Database classification
Mats Winther
Here is a proposal for classification of chess
databases. It builds on the functionality of today´s products on the
market. Please, comment on this.
There are three types of databases: (1) Key databases (ChessBase,
NICBase, ClubMate and others), (2) Tree-databases (Bookup, Chess
Assistant), (3) Key-tree databases (TascBase).
Key databases sort games and analytic material under keys analogous to
ECO keys. The user may define personal keys which are permanently
stored. When games are imported they are stored
under the defined keys. This is in accordance with the chessplayers
instinct to define a repertoire of his own. There are always
pre-defined keys so that it becomes easy to investigate
new variations. However, to define an exact repertoire requires a lot
of work. And the keys have to be altered as the theory is
re-evaluated. On the other hand, if the user doesn´t take on
this key definition labour then this type of database tends to become
merely a chess game archive on which certain search algorithms are
applied. Waiting for the search results can be somewhat frustrating.
This type of database is recommended to personalities who like to
bring order to chaos by creating a stabile repertoire.
Tree databases don´t use keys. Instead each game is regarded as a
variation in a big variation tree. By merging all games a tree with
variations and sub-variations is created. The user can
classify the games by creating different trees. For example, it´s
quite convenient to create a tree for each ECO-code or by a personal
definition. Tree databases easily create perspicuity
which facilitates learning. However, the tree easily becomes
contaminated if inferior games are imported. Then it becomes a djungle
of amateur variations mingled with grandmaster variations.
On the other hand, the perspicuity of these databases lessens the need
for a repertoire-tree completely cleansed from bad variations. These
databases suit players who are pragmatical,
readily changes variations and who often visits the theoretical
quagmire where reality tends to shift.
However, without a great amount of pre-defined keys, less cunning
players may feel a little lost in a wariegated variation tree. The
game material which generated the tree may hold bad variations
and others may be completely missing. This problem doesn´t exist in
key databases or key-tree databases which always have predefined keys
according to ECO, NIC etc.
Key-tree databases are tree databases wherein every node (half-move)
is a key. This is an attempt to combine the advantages of the key and
tree databases respectively. The ramification of the tree is reduced.
Comparatively, the tree databases let the tree grow all the way down
to the last move in every game. But in a key-tree database the user
preferably works like this: the tree is created interactively, a bit
at a time, by expanding the games that are stored in
the leaf-node (the last half-move). So the keys are created
automatically whilst doing theoretial study of the variations. This
reduces the contamination problem which is so typical
of the tree databases. The key databases´ demanding key labour is here
handled automatically by expanding of the games or by moving the
pieces at the board. However, here is missing the rapid expansion of
every game to the last move, provided by the tree databases.
By game importation the games are automatically sorted in under
respective key. No new tree branches are created like in the tree
databases. If the user wants additional expansion, this
has to be done with a command.
The only available key-tree database, TascBase, creates a graphical
tree in addition to the normal interface provided by Bookup and Chess
Assistant.
Conclusively I would like to mention that ChessBase now can merge
several games and by this it gets something of a tree database effect.
Correspondingly, Chess Assistant lets the user define positions which
can be loaded to find out if the position exists in the tree. In this
way the weakness of respective database type is somewhat compensated.
Mats Winther
Wybe Koopmans
Mats Winther wrote:
>Key databases sort games and analytic material under keys analogous to
>ECO keys.
Not always. I use for my Tasbase 2.0, not the ECO-keys, but the
NIC-key-classification (the old Nicbase). For example: not keys like B00,
E87 (ECO) etc, but SI (Sicilian), Al (Alekhine), KI (King-indian).
Regards,
Wybe Koopmans
ChessBase GmbH
Mats Winther wrote:
>
> Here is a proposal for classification of chess
> databases. It builds on the functionality of today´s products on the
> market. Please, comment on this.
>
> There are three types of databases: (1) Key databases (ChessBase,
> NICBase, ClubMate and others), (2) Tree-databases (Bookup, Chess
> Assistant), (3) Key-tree databases (TascBase).
>
Good classification. I will keep myself short to avoid endless ChessBase
plugs:
a) CB6 comes with a ready made ECO key of 45.000 variations.
b) CB6 can produce/modify key files automatically ("Refine key")
c) A function that is easily overlooked in Fritz5: Use CTRL-C (Button
'Classify') in a tree position to access the related openings key in the
current database. Then press CTRL-A in the database window to select all
games and hit ENTER to merge them.
--
Matthias
mailto:wuelle...@compuserve.com
http://www.chessbase.com
Please DO NOT mail to 'wuelle...@t-online.de' (dead letter box)
Komputer Korner
Your definition and segregation of the different classes of databases is a
very good attempt at wading through the differences, but please realize
that Fritz 5 also has a tree like CA and the information in the Fritz 5
tree is more extensive as regards to transpositions and that you can use
its' tree like an opening book. Also the trees in CA2, Fritz 5, and Bookup
do not retain the individual game header information. TascBase has it's
tree integrated into its' database view but it has no intrinsic advantage
by doing this since you can't edit the tree at the same time as watching
the analysis engine. All the trees/databases can carry bad lines/games and
no product will guarantee that only the best lines/games are kept.
ChessBase 6.01, Rebel 9 and Tascbase 2 can dump selected game openings
into a repertoire database or an opening book.
--
- -
Komputer Korner
The inkompetent komputer
If you see a 1 in my email address, take it out before replying.
Please do not email both me and the r.g.c.c. at the same time. I read all
the postings on r.g.c.c.
Mats Winther <mats.w...@swipnet.se> wrote in article
<34268a6...@nntpserver.swip.net>...
> Here is a proposal for classification of chess
> databases. It builds on the functionality of today愀 products on the
> market. Please, comment on this.
>
> There are three types of databases: (1) Key databases (ChessBase,
> NICBase, ClubMate and others), (2) Tree-databases (Bookup, Chess
> Assistant), (3) Key-tree databases (TascBase).
>
> Mats Winther
>
Mike Leahy
Mats Winther wrote:
>
> Here is a proposal for classification of chess
> databases. It builds on the functionality of today愀 products on the
> market. Please, comment on this.
>
> There are three types of databases: (1) Key databases (ChessBase,
> NICBase, ClubMate and others), (2) Tree-databases (Bookup, Chess
> Assistant), (3) Key-tree databases (TascBase).
>
> ECO keys. The user may define personal keys which are permanently
> stored. When games are imported they are stored
> under the defined keys. This is in accordance with the chessplayers
> instinct to define a repertoire of his own. There are always
> pre-defined keys so that it becomes easy to investigate
> new variations. However, to define an exact repertoire requires a lot
> of work. And the keys have to be altered as the theory is
> this key definition labour then this type of database tends to become
> merely a chess game archive on which certain search algorithms are
> applied. Waiting for the search results can be somewhat frustrating.
> This type of database is recommended to personalities who like to
> bring order to chaos by creating a stabile repertoire.
>
> variation in a big variation tree. By merging all games a tree with
> variations and sub-variations is created.
While BOOKUP is probably best labelled a 'tree' database, it generally
is not used for storing games at all. No games are needed to create
a perfectly keyed positional tree in BOOKUP.
With a couple exceptions, our databases contain no games. The variations
have been carefully edited to represent a complete repertoire or endgame
course or tactical puzzle set.
The automatic option of "contaminating" any book-on-disk with PGN games
with inferior variations is completely under the contol of the user. :)
[snip]
> Mats Winther
Mike Leahy
"The Database Man!" http://www.bookup.com