Text Mining and Self Organising Map Examples for AI/VISUAL Container Demo

This R script contains a set of text and data mining algorithms.

Options:
	-p INTEGER, --port=INTEGER
		bind to port [default = 8080]

	--shownodename
		display node name in title (helpful to distiguish app running in different docker containers) [default = FALSE]

	-s INTEGER, --somdim=INTEGER
		som dimension [default = 35]

	-i INTEGER, --somiterations=INTEGER
		som iterations [default = 100]

	-h, --help
		Show this help message and exit

Text Mining

• Normalized Compression Distance
• Multidimensional Scaling
• Affinity Propagation

For input a text or text file can be used with the following definition

<label1>: <any text>
<label2>: <any text>
...

A working Shiny application can run online within a browser or manually started wih the command source("main.r"). The application does the following steps:

1. cleaning input text, stopword reduction, remove punctation and numbers
2. runs the normalized compression distance to analyse dissimilarity of the text input
3. runs the multidimensional scaling to create a 2D projection of the dissimilarity data
4. runs affinity propagation of 2D projection to build groups of data

Self Organizing Map (SOM)

• Kohonen networks

Visualisation of Wine Data