Download chapter_4part1

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Database wikipedia , lookup

Entity–attribute–value model wikipedia , lookup

Open Database Connectivity wikipedia , lookup

SQL wikipedia , lookup

Microsoft Jet Database Engine wikipedia , lookup

Microsoft SQL Server wikipedia , lookup

Extensible Storage Engine wikipedia , lookup

Clusterpoint wikipedia , lookup

Relational model wikipedia , lookup

Versant Object Database wikipedia , lookup

Database model wikipedia , lookup

Transcript
BIT 3193
MULTIMEDIA DATABASE
CHAPTER 4 :
QUERING MULTIMEDIA DATABASES
Language expression that describes
the data to be retrieved from database.
Data item : as output
Information base : search is to be made
Condition : have to be satisfied for a
data item to be selected
Manipulation (editing and modifying
data)
Presentation
Analysis (indexing and searching)
• Multimedia queries can be of different types (deal with
different media and with different properties)
• The process of querying multimedia data is complex and
it can approach this in TWO ways:
• WHAT INFORMATION CAN BE RETRIEVED
• HOW THE INFORMATION CAN BE RETRIEVED
• There are THREE levels of complexity associated with
the “WHAT”:
a) Level 1 :
• Retrieval of primitive features such as color,
shape, texture, spatial location and object
movement.
• Query example:
“ find clips of objects flying from top-right to
bottom-left of screen”
b) Level 2 :
• Retrieval of logical features related to the
identity of the object within the media.
• Query example:
“ find clips of an aeroplane taking off”
c) Level 3 :
• Retrieval of abstract attributes associated with
an understanding of the nature or purpose of
the object.
• Query example:
“ find a picture of nutritional disasters"
• The “ HOW” can also be classified on the basis of
whether the information is retrieved by:
• Query on the Content (Content based query)
• Query by Example (QBE)
• Time Indexed Query
• Spatial Query
•Application Specific Query
• The content of media information is
described by the metadata associated
with media objects.
Content
Based
Query
• Hence, these queries have to be
processed by:
• accessing directly the metadata
• then the media objects
• Example:
“Show the details of the movie where
a cartoon character says:
“I like you”
• Have to be processed by finding a
similar object that matches the one in
the example.
Query by
Example
(QBE)
• The query processor has to identify
exactly the characteristics of the
example object the user wants to match.
• The similarity matching required by the
user can be on texture, color, spatial
characteristics or the shape of the
object.
• Matching can be exact or partial.
• For partial matching, the query
processor has to identify the degree
of mismatch that can be tolerated.
Query by
Example
(QBE)
• Then the query processor has to apply
the cluster generation function for the
example media object.
• These cluster generating functions map
the example object into an mdimensional feature space.
• Object present within this distance d
are retrieved with a certain measure
of confidence and are presented as an
ordered list.
Query by
Example
(QBE)
• Here the distance d is proportional to
the degree of mismatch that can be
tolerated.
• Example:
“ Show me the movie which contains
this song: Flying Without Wings”
• These queries are made on the
temporal characteristics of the media
objects.
Time Indexed
Query
• The temporal characteristics can be
stored using segment index trees.
• The query processor has to process the
time indexed queries by accessing the
index information stored using segment
trees or other similar methods.
• Example :
Time Indexed
Query
“Show me the movie 30 minutes after
its start”
• There are made on the spatial
characteristics associated with the
media objects.
Spatial
Query
• These spatial characteristics can be
generated as metadata information.
• The query processor can access this
metadata information to generate the
response.
• Example:
“ Show me the image where Ali is seen
to the left of Abu”
• Application specific descriptions can be
stored as metadata information.
Application
Specific
Query
• The query processor can access this
information for generating purposes.
• Example :
“ Show me the video where the river
changes its course”
• Involve single media.
• Example : text
Text Media
Query
Text
Index
Text
Database
Other
Media DBs
Response to Query
Figure 4.1 : Processing Single Media Query
• Query Process:
• Assuming the existence of metadata for the text
information:
