- Multiple-key index:
- Multiple-key index =
a special case of
a
multi-level index
using
different types
of search keys
in each level
(Multi-level index: see click here )
- Multiple-key index =
a special case of
a
multi-level index
using
different types
of search keys
in each level
- Schematically:
- Data on people who buy jewelry:
(age, salary (in $1,000)) A(25,60) D(45,60) G(50,75) J(50,100) B(50,120) E(70,110) H(85,140) K(30,260) C(25,400) F(45,350) I(50,275) L(60,260)
- A multiple-key index
on keys
(age, salary):
-
Partial Match queries
- The query specifies conditions on some dimensions but not on all dimensions
Example:
- Find all people wih
age = 25
Explanation:
- Use the age indenx to
find the index block(s)
for age = 25
- Then:
- Scan all entries in the salary index to find the records
- Use the age indenx to
find the index block(s)
for age = 25
Note:
- Multiple-key index
for partial match query
will only be
useful if
the first dimension is
given
- We cannot use
multiple-key index for
a query like:
- Find all people who earn $60,000 who buy jewelry.
-
Range queries
- Find objects that are located either partial or wholly within a certain range
Example:
- Find all people with:
35 ≤ age ≤ 50 50K ≤ salary ≤ 100Solution:
- Use age = 35 in the
first index and find
the first value that is
≥ 35:
- Use salary = 50K in the
second index and find
the first value that is
≥ 50K:
Then scan the index until salary > 100K
- Scan the
first index until
age > 50:
- Use salary = 50K in the
second index and find
the first value that is
≥ 50K:
Then scan the index until salary > 100K
- And so on....
- Use age = 35 in the
first index and find
the first value that is
≥ 35:
-
Nearest neighbor queries:
- The multiple key index
can help
in the processing
of Nearest neighbor queries
BUT:
- It involves a complicated expanding range search algorithm in "nearby branches of the index tree
- Sketch of the algorithm:
- Check for the
nearest neighbor in your
own bucket first:
- Expand the
search to a
sibling subtree only if
the distance is
smaller than the
current nearest neighbor:
If abs( x' - x ) ≥ distance(current nearest neighbor), you do not need to search that subtree !!!
- Check for the
nearest neighbor in your
own bucket first:
- The multiple key index
can help
in the processing
of Nearest neighbor queries
-
Where-am-I queries:
- Multiple-key index are not used in Where-am-I queries...