Pratchett guide
what is it for
The work has been done for Data Visualization course as a peer-graded programming assignment.
Main purpose of it was to utilize multidimensional scaling as it deeply inspired me.
Choosing a data set I decided to explore into the worlds created by Sir Terence David John Pratchett in his books.
When one just starts reading Pratchett it is usual not to follow chronological sequence since it does not have much sense and does not generally satisfy current passion; chronological order fits the second traverse more.
Frequent question a newbie reader asks is: Which book should I take next? There are so many directions to move: pick another book
- about interesting characters,
- from specific universe,
- by themes vicinity or
- by mix of factors.
This visualization is intended to answer the question according to several heuristics.
implementation
Digging into Pratchett's books I've picked the following criteria to search similarities on:
- universe where a story belongs,
- storyline, as far as Terry Pratchett's worlds are too detailed to be uncovered from narrow angle,
- main characters,
- minor characters set,
- leading themes,
- co-authorship: was a book written in co-authorship or it is 'pure Pratchett'
- literature form.
Having 305 remarkable characters and ~20 themes, each data (a book) is reflected as a point in 328-dimensional space.
Weighting similarities with different values (for example 1.7 for common universe, 0.3 for common minor character or 2.6 for common storyline) I calculate affinities between each pair of books and then, having affinities, get distances.
Original visualization had been done in presentation format, but here 'storytelling' is affordable, so this edition allows a user to change weights for criteria and get different views on the same data set.
default picture
In the default weights of choice preferences (highest weights) are given to common storyline and main characters.
Even experienced Pratchett fan can find interesting fact from the visualization. For example, the default picture
- surprisingly reveals how Death character dense through most of Diskworld storylines, letting them stay relatively isolated at the same time;
- that City Watch storyline is not far from the rest of Diskworld, but significantly decoupled.
- We see that The Long Earth series were written in co-authorship and, together with The Bromeliad Trilogy and Johnny Maxwell storyline, acting in some sort of real world, resides separately.
- We see also that Pratchett prefers novels and has only three books without siblings.
library used
The picture implementation relies on awesome D3js. Annotations, controls and the rest of the page are done with jQuery. Here is the precise list of third-party libraries exploited: