Queries analysis

Speakers opinions analyzed through specific words

Introduction

Visualization aim is to create a pattern that at a glance allow to understand what speakers (of which polarization) differ more from each other through the use of a specific language. Furthermore it's possible to navigate within groups (composed of ten most used words) to read them individually and get an idea of the speech made by each speaker. Instead it's possible to start from the words listed below and see who makes use.

How to read the visualization

There are different possible way to read the visualization. First, it's possible to observe the pattern created by the square at the top, where media cross their orientation. Each group of ten words represents the ten most used words from the specific speaker (tf highest value), those colored in gray are those used also by other speakers, and so less relevant for characterizing the speech made by these speakers. The words colored in blue instead are used only by the speaker they are associated. Then it's possible to display the words from the groups at the top, passing the mouse over the squares and clusters that most interest, and turn to see the words represented. Otherwise you can start from the specific words that interest and determine who uses them and in what amounts. Furthermore, the words in gray are ordered and represented by a gray gradation that goes from a darker to a higher nuance, based on the number of repetitions of the word within the visualization.

How it has been done

The first step in the visualization realization was to order in groups of ten squares the ten most-used words from each group of speakers of a specific polarization. In the part immediately below were lists words that are repeated more than once time in the visualization dataset, have been colored in shades of gray from darkest to lightest according to the number of repetitions of each ones, and the same color is associated with the related square on the top. In the part below the words that do not occur, which have been colored blue to create a colored pattern on the top of the display. In order to allow a simple and dynamic reading we realized an animation by associating each square of the underlying part (used as a legend) to the relative squares located in the pattern at the top, the aim is to allows to turn on (red coloring) the words which are selected moving the mouse hover the squares.

Findings

The visualization show the peculiarity of some words, we can make a few examples:
the word government is the most repeated and thus more present, is used by all the speakers except to the political sphere. But, looking the most characterizing (blue) word used by political sphere (against) the first is the name of a petition: Writers against mass surveillance, the second word refers to people, making use of the word citizen. Another example can be given from the first most word used by academics against: constitution, and academics in favor: FISC (last amendment which increases the range of surveillance by the NSA).

Metadata

Timestamp: 24/11/2014 - 11/12/2014

Data source: Google, Sven

Related Protocol

Download data a (8KB)

Download data b (1KB)