I've been studying the use of SQL to achieve the equivalent of FP-Growth pattern discovery of musical motifs. I'm using a rainbow table of 3.4million note vector trigrams (-88 to 88 for note vectors on a piano) to make this work. However I find that for the capabilities I want, this method is a bit rigid even if it's as fast as other loop&seek methods, some which allow weak matching.
However I am thinking of instead using R in conjunction with PostGIS geometric queries instead, with the goal of exploiting various features in PostGIS to analyze musical compositions, such as finding similar line segments and so on. Anyone know of any work on using GIS vector data as a solution for multidimensional time-series analysis of temporal datasets? Thanks in advance, Lee Rodgers
_______________________________________________ postgis-users mailing list [email protected] http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
