The dramaanalysis from quadrama

Genre classification stuff

Add genre classification stuff from Kleist paper

Read directly from XMI files

It should be possible to load data directly from a directory containing XMI files. Options:

Integrate a Glassfish server that runs locally, call it via URL
Call java code directly from R

The latter is probably easier to maintain and distribute, but needs a lot more new code.

Reporting is broken

The file Report.Rmd can not be found, because it is not installed in the correct directory.

Handle multiple document collections

It should be possible to work with texts from multiple collections side by side

Add corpus column
Switch figure to cast figure
Map CastFigure to Figure

frequencyTable grouping criterion

Grouping criterion should be settable. It's currently set to figure or drama only.

List all installed files

It should be possible to get a list of all available files.

Identify graph core

Identifying the independent set based on all drama characters would be cool.

Idea from Solomon Marcus:

Für die Szenenführung ist der Begriff des Kerns von besonderer Bedeutung. Die szenische Anlage eines Theaterstücks impliziert die Wahl eines Kerns von Charakteren, d.h. des Kerns eines Graphen, auf den sich die gesamte Aufführung stützt. Meistens läßt ein (selbst endlicher) Graph mehrere Kerne zu; es kann aber auch vorkommen, daß er überhaupt keinen Kern zuläßt.
(Marcus, 1973)

dictionary.statistics() on lemmas or surfaces

If should be selectable whether we count lemmas or surface forms

Odoardo is split into two figures in included data

warning message for n750.0 and rhzq.0

Warning messages:
1: In scan(file = file, what = what, sep = sep, quote = quote, dec = dec,  :
  invalid input found on input connection 'http://zwergdrossel.ims.uni-stuttgart.de:8080/drama.web/annotations/n750.0/de.unistuttgart.ims.drama.api.Utterance/de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Token'
2: In scan(file = file, what = what, sep = sep, quote = quote, dec = dec,  :
  invalid input found on input connection 'http://zwergdrossel.ims.uni-stuttgart.de:8080/drama.web/annotations/rhzq.0/de.unistuttgart.ims.drama.api.Utterance/de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Token'

Include annotation counting functions

Functions to quickly count the number of utterances/sentences/token/...

Configuration matrix figure component

The component contains a data.table instead of a vector.

Versioning and release workflow

Lemma - R (vs webtool)

normalisierung in r checken, da andere ergebnisse als auf website

Versioning of data files

Package needs to keep a record of date/version of currently downloaded archive.

Release 1.1.0

Personnel exchange

Integrate code to calculate

Drama change rate
Scenic difference
Hamming distance

into the package.

Suffix of dictionary file settable

Currently, files have to end on .txt, this should be settable.

Update stored data to be a data.table

Loading of files with default collection

It should be possible to specify a default collection and then just provide text ids. collectionId:textId should not be used anymore, if possible.

Plotting: Strip chart utterance points

`frequencytable()` produces weird output

Some results just can't be. Association between words and counts is lost somewhere.

figureStatistics export wrong column

Install XMI files locally via R

If we post XMI files to ~~figshare~~, they could be loaded from there.

Overhaul interface of dictionaryStatistics() function

The function should return a list with various components, one of them a matrix with the actual values, similarly to the configuration()-function

Wiki guides should become vignettes

http://kbroman.org/pkg_primer/pages/vignettes.html

Collection handling

There is currently no convenient way to handle the collections we have in quadrama/metadata.

Improve meta data export

multiple authors or dates

Dates are now a feature of the Drama FS and thus accessible directly
Author or Translator annotations can be extracted using getAnnotations() (verification needed)

Retrieval of meta data

DocumentMetaData features can not be retrieved right now.

Word frequencies for sub corpora

Add a method to generate word frequency lists for sub corpora

Fix encoding issues in Windows

On windows, all data is interpreted as latin-1 (cp 1252).
It has been suggested here to use readr to re-encode the data as UTF-8.

Function name conventions

All functions should use a unified name convention, probably lowerCaseCamel

Utterance statistics

Add function for collecting utterance statistics.

https://github.com/nilsreiter/QuaDramA-Internal/wiki/R-VW-Snippets#simple-figure-statistics-emilia-und-sara

Vignette about loading data

We need an explanation on how to load data

Add document id to DramatisPersonae export

loadText() returns N.A.-values for columns Speaker.figure_surface and Speaker.figure_id

loadText() returns NA-values for columns Speaker.figure_surface and Speaker.figure_id

Export as CSV is broken in some cases

When utterances don't have tokens in them, we get a line with less columns than expected. They should be either removed entirely or filled with null-values.

loadText(): load from factor

loadText() should be able to load from a factor, i.e. by applying as.character() before loading

Keyness calculation

Calculate keyness for words across figures or corpora.
First method:

Drama length

When quadrama/DramaNLP#10 is done, we can now longer count the tokens in the table to get the length of the drama, as some tokens might be present twice (if uttered by multiple figures)