Home

Welcome to the Relational-Network-Generator wiki!

The goal of this project is to ingest a list of trained relational statements and output a list of derived relational statements as well as generate a visualization of the output in graph form.

The script "relational permutation.R" now(10/17/2021) completes ingestion and outputs an edgelist for standalone use or network visualization in the form of a data table. The input structure is limited. Please read the code comments in the script for directions.

The script "Network graph visualizer.R" now(10/17/2021) ingests the edgelist table described above and outputs variants of network graphs depending on the filters and network options selected. Please read the code comments in the script for directions.

Next step(s) are to create an RMarkdown webpage that allows users to input their list of relational statements, select their visualization options, and outputs the edge list table and network visualization.

• Initially input and output would likely be in very simple forms like "a<b" or "DkuF."
• Long term goal for the website would be to be able to take human readable input in either simplified (e.g. dog less than cat) or sentence structure (The dog is less than the cat) form and output in kind.

Update October 28, 2024 I've wrapped the script in an RShiny app and hosted it at https://coachpatricksmith.com/rft-network-graph-demo for others to demo more easily. See the recent announcement for additional details. At this time, the simple form and a very rigid full sentence form input are implemented. Users can also select what edges are graphed in the visualization, and the output always comes in both edge list table and network graph form. I have not pushed the shiny app UI or the sentence form parsing to this repository. The rest of the functionality is the same.

Next Step(s)

• Add more relation functional classes to the derivation script (e.g., temporal, opposition, difference, hierarchy, & deictics)
• Improve the UI sentence form input to handle more diverse forms.
• Add a "Download" or "Save" option to the outputs.
• Improve the network graph visuals.