#82 Corpus Level Analysis

[Mona Baker]

• Corpus level analysis
Facets to visualise (univariate visualisation, eg. bar chart):
- organisation
- category
- publication_date
- format
- article_type
- blogs
- online_magazine
- journal

** Two-facet visualisation (e.g. mosaic plot)
- <format, organisation="">
- <organisation, date="">
- <journal, date=""></journal,></organisation,></format,>

• ... (there are 8C2 = 28 combinations, many of which would
  not make sense as graphs, such as <blogs, journal="">)</blogs,>

** Facets to be treated separately (e.g. tree-maps)
- section/type (???)
- visual

[Mona Baker]

Corpus level statistics not working for GoK.

[S Luz]

Usability improvements:

• use lighter background to improve readability
• format numerals: e.g. 957899334 -> 957,899,334
• consistency: place close window button ('x') on top right-hand side for consistency with the menu window.
• Abbreviate textual labels when they exceed the horizontal space allocated to the chart (e.g. reduce the text to a set number of characters and make it expandable upon clicking or similar)
• Open corpus-level stats on a separate browser window so as to allow users to see the stats while working with the concordancer/mosaic etc.

[S Luz]

Further usability improvements:

• the lollipops get clipped for low frequency items (at least on Firefox); see attached screenshot.
• when you abbreviate long items, add a '...' to the abbreviated items (e.g. "The Lancet Global Health'
  -> 'The Lancet...', rather than simply 'The Lancet') to avoid confusion. See screenshot.