0:06 from the previous lecture you studied on
0:10 powerful functionalities of dbms at the
0:13 same time it was clearly noted that
0:15 conventional relational dbms is not
0:18 designed for complex data types such as spatial
0:19 spatial
0:22 data in order to fully utilize the
0:24 proposed framework you should have a
0:27 firm understanding of spatial database
0:29 management system which is the topic of this
0:36 structure now let's start with the
0:39 question what is spatial database
0:42 management system in other words what is
0:44 the difference of spatial dbms from
0:46 conventional relational
0:50 dbms literally speaking spatial dbms is
0:53 designed for handling spatial data such
0:56 as Vector data and restor data for
0:59 handling spatial data it needs object
1:02 relational Ms in which abstract data
1:05 type in other words user defined new
1:07 data types can be
1:11 managed because spatial data is saved so
1:14 that spatial dbms needs spatial quality
1:17 language equipped with spatial
1:19 operations and spatial indexing for
1:22 quity optimization and so on so spatial
1:26 dbms can be defined as a dbms with
1:29 additional capability of handling
1:31 spatial data
1:34 now let's think about why conventional
1:37 relational dbms cannot deal with spatial
1:40 data in fact relational dbms can handle
1:44 spatial data but the problem is it is
1:47 not efficient the figure shows the
1:50 limitation the cadest map can be modeled
1:52 and the data can be saved in real
1:56 relational dbms however the special
1:59 components polygon Edge and point should
2:02 be save in three different tables which
2:05 makes simple spatial operation very
2:08 complicated because all the three tables
2:11 should be visited using time consuming
2:18 operation so we needed another way to
2:20 deal with spatial data one simple
2:24 solution was so-called dual architecture
2:27 in which relation dbms manage only at
2:31 data and a separate file system is used
2:32 for spatial
2:36 data it can overcome the inefficiency of
2:39 relational dbms however the Dual
2:42 architecture cannot provide powerful
2:45 dbms features such as transaction
2:48 management and many
2:52 others it could be used for only single user
2:58 applications when object relation tbms was
2:58 was
3:01 introduced and it can deal with abstract
3:03 data types in
3:06 1990s spatial data can be tightly
3:10 integrated with dbns in other words
3:14 column data type can be polygon line or
3:17 Point as a result full functionalities
3:27 architecture I mentioned object
3:29 relational dbms quite a few times already
3:30 already
3:33 let's discuss more about what it is in
3:37 more detail world dbms can be considered
3:40 as the middle ground to bridge the gap
3:43 between relational dbms and the objectoriented
3:45 objectoriented
3:47 programming which supports three
3:50 functionalities complex data with user
3:53 defined Class Type inheritance object
3:56 Behavior with method World dbms present
4:00 the best of the two consequently can
4:02 deal with spatial
4:06 data among many or dbms object relational
4:07 relational
4:11 dbms throughout this course I will make
4:14 use of post s square and S extension for
4:16 special data types post
4:20 GIS which are opsource software
4:23 discussed in the second
4:27 week the example is showing how to
4:31 create userdefined data type address is
4:34 the case which is composed of load C and
4:38 zip code and then using a square we can
4:41 create a table named restaurant which
4:45 has a column address the final s Square
4:49 actually inserts a restaurant named
4:53 spaga to restaurant table as you can see
4:58 or dbms can store a complex data type in
5:00 table which was not possible in
5:06 dbms the figure shows a series of
5:11 processing in SQL window of postgis to
5:14 create a data type create a table and
5:17 insert a record which was described just
5:26 slide now we have a method to store
5:28 spatial data in
5:31 dbms the next issue is how to retrieve
5:36 spatial data from dbms as you studied it
5:39 for the retrieval process square is also
5:43 used that means that you need the square
5:47 with spatial operations for spatial data
5:50 SQL 3 which is a standard SQL
5:51 established in
5:55 1999 allows to support spal data type
5:58 such as Point curve surface geometri correction
6:00 correction
6:03 and operations such as spal reference
6:06 envelope some booing operations of Pop
6:09 topology and spatial functions such as
6:13 distance buffer insertion intersection
6:21 others joint operation is to connect two
6:25 tables based on clance check on common
6:28 field join is a key operation at the
6:31 same time a very expensive operation in
6:33 dbms C
6:37 processing actually indexing is to
6:38 basically speed up joint
6:41 operation likewise spatial joint to
6:45 connect two table based on spatial
6:47 relationship is also very powerful at
6:49 the same time
6:52 expensive in other words timeconsuming
6:55 operation instead of Correspondence or
6:58 exact matching of in regular joint
7:00 operation space joint is based on
7:04 special relationship such as intersect
7:08 contain covers and so
7:10 on let's take a look at an example of
7:14 spatial joint the quy is what subway
7:18 stations are located in Little Italy in
7:20 New York
7:24 City for the quy we need two tables New
7:27 York City neighborhood table which is
7:31 based on polygon and Subway table based on
7:33 on
7:38 point the query sets two variables n and
7:43 S to point the two tables and then find
7:46 Lally from New York City neighborhood
7:49 table and find the subway station of
7:53 which location is within Little Italy
7:55 then finally Lally and the subway station
7:56 station
8:00 names please note that for spatial
8:04 relationship a bulling function s within
8:07 is used for spatial joint
8:11 here the figures on the side illustrate
8:14 the query processing and the result of
8:21 view I brought in another example of
8:24 spatial joint the previous example was
8:28 spal joint between polygon and point
8:31 this time it is between polygon and
8:35 line the query is what are the RADS
8:39 which cross the boundary of Corona
8:41 neighborhood the S square is similar to
8:43 the previous one but this time the
8:45 spatial relationship is defined by a
8:48 bullan function s
8:51 crosses from all the street of New York
8:55 City the Quarry retrieves only Street
8:57 which meet the boundary of Corona neighborhood
9:02 and now you are looking at the Quality
9:05 result in table View and map
9:08 view now let's think about the same
9:12 query with different data set the query
9:15 is what are the
9:17 laws which cross the boundary of D
9:18 County in
9:23 Wisconsin for the query we are using
9:26 county map and load network of the whole
9:29 country for the query all the Road from
9:33 Alaska to Florida should be checked if
9:36 they meet the boundary of B County how
9:41 about that would it work yes it works
9:51 time that's why spatial indexing is
9:55 required in spatial dvms for efficient
9:57 quity processing of the previous spal
10:01 joint what if we have to check the rows
10:04 only around D count in
10:06 Wisconsin the quality processing can be
10:08 done very
10:11 quickly spatial indexing is to provide a
10:14 better search performance of spatial
10:18 context for that R Tree which is a
10:22 extension of B tree was
10:24 introduced there are many other spal
10:27 indexing method but Archer is considered
10:30 as the standard additionally it's worth
10:33 mentioning geohash which is a powerful
10:35 method for spatial data searching and
10:38 organization which is going to be used
10:46 Data the figure is composed of point and
10:49 rectangles for the given spatial data
10:53 you can apply Arch tree based on MBR
10:56 which stands for minimum bounding
10:58 rectangles you are looking at the given
11:02 spal data set which are organized by two
11:08 different level of MBR m n o and P
11:11 represent four higher level mbrs and
11:14 each MBR has three or four lower level
11:19 mbrs for example M has AB b c d as a
11:21 lower level
11:24 MBR hierarchical structure can be
11:27 transferred to three structure as you
11:29 can see now which is
11:35 R3 similarly to B Tre it stores data
11:38 sets on the Le noes and it is a balanced
11:42 tree quad tree is another spatial data
11:44 indexing method as you can see the
11:48 example qu tree partitions to this space
11:51 into four quadrants
11:55 recursively and save the data to the end
11:58 nodes the figure shows the corresponding
12:00 point region quad
12:05 tree each node has exactly four child
12:09 nodes and the data is saved to the end
12:11 nodes which one do you think is better
12:14 than the other R Tree or quod tree of
12:17 course it depends on your spatial data
12:20 set generally speaking if you have
12:24 unevenly distributed data set R Tree
12:27 would be better because it can balance
12:29 the shape of the tree structure
12:32 on the other hand if you have evenly
12:35 distributed Point data quad Tre would be
12:41 choice now let's take a look at real
12:43 example of applying spal data indexing
12:46 and its performance the given query is
12:49 what are the name of the rows in
12:53 manhatton that is more than 30 ft width
12:56 in which SD contain function is used for
13:00 spatial relationship for the spatial
13:03 without indexing on the left and with
13:05 indexing on the right the difference of
13:10 the Quality Pro performance is 47 second
13:11 versus 1
13:14 second it is significant
13:17 Improvement what if we deal with the
13:19 load networks of the whole country the
13:29 phenomenal now rest the difference
13:32 between GIS and spatial dbms again we
13:35 discussed it in the second week GIS has
13:38 a variety of spatial data handling
13:41 capability but only to a certain degree
13:45 s dbms spal dbms is basically designed
13:48 for spatial data management with all the
13:51 wonderful features inherited from
13:59 dbns in this lecture
14:01 we started with the definition of
14:05 spatial dbms and object or relational
14:08 dbms was introduced as a dbms solution
14:11 for complex data types such as spatial
14:15 data Square for spatial data and spatial
14:18 joint was discussed finally two
14:20 different spatial indexing methods were
14:23 introduced and performance enhancement
14:26 with real example was given for your
14:28 understanding of the value of indexing method
14:30 method
14:32 all right this is the end of this
Chrome Extension