U.S. patent application number 13/835376 was filed with the patent office on 2013-08-08 for music recommendation method with respect to message service.
This patent application is currently assigned to Samsung Electronics Co., Ltd.. The applicant listed for this patent is Samsung Electronics Co., Ltd.. Invention is credited to Hyoung Gook Kim, Jung Eun KIM.
Application Number | 20130204878 13/835376 |
Document ID | / |
Family ID | 39061760 |
Filed Date | 2013-08-08 |
United States Patent
Application |
20130204878 |
Kind Code |
A1 |
KIM; Jung Eun ; et
al. |
August 8, 2013 |
MUSIC RECOMMENDATION METHOD WITH RESPECT TO MESSAGE SERVICE
Abstract
A music recommendation method and a music recommendation system
are provided. The music recommendation method includes: selecting
music files according to a theme of the message service and music,
a mood of the music, a similarity between content of the message
service and content of the music; and recommending selected music
files to a user.
Inventors: |
KIM; Jung Eun; (Yongin-si,
KR) ; Kim; Hyoung Gook; (Yongin-si, KR) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Samsung Electronics Co., Ltd.; |
Suwon-si |
|
KR |
|
|
Assignee: |
Samsung Electronics Co.,
Ltd.
Suwon-si
KR
|
Family ID: |
39061760 |
Appl. No.: |
13/835376 |
Filed: |
March 15, 2013 |
Current U.S.
Class: |
707/741 |
Current CPC
Class: |
G06F 16/685 20190101;
G06F 16/68 20190101; G06F 16/61 20190101; G06F 16/635 20190101 |
Class at
Publication: |
707/741 |
International
Class: |
G06F 17/30 20060101
G06F017/30 |
Foreign Application Data
Date |
Code |
Application Number |
Dec 13, 2006 |
KR |
10-2006-0127171 |
Claims
1. A music recommendation method with respect to a message inputted
by a user, for transmitting via a message service in a personal
communication terminal, the method comprising: selecting music
files according to a theme of the message, a theme of a music, and
a mood of the music, determining whether the number of selected
music files is greater than a predetermined number, and, if the
number of selected music files is greater than the predetermined
number, then further selecting music files from the selected music
files according to a similarity between content of the message and
content of the music; and recommending the selected music files to
the user.
2. The method of claim 1, wherein the selecting of the music files
further comprises: selecting target music files; and classifying
the target music files and the message according to a theme,
selecting at least one music file which corresponds to the theme of
the message according to a result of the classification, from among
the target music files, and generating a list including the at
least one music file.
3. The method of claim 2, wherein the selecting of the music files
further comprises: classifying each of music files of the list
according to a mood, and eliminating at least one music file, whose
mood is inappropriate for the theme of the message, from the
list.
4. The method of claim 2, wherein the classifying of the target
music files comprises: classifying lyrics of the music files
according to the theme; classifying a title of the music files
according to the theme; classifying the message according to a
theme; merging a result of the classification of the lyrics with a
result of the classification of the title; and the classifying the
target music files according to the theme, is based on a result of
the merged classification.
5. The method of claim 4, wherein the classifying of the lyrics,
the title, or the message according to the theme comprises:
selecting a feature to be used for the theme classification from
music files, stored in a database, or a collection of the message;
indexing one category using the selected feature; pre-processing
the music files stored in a personal mobile terminal and a message
inputted by the user; indexing the music files and the message
according to a result of the pre-processing; and allocating each
category to the indexed music files and the message based on a
result of the indexing of the category.
6. The method of claim 1, wherein the further selecting music files
from the selected music files according to a similarity between
content of the message and content of the music comprises: indexing
the content of the message and the content of the music files;
calculating a similarity between the indexed content of the message
and the indexed content of the music files; comparing the
calculated similarity with a threshold value; and selecting the
music files when the calculated similarity is greater than the
threshold value.
7. The method of claim 1, wherein the recommending of the selected
music files to the user comprises: arranging the music files
further selected according to the similarity; and providing a list
of the arranged music files.
8. A computer-readable storage medium storing a program for
implementing a music recommendation method with respect to a
message inputted by a user, for transmitting via a message service
in a personal communication terminal, the method comprising:
selecting music files according to a theme of the message, a theme
of a music, and a mood of the music, determining whether the number
of selected music files is greater than a predetermined number,
and, if the number of selected music files is greater than the
predetermined number, then further selecting music files from the
selected music files according to a similarity between content of
the message and content of the music; and recommending the selected
music files to the user.