• the indexed file is accessed first
• and the information is presented to the user
• Involve more than one media.
• Example : text and image
Text &
Image Media
Query
Text
Index
Text
Database
Image
Index
Image
Database
Response to Query
a) Accessing Text Index First
Text &
Image Media
Query
Image
Index
Image
Database
Text
Index
Response to Query
a) Accessing Image Index First
Figure 4.2 : Processing Multiple Media Query
Text
Database
• Query Process:
• can be in TWO different ways:
a) Accessing Text Index First
• Select an initial set of documents
• This set of documents are examined to
determine whether any documents contains
the image object specified in the query.
• This implies that documents carries the
information regarding the contained images.
b) Accessing Image Index First
• Select a set images.
• Examined to determine whether images are
part of any document.
• This strategy assumes that the information
regarding the containment of images in
documents are maintained as a separated
information base.
• Access to multimedia information must be quick so that
retrieval time is minimal.
• Metadata must be stored using appropriate index
structure to provide efficient access.
• Index structures to be used depend on the media, the
metadata and the types of the query.
• SQL language is relevant to multimedia data.
Standard language for dealing with
relational databases
• developed from an earlier Structured English Query
Language, SEQUEL.
• with a lot of improvement, SQL become the de facto
standard of the database world.
• includes the specification for a data dictionary called the
information schema.
• Involves TWO process:
• create the structure of the table
• followed by inserting the data
• example:
• Create the structure
CREATE TABLE employee
(employee_no
CHAR (4),
employee_name
VARCHAR2(30),
salary
NUMBER(6,2))
• Insert:
INSERT INTO employee
(employee_no
CHAR (4),
employee_name
VARCHAR2(30),
salary
NUMBER(6,2))
VALUES
(‘123B’, ‘Ali Ahmad’, ‘1666.22’)
• Update:
UPDATE employee
SET employee_name =‘Abu Ahmad’
WHERE employee_no = ‘123B’
• SQL statements:
SELECT
(select list)
FROM
(table list)
WHERE
(search condition)
• Example:
SELECT
employee_name
FROM
employee
WHERE
employee_no = ‘123B’
• Text objects consist of words.
• Easy to recognize because within the body of text can be
delineated by spaces.
• Words consist of character strings.
• These features enable to process easily using query
language such as SQL.
• However there are some problems with words
• meaningless words such as “and, but, the, from, to”
• these are called stop words
• deal with them by removing or ignoring them
• synonymy when a word has the same meaning as
another word
• polysemy when a word has more than one meaning
in different context
• Principles of techniques specialized for text:
• text retrieval (TR)
• Schema-directed extraction (SDE)
• query-directed extraction (QDE)
Text
Retrieval
• simplest form of processing
• documents are returned in the result
set if they are considered relevant to
the query
• method to achieve:
• exact matching
• inexact matching
• proximity searches
• intelligent searches, ect.
• using BOOLEAN queries
SELECT cocoa_name, price
FROM cocoa_list
WHERE region = ‘Ghana’ AND category =
‘ABC’
• the uncertain spelling problem can be solved by using
SOUNDEX function
SELECT cocoa_name, SOUNDEX (cocoa_type)
FROM cocoa_list
WHERE SOUNDEX (cocoa_type)=SOUNDEX
(‘couverture’)
Coco-type
SOUN
------------------------------couverture
C130
• SOUNDEX codes
Table 4.1 : SOUNDEX codes
Character
SOUNDEX Codes
B,F,P,V
1
C,G,J,K,Q,S,X,Z
2
D,T
3
L
4
M,N
5
R
6
• vowel sounds “A,E,H,I,O,U,W,Y” are not assigned a
value.
• can be used in both exact and inexact search conditions.
• LIKE keyword is used to match a column with a known
string or part of a string
• example:
SELECT coco_type
FROM coco_list
WHERE coco_type LIKE ‘cou%’
• also use LIKE to search topics within character data
• example:
SELECT coco_type
FROM coco_list
WHERE character LIKE ‘%chocolate%’
• This query can be expressed using natural language as
“Give me type of each cocoa that has character that is
described by a string that contains the substring
“chocolate” with any number of characters before or
after”
QueryDirected
extraction
• it is useful for semi-structured text
documents such as business or scientific
domains where documents are laid out
to conform to a pattern
SchemaDirected
extraction
• process that manipulates large
collections of related text objects such
as e-mail
• schema :
e-mail {sender, receiver, date, subject,
persons, places}