Talk:Functional dependency

Makes databases sound harder than they are

This entry needs great simplification. It makes databases look like an alien and incomprehensible concept. 203.171.78.120 (talk) 02:29, 28 May 2008 (UTC)

Is actually well read and well written, but people don't read textbooks so they don't know

Rating it as low importance is also a sign of ignorance of the Wikipedian authority, unfortunately, because functional dependency is a concept that actually gives more coherence to the process of normalization, and is necessary for relating the duality of normal database terminology , ( primary key, alternate key, foreign key, table), and the language of the research papers that make normalization a rigorous part of database design ( functional dependency, normal forms, dependency preservation, lossless-join decomposition, closure sets of FDs using Armstrong axioms ). —Preceding unsigned comment added by 218.215.151.38 (talk) 14:45, 6 April 2010 (UTC)

Less jargon?

anyone explain this in a better way? I have an exam in Data management soon and I can't find one good explination of Full Functional Dependency because they all go into jargon la-la land using very specific database jargon, and understanding it is like trying to understand something in Arabic via Latin while your main language is Russian! Just a simple, (very plain) English explination of what all this means would be very much appreciated.

As far as I can tell/decypher, it full functional dependency is when you have two attributes, and one of them will only ever have one value for each of the other one, and partial dcependency is when there can be several values of that attribute at once for one value of the other. For example, if you have an Aircraft carrier, it will have one captain, so the relationship Aircraft Carrier -> Captain is fully functionally dependant, but the Aircraft Carrier will have many planes, so the relationship AC -> Planes is only partially dependent. Is this close to the mark? 81.129.82.108 06:21, 11 January 2007 (UTC)

Relation?

The link to the Relation disambiguation page doesn't seem to be relevant to this database-specific use of the term 'relation'. --VinceBowdren 15:53, 16 August 2006 (UTC)

The definition that 'each X value determines at most one Y value is incorrect. It should be 'each X value determines EXACTLY one Y value.

The test to detemine functional dependency: X->Y TEST: If t1[X]=t2[X] then t1[Y]=t2[Y] i.e. Whenever two tuples agree on their X values, they also agree on their Y values.

Examples?

This could really use a specific example or two. —The preceding unsigned comment was added by 209.208.140.194 (talk) 22:41, 26 February 2007 (UTC).

Yes, please add some examples. I think that the Aircraft Carrier -> Captain example is correct but the AC -> Planes example misses the mark. AC-> Planes is just a 1:M (one to many) relationship between two separate entities (the Aircraft Carrier and the Plane are two separate entities). Functional dependency has something to do with data within a given table (basically a synonym for "Relation"). But I am running out of steam here and can't quite get all the way to a good example. Can someone please help?

Small suggestion

Might it be worth putting in a redirect from Functional Dependency for people like me who get confused when that comes up with no page? (86.20.35.62 12:43, 28 March 2007 (UTC))

Irreducible function depending set

Surely this should say "A functional depending set S is irreducible if the set has the three following properties:" and not that the set is "not irreducible" if it has the properties? Maybe my understanding of it is incorrect but surely if the set had those properties then there would be no way it could be reduced any further. If i am wrong then surely it would be better to say that it is reducible, instead of using a double negative in saying it is "not not reducible" 86.0.59.40 14:10, 21 May 2007 (UTC) Razamafez

Properties of functional dependencies

Original: Extension: If X -> Y and W -> Z, then WZ -> YZ Suggest (with some trepidation): Extension: If X -> Y and W -> Z, then WX -> YZ--Jim P. 01:28, 12 July 2007 (UTC)

The latter implication you propose is a theorem by X → Y implies WX → Y (by Pseudotransitivity) and W → Z implies WX → Z (ibid) and these together imply WX → YZ (by Union). You can also prove that by computing the closure of WX under {X → Y, W → Z}, yielding XYWZ, which includes YZ. Your first proposed implication is incorrect though because the closure of WZ under {X → Y, W → Z} is only WZ, which does not include Y. You can trivially construct a counterexample based on that too: {(W:1, Z:2, Y:3), (W:1, Z:2, Y:4)}. Which reminds me that I should add the notion of closure to the article... Tijfo098 (talk) 16:55, 20 November 2012 (UTC)

FP

This needs examples of functional dependency in Functional Programming languages, namely Haskell —porges_(talk) 08:50, 31 July 2007 (UTC)

It's not the same concept as far as I can tell, so it should not be covered on this page. Added a hat note to that effect. Tijfo098 (talk) 13:38, 20 November 2012 (UTC)

Or maybe it is related, judging by Mark P. Jones's paper abstract [1]. I've not read the paper yet, so no comment on the details. Tijfo098 (talk) 13:41, 20 November 2012 (UTC)

Ok, I've looked at the papers. It's not sufficiently related to cover here. Tijfo098 (talk) 15:00, 20 November 2012 (UTC)

Definitions flip-flops between sets of attributes and attributes

In the definition sometimes X and Y denote attributes and sometimes sets of attributes. This should be cleaned up as it makes the definition incorrect.

64.180.111.6 (talk) 00:16, 27 May 2011 (UTC)

I fixed that. The problem stemmed from the fact that some textbooks (e.g. 1st ed. of Garcia-Molina, Ullman and Widom) were actually defining a fd as taking only one attribute on the right. Given the union and decomposition properties (aka split/combining property), it doesn't make much practical difference though. Tijfo098 (talk) 15:03, 20 November 2012 (UTC)

Soft FDs

Quick note here to add a summary of soft FDs and their applications to query plans [2]. Tijfo098 (talk) 07:26, 24 November 2012 (UTC)