9. A music recommendation system with respect to a message inputted
by a user, for transmitting via a message service in a personal
communication terminal, the system comprising: a music file
selection unit to select music files according to a theme of the
message, a theme of a music, and a mood of the music, filtering
unit to determine whether the number of selected music files is
greater than a predetermined number, and, if the number of selected
music files is greater than a predetermined number, then to further
select music files from the previously selected music files
according to a similarity between content of the message, and
content of the music; and a recommendation unit to recommend the
selected music files to the user.
10. The system of claim 9, wherein the music file selection unit
comprises: a selection unit to select target music files, classify
the target music files and the message according to a theme, select
at least one music file which corresponds to the theme of the
message according to a result of the theme classification, from
among the target music files, and generate a list including the at
least one music file.
11. The system of claim 10, wherein the music file selection unit
further comprises: a first filtering unit to classify each of music
files of the list according to a mood, and to eliminate at least
one music file from the list based on the mood.
12. The system of claim 10, wherein the music file selection unit
further comprises: a lyrics theme classification unit to classify
lyrics of the music files according to a theme; a title theme
classification unit to classify a title of the music files
according to a theme; a message theme classification unit to
classify the message according to a theme; a classification result
merging unit to merge a result of the classifications of the lyrics
with a result of the classifying of the title; and a music
selection unit to select the plurality of the music files based on
a music collection classified according to the theme and the
message classified according to the theme.
13. The system of claim 12, wherein each of the lyrics theme
classification unit, the title theme classification unit, and the
message theme classification unit comprises: a feature selection
unit to select one of a feature of the title, the lyrics, or the
message, which is used for the theme classification from a
database; a category index unit to index one of a category of the
title, the lyrics, or the message using the feature; a
pre-processing unit to pre-process one of the title, the lyrics, or
the message stored in the user personal communication terminal; an
index unit to index the music files and the message according to a
result of the pre-processing; and a category allocation unit to
allocate each category to the indexed music files and the message
based on a result of the indexing of the category.
14. The system of claim 9, wherein the filtering unit comprises: a
message index unit to index a content of a message inputted by the
user; a lyrics index unit to index a content of a title and lyrics
of the previously selected music files; a similarity calculation
unit to calculate a similarity between (1) the indexed content of
the title and lyrics, and (2) the indexed content of the message; a
comparison unit to compare the calculated similarity with a
threshold value; and a selection unit to select the music files
when the calculated similarity is greater than the threshold
value.
15. The system of claim 9, wherein the recommendation unit
comprises: a music arrangement unit to arrange the music files
further selected based on the similarity; and a music list
providing unit to provide a list of the arranged music files.
Description
CROSS-REFERENCE TO RELATED APPLICATION
[0001] This application is a continuation of U.S. Ser. No.
11/889,622, filed Aug. 15, 2007, the disclosure of which is
incorporated herein in its entirety by reference. This application
claims the benefit of Korean Patent Application No.
10-2006-0127171, filed on Dec. 13, 2006, in the Korean Intellectual
Property Office, the disclosure of which is incorporated herein by
reference.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The present invention relates to a music recommendation
method which can automatically recommend appropriate music when
transmitting a message service in a personal communication
terminal, and a system using the method. More particularly, the
present invention relates to a music recommendation method which
can select a music file according to a theme of the music, a mood
of the music, and a similarity between content of a message service
and content of the music, and automatically recommend a user the
selected music file.
[0004] 2. Description of Related Art
[0005] Currently, since a personal communication terminal such as a
mobile phone tends to provide various multimedia functions, there
is a tendency in a message service to also provide a multimedia
messaging service (MMS) that may attach to transmit a photo, music,
and a moving picture included in the message service. An amount of
use of the MMS will be rapidly spread since it is possible to
transmit a long e-mail or multimedia contents in the personal
communication terminal. However, various functions which enable a
user to easily use the MMS are needed since the user has an
aversion to use the MMS due to inconveniences of procedures for
transmitting the MMS and a user interface.
