in reply to Markov Chain automata class

  1. First glaring error - 'use Security::Monitoring::Utils;' - that isn't included so I can't run it. When I comment that out, it seems to run fine.

  2. I'd give your POD a review with 'podcheker'. Run 'podchecker -w -w filename'; I get a lot of errors running your POD through it.

  3. Once that's corrected, adding to / organizing your POD so it's helpful to a first time user would be great. I was able to deduce from trial and error the basics I needed to get something running (see below), but a SYNOPSIS section usually includes this. What's in your SYNOPSIS section is usually put under a DESCRIPTION section.

FYI, I ran this on Windows x64 with Strawberry 5.18.1 (64-bit) and it ran fine with my \usr\ install from GnuWin32 (which included /usr/share/dict/words'.

my $sec = Security::Monitoring::Detection::Markov->new(10); $sec->sys_learn(); print "finger = " . $sec->get_score('finger') . "\n"; print "alkajsdfa = " . $sec->get_score('alkajsdfa') . "\n"; print "xxxphotoblatxxx = " . $sec->get_score('xxxphotoblatxxx') . "\n" +;

Output:

finger = 1.89777628064316e-006 alkajsdfa = 2.84666442096473e-006 xxxphotoblatxxx = 4.74444070160789e-006

Replies are listed 'Best First'.
Re^2: Markov Chain automata class
by QuillMeantTen (Friar) on Aug 24, 2015 at 18:47 UTC

    Thank you for your input, I'm sorry about the Utils module, its just a wrapper around nstore and retrieve so I can call it from everywhere in my code

    I'll run it through podchecker tomorrow and correct those silly mistakes
    I'll also reorganise the documentation to make the synopsis part really useful

    I'm most interested in your feedback about the scores, I have no idea what kind of threshold should be used for alerts, whether 10% is or ain't enough.

    I was thinking about studying the distribution of scores for the words of the dictionary around the mean (hopefuly looking for anything that could tell me it follow a normal distribution) but I'm not sure whether its the way to go

      Other than a cursory knowledge of the term, I needed Wikipedia to help with Markov chaining and the likes you talk about. I'm afraid I can't be of any help on the scores output, simply the overall architecture and user-friendliness of the module.

        Here is the new batch, I have used podchecker as you told me, now I dont have any more warnings.
        I have also rewritten the Synopsis section and the description section putting everything in its own place.
        It does look nicer on a perldoc.
        Eager for feedback and ways to get better, as always.