Word Clouds: Can they be misleading?

Theoretical (Analytical):

Practical (Implementation):

Literature Work:


Overview

As revealed in the VALCRI project, some of the criminal analysts like word
clouds as it helps them construct a narrative.  But can this narrative be distorted
by the order/arrangement/size of the words?

Problem Statement

If the mental construction of a narrative can be altered, then this is obviously a
serious problem, given the context of the work.  So the initial study would be to
investigate whether the order (position) that words are presented in word clouds
affects their interpretation. Other factors such as the size (or prominence due to
colour) of the words can be investigated as well. Also, are there ways of
presenting the words which do not bias the constructed narrative?  A literature
search reveals that a fair amount of work has been done on word clouds but
nothing specifically on constructing narratives.

Tasks

Design and conduct user experiments to investigate whether the narrative
created from a word cloud is influenced by the word order and other presentation
factors. An additional task would be to investigate others ways of presenting a list
of words which do not bias their interpretation.

Requirements

Reasonable programming skills and preferably experience of designing and/or
undertaking a user study. Note that we have already developed a web-based experimental framework (P5 & JavaScript) for another project, which can be adapted to conduct user studies.

Scope/Duration/Start

  • Scope: Bachelor/Master

    The potential scope of this project is large, but can be adapted to suit the particular timescale and interest of the student.

Contact

References

  • Ennis, A. (2010). Indicators of Content: the role of word clouds in the creation of
    summaries
     (Doctoral dissertation, Loughborough University). Available here:
    www.jibs.ac.uk/prize/ennisandrea2010.pdf