We have been monitoring all tweets around the Museum Week since a month before the event up to the end of March. This year Museum Week has used 7 different hashtags, one for each day (from March 23rd to March 29th) plus the general one: #museumweek.

People have followed the suggested rule and most of them have changed the hashtag, as it is shown in the following Topsy chart, where you can see the first 3 days. In last year’s Museum Week, 40,000 users were involved (people, museums and organizations), and they published around 180,000 tweets. This year, the figures have more than tripled: almost 600,000 tweets published by more than 140,000 users. There has also been a significant increase in the number of museums engaged in Museum Week.

Museum week whole graph

Even in these cases we analyze the complete dataset and we perform the community analysis, the influence rankings and the calculation of the descriptive measures referring to the whole graph (such as the average distance, the diameter or more sophisticated measures as the clustering coefficient) over all the gathered tweets.

Communities and how do we detect them in a graph

The procedure is technically very complex and there are different algorithms that analyze graph structures. You can find a brief definition of the main algorithms at Wikipedia, but even with this high-level description, the topic is quite difficult.

The main point is that we analyze the graph structure, split it in different communities and then (and only then) label them. In fact, the list of communities is the result of the analysis; so, if we talk about a French community is because the graph structure shows that there is a distinct group or community whose members are mainly French. That does not mean that all French users are in this community or that all users in the community are French, but most of them are.


Nobody has asked about this one, but it is important to understand what does it mean. A connected component is just a group of nodes that are connected through a path; the path may contain as many steps as needed.

When analyzing a museum’s environment on Twitter, as in our 2015 Museums and the web paper, the data gathering procedure is defined in such a way that all resulting Twitter users belong to a single component.

There are a few things that stand out to the experienced eye. 2015’s conversation is much larger, yet more compact. This is a good starting point to understand what’s happening and how has Museum Week evolved since last year:

  • Both the number of users and the number of tweets has tripled – the exact factor is 3.3 for users and 3.4 for tweets.
  • Last year, around 600 museums from few European countries participated, while this year Museum Week has involved 2,800 museums from 77 countries all over the world.