Feedback Needed: New Tagging Method
TappedOut forum
Posted on Jan. 26, 2017, 3:49 a.m. by Femme_Fatale
I present a new method of tagging our cards ...
Instead of keywords, cards are connected by what defines them and what traits they might have that aren't traditionally searchable by the normal search engines. Examples:
- A simple tag of "suicide" will show you all the cards that have life loss drawbacks but powerful P/T and effects.
- "Requests (Suicide)" will show all the cards that benefit from life loss drawbacks.
- Looking for the appropriate prowess/Kiln fiend enabler? "Combat Trick" might just be up your alley.
- How about methods to pitch "Target (Reanimation)"s to the grave? "Discard (Self)" and "Mill (Self)" can most certainly be of benefit.
The problems with generating such a list are:
- A. What should be included.
- B. How to eliminate redundancy.
- C. How to encompass as much variance as possible.
So why use this tagging method?
First off, the keyword section of this site doesn't have much of a value to us. It is used for nothing beyond searches and sorting the deck by said keyword, and even then situations like Action Words and Keyword Actions are not properly tagged here or by WotC. All forms of actions and keywords can be searched simply by typing that word into the text box field and voila! You'd get the same, if not even better, results. Keywords are still going to be a necessity as some combos require a specific keyword to function, but regardless, a larger tag database would be a great boon.
Next, comes the combase issue. Our new combo database has come with a few problems in maintaining the database, in that it simply doesn't make sense to put in every possible iteration of a combo that can blink a creature, sacrifice a creature for a benefit and a creature that can generate 2 other creatures when it enters the battlefield. Or any possible iteration with Ghave, Guru of Spores for that matter. In this, a system that can quickly and easily tag multiple cards together for synergy or an infinite combo without requiring the massive amounts of database work would be beneficial.
Finally, our draft sim has the potential to be much more than what it is, but in order to do so our draft bots need to have some sort of guidelines to follow when drafting.
However, we don't have to start from scratch.
Much of T/O is built on taking ideas from other sites and incorporating it into our own site. And even now, we will be taking ideas from other sites to forward this tagging process. The two sites I'm particularly interested in are :
Cube Tutor has attempted to crowd source information about what cards do. They are pretty successful in this attempt, though the methods of crowd sourcing currently leaves much to be desired as almost every single card has tags that will take years to fully be implemented. Their tag system is based entirely on cube archetypes, though there are plenty that are questionable for their inclusion since there isn't an easy method to search those tags.
Cube Tutor also has a "strategy guide" for their superior AI bots to follow, these are basically a guide for which colours are preferred, preferred cmc curve, and what tags should and shouldn't you include. The tags we will be implementing will allow us to also implement our own strategy guide, should yeago have the time to code it.
Where Cube Tutor is focused entirely on the draft environment, and will provide us with the largest portion of our tags, Essential Magic provides us with an insight in how the combase should be structured. Not only does it have a description base and whether or not it is approved, but it also allows for comments on the combo and ratings on the combo. The various amounts of combos lets us gain an insight into what sort of identifiers we need to account for the variability of creatures, and how the combo suggestion form should be structured.
I ask you, the community, for help.
Help in figuring out what tags should and shouldn't be used. There are 3 categories for our tags.
- Draft: A draft tag would be one that supplements potential draft archetypes like discard, reanimator, +1/+1 counters, fliers, tokens, RDW ect.
- Combo: A combo tag would be one that identifies a key piece of a combo. It is important to try and generalize the tag here and not worry about whether or not the tag results in synergy or infinite.
- Search: A search tag would be one that brings together common effects under one tag. Meddling Mage for meddle effects, Hypnotic Specter for saboteur, Kiln Fiend for Trigger (Cast Instant) and Trigger (Cast Sorcery) etc.
When suggesting a tag, please mention which categories you are prioritizing for this tag, but make sure to keep in mind that the more useful it is for the other categories, the more likely the tag will be implemented.
Here is cube tutor's tag list and derived from this our own tag list. Feedback is greatly appreciated, and if you think that a tag from cube tutor's list should be included, or a tag from our own list shouldn't be included do detail as to why.
Thank you in advance for your cooperation.
Femme_Fatale says... #3
It's purposes are hence:
- To give us a method in which to code draft bots to follow algorithms and archetypes for a better draft simulation.
- To give us a method to better streamline the management of the combase instead of having to individually go through each combo and compare it to existing ones and comb through gatherer to find the better iterations of it.
- To allow users to have an alternate method of searching cards. It would be clearly defined as "tags", probably right above the keywords one if we plan to keep it.
I've already been working with yeago on this on our discord server, its purposes can present us with new ways to organize and provide data for future use. Cube players will definitely like this also as sorting by tags can give them a clear idea of what is present in their cube and what needs to be tweaked to better balance each archetype.
January 26, 2017 4:53 a.m. Edited.
MagicalHacker says... #4
I can see your points about the keyword sort not being useful. However, the times I have used it, I always wished I could instead search for words or phrases in my deck list. For example, let's say I am curious as to how many cards in a deck care about basic lands. I would be able to type "basic" into the rules text search and see the cards separated into which ones have and which ones don't have that particular word. Another example is searching a deck to see how many warriors you have in the deck. Again, search in the subtype text search for warriors and get the deck seperated into warriors and nonwarriors.
I think this would be unimaginably more useful since it means each user could search for what they care about. On top of that, there would be no creation or management of categories, which I think is important as well.
January 26, 2017 9:21 a.m.
I think we should identify the main keywords we want before we go and try to revamp the system.
January 30, 2017 7:39 a.m.
Femme_Fatale says... #6
No no, keywords are either being untouched or just simply tossed as the hold nothing of benefit for us, and with MagicalHacker's suggestion would have truly no use. As for trying to identify what tags to use Atony1400, that is the entire purpose of this post, to get people to critique the tags and to supply other systems that work with the combase to get it functional.
February 1, 2017 4:49 a.m.
I'm a bit late to the discussion, but why even bother building the list yourselves? It seems like a huge amount of work that will require a huge amount of maintenance, both of which are wholly unnecessary - you already have access to users' decks, and i'd bet a reasonably useful/significant number of users utilize Custom Categories. You could just use the categories already defined by the userbase. This would constantly evolve as new cards and playstyles are introduced.
EmblemMan says... #2
I am not trying to be negative or shoot down your idea I am just giving my honest opinion. I would never use this. I would also not think to use it unless I knew it existed which I do not think most new users would. It also just seems like a pretty complicated thing to set up with very little pay off.
January 26, 2017 4:48 a.m.