Bag-Of-Words To Visualization-Map

The utility visualizes documents stored in Bag-Of-Words input file ("-i"). Calculated positions of documents is than stored in .viz file ("-o"). Semantic space can be used to help at lowering dimensions ("-ssp"). Extra reliefs can also be calculated on already calculated positions of documents ("-viz").

The parameter "-ssp_dim" determines to how many dimensions input documents should be reduced using semantic space (if semantic space is given as input). The parameters "-max_step" and "-min_diff" determine how accurate visualisation will be. The higher the "-max_step" parameter and lower "-min_diff" parameter, the more exact it will be but it will also take longer to compute.
The parameter "-numrlf" determines number of reliefs that will be calculated. The parameters "-rlfx" and "-rlfy" determine the resolution of calculated reliefs. The parameters "-sigma" and "-k" determine the diversity of reliefs.

usage: Bow2VizMap.exe
-i:Input-BagOfWords-FileName (default:'')
-ssp:Input-Sematic-Space-FileName (default:'')
-viz:Input-Sematic-Space-FileName (default:'')
-o:Output-Points-FileName (default:'')
-ssp_dim:Reduced-Space-Dimension (default:50)
-max_step:Max-Number-Of-Iteration (default:1000)
-min_diff:Min-Iteration-Difference (default:0.0001)
-numrlf:Number-Of-Reliefs (default:4)
-rlfx:Width-Of-Reliefs (default:1000)
-rlfy:Height-Of-Reliefs (default:1000)
-sigma:Sigma-Parameter-For-RBF-Kernel (default:0.03)
-k:Sigma-Decreasing-Coficient (default:0.5)

Example 1:
Bow2VizMap.exe -i:fp6-ist.bow -ssp:fp6-ist.ssp -o:viz.viz -ssp_dim:100 -max_step:2000 -numrlf:10 -simga:0.05

The above examples visualizes documents from fp6-ist.bow by first projecting them to semantic space of 100 dimensions. It can do at maximal 2000 iterations and 10 reliefs are precalculated.