tag:blogger.com,1999:blog-1417239720543668306.post1729215412078412666..comments2023-10-26T13:49:05.161-07:00Comments on ElasticTruth: JonBenet Ransom Note Analysis Using Syntactic Ngrams -- Or Taking The Words Away And Looking At Structure.Tom Bergerhttp://www.blogger.com/profile/03196089855453332363noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-1417239720543668306.post-85684668879794947212018-12-11T21:14:02.195-08:002018-12-11T21:14:02.195-08:00Hi Asimos, thanks for your comments. Regarding P v...Hi Asimos, thanks for your comments. Regarding P values, I am using NPC TEST, non parametric combination software http://static.gest.unipd.it/~salmaso/NPC_TEST.htm and also the R script version by Devin Caughey of MIT. This makes no assumptions about the distributions and allows analysis for small samples. I haven't used it for this analysis but I am gearing up for major new posts. I haven't updated for a long time, getting up to speed with various statistical procedures and graphics. I will have all new material shortly. Thanks for your support and suggestions, I take on board what you say about P values. Cheers.Tom Bergerhttps://www.blogger.com/profile/03196089855453332363noreply@blogger.comtag:blogger.com,1999:blog-1417239720543668306.post-92118060805851374892018-12-11T03:08:20.996-08:002018-12-11T03:08:20.996-08:00Great article. It's a shame that this type of ...Great article. It's a shame that this type of AI-based analysis has not been established in forensics. If done properly, the result could be as reliable as that of DNA analysis (if not more so, since we are certain of the source of our data).<br /><br />It would be interest to get some p-values into the result in order to assess their statistical significance (the probability that the conclusions were not reached by chance). If you get significant p-values you should publish your work in an academic journal. Asimoshttps://www.blogger.com/profile/15924931164039897272noreply@blogger.com