ransomts / litexplor Goto Github PK

https://rstudio.github.io/shinytest/articles/shinytest.html

We've got some functions written in the eric interface, how are those included in the user_interface.rmd?

https://docs.ropensci.org/aRxiv/

https://serpapi.com/google-scholar-api

This one is lowest priority

take the years from the interface and hook them into the existing data communication functions, default to Jan 1 for the beginning and Dec 31 for the end

heatmap between two term groups, rows and columns are combinations of terms within group, groups being compared are selected by two drop down menus

intragroup:
Hasse diagram, where nodes are labeled with paper counts, possible color shading

extra

• flame graph https://www.brendangregg.com/FlameGraphs/cpuflamegraphs.html
• dendrogram http://r-statistics.co/Top50-Ggplot2-Visualizations-MasterList-R-Code.html#Hierarchical%20Dendrogram
• logic puzzle representation?

http://r-statistics.co/Top50-Ggplot2-Visualizations-MasterList-R-Code.html

https://docs.ropensci.org/rplos/

https://peer.asee.org/advanced_search?q=A+OR+B&collection_id=&year=&published_after=&published_before=

https://connect.ebsco.com/s/article/EBSCOhost-API?language=en_US

https://github.com/r-lib/progress
this library might work well, unless there's a shiny bar that would be better for the UX

note: don't use the dplyr estimated progress, that's been deprecated

https://docs.ropensci.org/jstor/

get the heatmap visualization to be interactive through plotly:
https://www.data-to-viz.com/graph/heatmap.html

a kable seems like it would be ideal if the reactive side isn't too hairy

first column would be the human readable eric search query, second column is just the number of papers returned. Want to keep this very simple so at least the head of it can be displayed to the user to demonstrate functionality

Found proquests api: documentation

Here's their example curl request:

curl -H "Authorization: Bearer 8461684a-5a79-407c-9321-93d8d719ddeb" -X POST -H "Content-Type: text/xml" --data "test and resultsmedlineprof50" https://api-dialog.proquest.com/v1/search > results.xml

So it'd need a different query formatter, plus I don't think it supports json, so the xml would need parsed. Also unsure what's going on with their versions, seems like v1 is still in production but v2 is documented?

https://eric.ed.gov/?api

With some searches there are hundreds of thousands of results and other there are just 10 or 15. This poses an issue when it comes to displaying a plot of all the values. Currently we are either displaying all values limited only by the max or we're capping the plot to where the small data values are visibly but others are off the charts.

Potential Fixes

• Logarithmic scaling
• Interactive plot (zooming in and out)
• Different plot type

https://pkgdown.r-lib.org/

Get the github io page hosted as well

We have a weird set manipulation requirement to manage our search terms, and have a first implementation at 5cec144, thanks!

Thoughts on how to extend to n dimensions:

• recursive application of the 2 dimensional implementation
• binary encoding of all search terms, where search term groups are sections of bits. iterate over integers to generate our set
• R answers from this code golf page, no reason not to let R help us whenever it can