitations, but keep in mind that it’s all free to use, run through grants, and ALL OPEN SOURCE (contribute!! :))
One final note, be careful with Media Cloud as it can make time magically disappear! This started as a quick blog post, but then I fell down the rabbit hole. Great work, Media Cloud team!