[0006] Currently, the personal communication terminal can transmit
the music file while transmitting a message of a message service
via the MMS, and can store various types of music files in a memory
where the music files are stored according to a capacity of the
memory is increase. However, a conventional communication terminal
has a problem in that, it takes a great amount of time and a great
amount of effort to search for an appropriate music file for the
message service among stored music files via the MMS. Accordingly,
when the user uses the MMS via the personal communication terminal,
a new method capable of easily searching for a music file for
attachment is needed.
[0007] In the conventional art, as an example of selecting
appropriate music for an e-mail, there is a method which can
automatically select music based on an impression of the music. In
the conventional method, character strings are detected from the
e-mail, the detected character strings are converted into an
impression value using a conversion table, an impression value
database of the music is compared with the impression value, and
consequently the appropriate music is selected. In this instance,
the impression value indicates a reference value which shows
emotions felt by a user when the user feels the music. The
impression value is analyzed from a physical feature of a music
signal, and there are impression values such as extreme,
liveliness, refresh, simplicity, tenderness. However, in the
conventional art, there is a problem in that, the appropriate music
is not accurately selected since the method exclusively relies on
the impressions of the music, i.e. the appropriate music with
respect to the e-mail is selected by exclusively using the
impressions of the music, accordingly there is a probability that a
selected music does not corresponds to the e-mail.
[0008] Also, a personal communication terminal using the
conventional art has a problem in that, a user is required to
navigate a plurality of selected music on a limited small screen,
and select appropriate music after checking the navigated music
when a great number of music having an identical impression value
exist.
[0009] Also, the personal communication terminal using the
convention art has a problem in that, a recommendation rank of a
plurality of selected music may not be rated since music having an
impression value, corresponding to an e-mail, is randomly
displayed.
BRIEF SUMMARY
[0010] An aspect of the present invention provides a music
recommendation method which can select a music file according to a
theme of the music, a mood of the music, and a similarity between
content of a message service and content of the music, and
automatically recommend to a user the selected music file, and a
music recommendation system using the method.
[0011] An aspect of the present invention also provides a music
recommendation method which can classify a title of music, lyrics
of the music, and a text of a message service according to a theme,
compare the classified theme, and select the music as a result of
the comparison, and a music recommendation system using the
method.
[0012] An aspect of the present invention also provides a music
recommendation method which can recommend music, which is matched
with a theme of a message service, by classifying music according
to a theme, and also classifying the music according to a mood in a
personal communication terminal, and a music recommendation system
using the method.
[0013] An aspect of the present invention also provides a music
recommendation method which can accurately select music, which is
matched with a message service, by calculating a similarity between
content of lyrics and content of the message service, and a music
recommendation system using the method.
[0014] According to an aspect of the present invention, there is
provided a music recommendation method in a personal communication
terminal, including: selecting music files according to a theme of
the message service and music, a mood of the music, a similarity
between content of the message service and content of the music;
and recommending selected music files to a user.
[0015] According to another aspect of the present invention, there
is provided a music recommendation system including: a music file
selection unit selecting music files according to a theme of the
message service and music, a mood of the music, a similarity
between content of the message service, and content of the music;
and a recommendation unit recommending selected music files to a
user.
[0016] Additional and/or other aspects and advantages of the
present invention will be set forth in part in the description
which follows and, in part, will be obvious from the description,
or may be learned by practice of the invention.
BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The above and/or other aspects and advantages of the present
invention will become apparent and more readily appreciated from
the following detailed description, taken in conjunction with the
accompanying drawings of which:
[0018] FIG. 1 is a diagram illustrating a music recommendation
system with respect to a message service according to an embodiment
of the present invention;
[0019] FIG. 2 is a diagram illustrating an example of a
configuration of a music file selection unit of FIG. 1;
[0020] FIG. 3 is a diagram illustrating an example of a
configuration of a selection unit of FIG. 2;
[0021] FIG. 4 is a diagram illustrating an example of a
configuration of a second filtering unit of FIG. 2;
[0022] FIG. 5 is a diagram illustrating an example of a
configuration of a theme based music selection unit of FIG. 3;
[0023] FIG. 6 is a diagram illustrating an example of a
configuration of a lyrics theme classification unit, a title theme
classification unit, and a message theme classification unit of
FIG. 5;
[0024] FIG. 7 is a diagram illustrating an example of a theme
matching table and a mood matching table according to an exemplary
embodiment of the present invention;
[0025] FIG. 8 is a flowchart illustrating a music recommendation
method with respect to a message service according to another
embodiment of the present invention;
[0026] FIG. 9 is a flowchart illustrating an example of operations
of analyzing themes of FIG. 8;
[0027] FIG. 10 is a flowchart illustrating operations of
classification of music files according to a theme of FIG. 9;
[0028] FIG. 11 is a flowchart illustrating an example of operations
of selecting music files according to the theme of FIG. 9;
[0029] FIG. 12 is a flowchart illustrating an example of filtering
of a plurality of music files according to a similarity of FIG. 9;
and
[0030] FIG. 13 is a flowchart illustrating an example of operations
of classification of a theme of lyrics, a theme of a title, and a
theme of a message service of FIG. 11.
