The process starts with the user navigating the Web and identifying a raw data- set. For instance, the Scimago Journal Country Rank 10 . The site provides links to visualize their data 11 but the parameters these visualization took are not the same as the ones provided in the rank, and if you already had filtered the data in the ranking you need to move to another context to start filtering again and visualize the data. So, consider that a user wants to take his customized ranking as the dataset to visualize and he chooses all the journals in the Computer Science field 12 . His aim is quickly identifying the journal with more references per document among the first 50 top-ranked journals by Scimago.

As you can observe in the next video, the site presents 50 results by page, therefore all the information required by the user is present by default. The third column in the table presents the ranking for the journals, and the last column, the number of references per paper. The user can choose a column to sort the data, but it applies to the whole dataset. This means that if he orders it by ”refs by document”, he may miss the first 50 ranked sites. The first step, then, is to create his dataset of interest; a subset of the full dataset provided by the site. To do so using our tool, he can press the browser action (a button in the browser’s toolbar) of the ALVis extension.