in the database
ilationship is
lat properties,
icular property
what properties
th the property
on of adjacency
plicitly encode
SURVEY
CONTROL
V
files
s that will be
i to a specific
y to cover only
i quick response
desirable that
ist in the area
ng the inquiry,
dex is being
will be divided
l tiles. These
in terms of the
s is no plan to
h map sheets or
f the earth’s
timize the size
>t is more than
retrieval time,
iber of entities
i tile, and the
sgion covered by
n been proposed
ept listing the
red by the tile,
to restrict the
e entities that
of interest.
1 be defined by
Dverlapping the
fy the entities
istrates how the
. Usually, the
m required, but
>ical approaches
e discussed in
The relational
arity, as shown
M's DB2 (Date,
it is necessary
se analogue of
processing, and
mnections among
:k approach, on
>rd types, the
so necessary to
INDEX_1
R0AD_17 PT 921 .. . PT 1004
INDEX_2
WELL_2 3
PLAN 24 PT 5197 . . . PT 5203
INDEX_3
R0AD_17 PT_1004 . . . PT_1029
PLAN_92
INDEX_4
R0AD_17 PT_1029 . . . PT_1102
PLAN 24 PT 5203 . . . PT 5197
sets. Restricted access (such as read-only access)
for certain classes of users, oi different "user
views", can be provided by means of different
subschemas. And, on the VAX-11 DBMS in particular,
retrieval efficiency can be optimized by means of a
storage schema, which provides details on such
things as indices, and attempts to store related
items on the same page of disk storage.
5.2 Infological Model
The infological model is related to human concepts
and real world representation. In other words the
data are organized in terms of representing reality
the way humans perceive it. The most common
infological model is a hierarchical tree structure
which would break the data into levels as shown in
Figure 4. This type of model is feasible to
implement in a database but it does have some
limitations. If there are a large number of classes
near the root of the tree it is difficult to
navigate around the data to establish and utilize
the various relationships. This type of model also
has a problem with data redundancy, which arises
from the need for each entity to have its own
non-unique attributes. This results in
nonstandardized entity/attribute records, where a
geology polygon would require fewer record fields
than the same polygon representing surface cover.
Figure 3. Spatial indexing
define connections between different record types.
These connections are defined by means of DBTG
(Database Task Group) sets.
The relational approach offers the advantage of
flexibility, since there are no pre-defined link
ages. Any query that can be formulated in the query
language can be accomplished. For this reason,
relational DBMSs, such as ARC/INFO (Dangermond and
Freedman, 1984), are popular for LRIS design. The
network approach, on the other hand, offers the
potential advantage of greater efficiency of those
queries for which suitable linkages have been
pre-defined. Of course, when suitable DBTG sets
have not been pre-defined, a query may either be
impossible or very inefficient to answer. In a
network database one is said to "navigate" through
the database, and in such a database system it is
necessary to structure the record types and DBTG
sets in such a way tnat frequently occurring queries
can be satisfied by relatively simple navigation
through the database.
As a general rule, the inclusion of more DBTG sets
improves retrieval speed, but may make the speed of
updating, both insertion of new records and changing
values in existing records, slower, because more
pointers and indexes must be changed. In the
current project much of the data is not volatile;
much of it will change rarely, if ever. Even those
data that do change will not need to be updated in
"real-time". Consequently, the design of the
database is oriented toward efficient retrieval at
the cost of slower updates.
GEOLOGY
SURFICIAL BEDROCK
ALLUVIUM COLLUVIUM AEOLIAN SEDIMENTARY METAMORPHIC IGNEOUS
Figure 4. Hierarchical infological model
With these limitations in mind, a different
infological model has been designed. This model
also takes into account the nature of the network
database model, which allows definition of backwards
and forwards file pointers. This infological model
utilizes paradigm files to reduce redundancy and to
allow easier navigation and establishment of rela
tionships. A paradigm file contains a set of
standardized entity attribute records as shown in
Table 2.
Table 2.
Geologic bedrock paradigm
file.
Geologic
Geologic
Formation
Rock
Structure
code
age
name
type
D
Devonian
undiff
Dbf
Devonian
Banff
Dbw
Devonian
Bow
J
Jurassic
undiff
Jcl
Jurassic
Coleman
This prototype LRIS will use the VAX-11 DBMS,
which is a "CODASYL compliant" network data base
management system, and which runs under the VMS
operating system on a DEC VAX-11/750. A good
Fortran interface, and the retrieval efficiencies
that are possible with a network database system are
among the reasons for choosing it. As with all
CODASYL network DBMS's, a database is defined by
defining a schema, which contains, among other
things, definitions of the record types and the DBTG
These paradigm files are very stable, allowing easy
updating and multiple entry to and from various
different entities. Thus, it is possible to
standardize the basic entity attribute record
structure by utilizing the pointers to these files
as shown in Figure 5. This reduces data redundancy
in many cases and provides a framework for the
definition of the relationships between the various
data.