Re: [agi] Search text for 999999999999999999 words in no time

2019-09-10 Thread rouncer81
The data in the tree is the search items -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T44eb904095b7612b-M30e9418285af9ea76fe8ff32 Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Search text for 999999999999999999 words in no time

2019-09-10 Thread Stefan Reich via AGI
My code just does what yours does, lol. > 12 seconds to check the full 132MB for any of 72 words Hmmm... I am actually not _that_ impressed. Searching 10 MB a second? In RAM? The search tree should fit in the cache, so that can't be the problem. On Tue, 10 Sep 2019 at 12:03, wrote: > Correct.

Re: [agi] Search text for 999999999999999999 words in no time

2019-09-10 Thread immortal . discoveries
Correct. I put the search items in a tree instead of putting the data in a tree. The little tree searches the data. What exactly does your code do? I made it faster: https://www.dropbox.com/s/5lym9vl2fttrmgd/FasterFastSearch.rar?dl=0 268MB becomes 132MB and takes 12 seconds to check the full 132M

Re: [agi] Search text for 999999999999999999 words in no time

2019-09-09 Thread rouncer81
Accelleration structures is the way to get really fast matches,  even if you are on a really powerful parallel computer that can afford brute searching you still go alot quicker if you do it with a tree of some variety.    If you want to get a little robot to run around like a little insect or s

Re: [agi] Search text for 999999999999999999 words in no time

2019-09-09 Thread Stefan Reich via AGI
Actually I made a class doing the same thing, "OccTree " (occurrence tree). With the "add" method, you can add a whole word list at once. It was made in 2017. Do I beat you? :-D On Mon, 9 Sep 2019 at 21:01, Stefan Reich < stefan.reich.maker.of@googlemail.com

Re: [agi] Search text for 999999999999999999 words in no time

2019-09-09 Thread Stefan Reich via AGI
Interesting. The Huffman compression seems quite unrelated to the actual algorithm, but I see you do that to save memory. Your description is not entirely clear to me. You search for _multiple_ words at once (possibly very many without loss of speed), right? But the system only finds full words (o

[agi] Search text for 999999999999999999 words in no time

2019-09-08 Thread immortal . discoveries
Here's one of the things I just made. Try it out. I tested it on Windows. You can search for any word in a huge amount of data really fast as if it is a 1 word search. Runs on CPU. You can swap the 200MB in the src folder. Run it in Visual Studio 2019. You can edit the words I search for in main