pileofrogs,
This really isn't on topic, so let me suggest turning it into a challenge that is.

Assume we have a list of multi-token words. In fact, assume that it is obtained using the following code:

#!/usr/bin/perl
use strict;
use warnings;
use WWW::Mechanize;

my $mech = WWW::Mechanize->new(autocheck => 1);

for ('a' .. 'z') {
    my $url = "http://wordlist.com/index-$_.htm";
    eval { $mech->get($url) };
    if ($@) {
        warn "Unable to get link for letter '$_': $@\n";
        next;
    }
    for my $link ($mech->links) {
        my $word = $link->text;
        next if ! $word || index($word, ' ') == -1;
        print "$word\n";
    }
}
[download]

The challenge is to make the longest chain of multi-token words where the end of one word overlaps the beginning of the next word by at least one token. Each word may only be used once in the chain. Here is an example:

area code
code of ethics
ethics committee
[download]

Here are some (likely not all) edge cases that I thought of.

• Overlapping tokens must match exactly - for instance, 'area code' can't be followed by 'coded message' because 'code' ne 'coded'
• Each new word must increase the chain length by at least 1 token. For instance, you can't claim 'bloody murder' follows 'yell bloody murder' because 'bloody murder' is completely within the preceding word

I realize that this particular list stinks but I couldn't find a better one. If you use an alternate source, please link to it so that others may compete using the same list. Oh, I am pretty sure there is a fairly well known computer science problem hidden within so heuristics solutions are likely necessary.

It should be fairly obvious, but here is a hint if you are having a hard time dealing with such a large list: