Page 1 of 1
Searching using Warmachine faction names - too common?
Posted: Sat Oct 06, 2012 3:10 am
by mysticarcher
I've been trying to use the search feature to check the Bits and Parts subboard for Warmachine bits (since where else could I find them easily)
But the search feature keeps coming back with:
The following words in your search query were ignored because they are too common words: menoth.
You must specify at least one word to search for. Each word must consist of at least 3 characters and must not contain more than 14 characters excluding wildcards.
This is obviously problematic since I would like an automated search for searching somewhere like the Bits and Parts section where the vast, vast majority of posts will be for 40k bits which I don't really care about.
I'm not sure there's really anything to be done about it, but I was hoping there might be.
Re: Searching using Warmachine faction names - too common?
Posted: Tue Oct 09, 2012 9:40 pm
by MagickalMemories
Nope. It's all about the commonality of the words you're using.
What words are you searching?
Eric
Re: Searching using Warmachine faction names - too common?
Posted: Tue Oct 09, 2012 10:17 pm
by mysticarcher
Nope. It's all about the commonality of the words you're using.
What words are you searching?
I've been using the faction names (Cryx, Menoth, Cygnar) as I thought using the specific warjack names might cause me to miss posts due to people's wording:
For ex. if someone listed like "Cryx helljack bits" and I was looking for some arms to make a reaper helljack searching for 'reaper' wouldn't bring that listing up.
Re: Searching using Warmachine faction names - too common?
Posted: Wed Oct 17, 2012 5:32 pm
by Plarz
I took a look at the search settings, and currently any word that appears in more than 5% of all the posts on the board is considered too common. Considering we have roughly 350,000 posts, that means if you were just under that 5% mark, your search would return ~17,500 posts.
Due to the nature of our forum, only the first 20-30 would be relevant, since folks post their goods every week or so, so after 30ish (to pull out a random number) you'll start seeing the same people's stuff.
The PP section of the board has 606 posts (at the time this post was made) out of the roughly 350,000. If everyone plays two factions, let's assume that 2/11 = 1/5 for a gracious 20% of the posts will contain any given faction name. 20% of 600 is 120, which is about 0.03% of the posts. In fact, all 600 posts in the PP board only constitute 0.15% of the total posts on the board. Now, this ignores any posts in the BTR section, or any of the conversational forums, so the actual number of posts that contain "Menoth" will be higher.
So, obviously the (very fudged) math doesn't support the fact that the search function won't let you use "Menoth".
I've upped the common word threshold % from 5% to 15%, and I'm rebuilding the search indexes in the database to implement that change. This should re-compute which search words are considered "too common" (ie, appear in more than the new 15% of the posts on the board).
We'll see where this lands us. Unfortunately, rebuilding said indexes takes several hours, so this is not a very fast process of trial and error to get searching working better.
Re: Searching using Warmachine faction names - too common?
Posted: Wed Oct 17, 2012 7:22 pm
by montaa
I got 58 responses for Menoth.
90 for Cryx.
83 for Khador.
81 for Signar.
16 for Mercenaries.
320pm EST 10/17/12
Re: Searching using Warmachine faction names - too common?
Posted: Wed Oct 17, 2012 7:50 pm
by Plarz
Good, the re-indexing is working. It's not done yet, but It will be by the end of the day today.