From a reply by --CELKO-- to another post of mine I learned that
the data I'm working with is a poor fit for RDBMSs, and is better
captured with an Entity-Attribute-Value (EAV) model. (I.e. I'm
working with a huge number of entities, and a huge number of
potential attributes that these entities can have, but any one
entity has relatively few attributes.)

Does anyone know of off-the-shelf software to manage data organized
according to the EAV model?

Also, I am interested in any theoretical analysis of EAV, and also
in design tricks to minimize as much as possible the shortcomings
of RDBMS for representing this kind of data.

Any leads on any of these questions would be much appreciated.

Thanks!

kj

--
NOTE: In my address everything before the first period is backwards;
and the last period, and everything after it, should be discarded.

kj <socyl (AT) 987jk (DOT) com.invalid> wrote:




The whole point is that you should *_avoid_* using the EAV method and
try to properly model your data.



Paul...



--

plinehan __at__ yahoo __dot__ __com__

XP Pro, SP 2,

Oracle, 9.2.0.1.0 (Enterprise Ed.)
Interbase 6.0.1.0;

When asking database related questions, please give other posters
some clues, like operating system, version of db being used and DDL.
The exact text and/or number of error messages is useful (!= "it didn't work!").
Thanks.

Furthermore, as a courtesy to those who spend
time analysing and attempting to help, please
do not top post.

"kj" <socyl (AT) 987jk (DOT) com.invalid> wrote

Check out:

http://hdf.ncsa.uiuc.edu/

You're welcome.

--
Mail sent to this email address is deleted unread
on the server. Please send replies to the newsgroup.

In <mnrjg19n0rqanvu70jpi8578crpqablk06 (AT) 4ax (DOT) com> Paul <paul (AT) see (DOT) my.sig.com> writes:

"Entity-Attribute-Value (EAV) databases... are used in domains
where the number of potential descriptors (attributes) describing
an object is a couple of orders of magnitude greater than the
actual number of descriptors for a given object. For example,
when dealing with patient data across all clinical specialties,
the number of history elements,symptoms, clinical examination
findings, lab tests and so on ranges in several tens of thousands,
and this number is constantly growing. Yet, for a given patient,
not more than a few dozen types of positive or significant
negative findings are actually relevant. That is, the data is
highly sparse, and a set of conventional relational tables,
with one finding per column, would result in much wasted space,
because most columns would be null."

My data fits this description perfectly. Hundreds of thousands of
entities *and* hundreds of thousands of attributes. How do *you*
"properly model" such data, hm? As far as I can tell, the only
way to "*_avoid_* using the EAV method" in this case is to quit my
job.

kj
--
NOTE: In my address everything before the first period is backwards;
and the last period, and everything after it, should be discarded.

Do not use EAV; use XML instead. It still stinks, but it is better
than using SQL.

In <1124753679.540992.295430 (AT) f14g2000cwb (DOT) googlegroups.com> "--CELKO--" <jcelko212 (AT) earthlink (DOT) net> writes:

I don't know the first thing about XML servers. What's out there
that is comparable to PostgreSQL (the RDBMS we use)?

kj

--
NOTE: In my address everything before the first period is backwards;
and the last period, and everything after it, should be discarded.