context-based classification and

sounderslipInternet and Web Development

Oct 22, 2013 (3 years and 7 months ago)

65 views

A Semantic Web ontology for
context
-
based classification and
retrieval of music resources

Alfio Ferrara, Luca A. Ludovico, Stefano
Montanelli, Silvana Castano, and Goffredo
Haus

ACM Transactions on Multimedia Computing, Communications and
Applications, Vol. 2, No. 3, Pages 177


198

I. Overview

A.
A new
semantic description

of music
resources based on
context

and
genre

B.
A new method of assigning music
resources to genres based on context

C.
An application to searching

II. Semantic Description

A.
The World Wide Web: human readable

B.
The Semantic Web: machine readable

C.
Ontologies

III. The Ontology

Genre

Context

IV. Context

A.
Ensemble

B.
Rhythm

C.
Melody

D.
Harmony

IV. Context: Ensemble

Ensemble

Part

Instrument

#

has

has
instrument

has number
of players

Part

has

IV. Context: Rhythm

Rhythm

Episode

# / #

#

has

has time
signature

has number of
measures

Episode

has

IV. Context: Melody

Melody

Fragment

Note

has

has
highest
pitch

has lowest
pitch

Note

has scale
with first
degree

Note

Octave

Pitch

Accidental

has

has

has

IV. Context: Harmony

Harmony

Chord

Scale

has

Fundamental Degree

Bichord

#

Modifier

has

has

has

has

has degree
distance

V. Genre

A.
Dimensions

1.
ensemble (e.g., string quartet)

2.
dance type (e.g., waltz)

3.
critical (e.g., Romantic)

4.
form (e.g., fugue)

V. Genre

B.
Fuzzy membership:
“Along the genre
dimension ‘Critical,’ resource R belongs
to the category ‘Romantic’ with degree
0.7.”

C.
Mapping from context to genre
(incomplete):
“If number_of_parts = 4
and for each part number_of_performers =
1, then along the genre dimension
‘Ensemble,’ R belongs to the category
‘Quartet’ with degree 1.0.”

V. Genre

D.
Other classification methods

1.
manual

2.
prescriptive (supervised)

3.
emerging (unsupervised)

VI. Recap: The Ontology

Genre

Context

Mapping

VII. Application to searching

A.
Context
-
based discovery

B.
Genre
-
based discovery

1.
query by genre

2.
query by target

VII. Application to searching

Query by genre

Query by target

VIII. Next steps

A.
More classes and properties

B.
Comparing different genre taxonomies

C.
Extension to other media resources

The MX Formalism

A.
Layers

1.
general

2.
logical

3.
structural

4.
notational

5.
performance

6.
audio

The MX Formalism

B. Spine

Time

Space