Single Application with all Text Visualizations?

[vlandham]

The current form of phase 1 of this project is to have each visualization exist separately from one another. Each will live in a separate repo and be configurable through JSON-style input.

Processing and analysis will happen via the scripts provided. Input to the visualization will be output from one or more processing scripts.

It could be desirable as an end user to have all visualizations accessible via a single web front end. This application would provide the ability to input files from many of the processing scripts and then provide some GUI around selecting which visualization to match up to the input.

Is this aggregate application a priority for phase 1? or should we focus on getting more visualizations developed - with an eye towards this kind of future as a desirable end point?

[arnicas]

GUI aside, I like the idea of people who are a little more savvy being able to combine vis on a single page. We could provide a sample of how to do that in the Learn site, as you guys call it...

Maybe a GUI to select them (and for other aspects, like to upload docs/stopwords etc) could be a P2 for V1. I'd like to get as many different kinds of useful vis as we can make, independently, before we start combining, although I expect in the use cases we'll see some things that would go together well popping up.

[iros]

I think we could have a single aggregate repo at the end that pulls in all the other visualizations as submodules and builds a GUI, but we should maintain and build the individual visualizations in separate repos so that we are able to better track work, accept issues and generally maintain order of the repos. I think with multiple visualizations in one repo, too much "mysterious glue code" would be too easy to create, imho.

[arnicas]

A demo of a lot of them combined on one page would be a nice "documents report card" though -- maybe heading towards a single GUI view in a P2.

Image a report card where you upload a bunch of docs, and get for them in a nice comparative fashion:

Relative lengths Word counts, sentence counts, maybe punctuation counts (if you ask for it) Most common words in each part of speech (exluding stops), most interesting by TF-IDF (could be word clouds) Clustering result showing their relationship to each other Maybe a timeseries sentiment graph of each, using the simple sentiment wordlist searches I added in data

Optional search in the docs, using the concordance plots display.

Might be fun as a demo at the end.