DETAILED DESCRIPTION OF EMBODIMENTS
[0031] Reference will now be made in detail to exemplary
embodiments of the present invention, examples of which are
illustrated in the accompanying drawings, wherein like reference
numerals refer to the like elements throughout. The exemplary
embodiments are described below in order to explain the present
invention by referring to the figures.
[0032] FIG. 1 is a diagram illustrating a music recommendation
system 100 with respect to a message service according to an
embodiment of the present invention.
[0033] Referring to FIG. 1, the music recommendation system 100
includes a music file selection unit 110, a recommendation unit
120, and a music file collection 130.
[0034] A user inputs, via an input device of the music
recommendation system 100, a message service to transmit.
Accordingly, the music recommendation system 100 may receive the
message service via various input devices such as a keyboard, a
keypad, and a touchpad, and the like, (not shown) from the
user.
[0035] Also, when the user wants to transmit the message service by
attaching a music file, the user selects whether to use the music
file collection 130 stored in a user terminal, or to use a music
file which is downloadable via a music download service on an
Internet connection, as the music file to be attached to the
message service. Accordingly, the music recommendation system 100
may receive a selection from the user whether to use the music file
collection 130 or to use the downloadable music file.
[0036] The music file selection unit 110 selects a music file
according to a theme of the message service, a theme of the music,
a mood of the music, and a similarity between content of the
message service and content of the message service. Specifically,
the music file selection unit 110 classifies the message service
and the music file according to the theme of the message service,
the theme of the music, the mood of the music, and the similarity
between content of the message service and content of the message
service, and selects a music file according to a result of the
classification. Hereinafter, a configuration and operations of the
music file selection unit 110 will be described in detail by
referring to FIG. 2.
[0037] Referring to FIGS. 1 and 2, the music file selection unit
110 includes a selection unit 210, a first filtering unit 220, and
a second filtering unit 230.
[0038] The selection unit 210 selects target music files,
classifies the target music files and a message service according
to a theme, and selects, from among the target music, a music file
which corresponds to the theme of the message service according to
a result of the classification. Hereinafter, a configuration and
operations of the selection unit 210 will be described in detail by
referring to FIG. 3.
[0039] Referring to FIGS. 1 through 3, the selection unit 210
includes a target music selection unit 310 and a theme based music
selection unit 320 which selects music according to a theme.
[0040] The target music selection unit 310 selects either a music
file collection stored in a user terminal, or a music file which is
downloadable via a music download service on an Internet
connection, as the target music files. Specifically, the target
music selection unit 310 may select either the music file from
among music files in the user terminal, or the music file which is
downloadable using the music download service on an Internet
connection, as the target music files. The music recommendation
system 100 receives a selection for the music file as the target
music file, from the user via the target music selection unit
310.
[0041] The theme based music selection unit 320, selecting the
music according to the theme, classifies the target music files
according to the theme, and selects a plurality of music files
according to the classified theme from among the target music
files. Specifically, the theme based music selection unit 320
selects, from among the target music files, the plurality of music
files according to the classified themes of the target music files
and message service. As described, the music recommendation system
100 of FIG. 1 analyzes the theme of the target music files and the
theme of the message service via the theme based music selection
unit 320, and selects, from among the target music files, the
plurality of the music files appropriate for the analyzed themes of
the message service. Hereinafter, a configuration and operations of
the theme based music selection unit 320 will be described in
detail by referring to FIG. 5.
[0042] FIG. 5 is a diagram illustrating an example of a
configuration of the theme based music selection unit 320 of FIG.
3.
[0043] Referring to FIG. 5, the theme based music selection unit
320 of FIG. 3 includes a lyrics theme classification unit 510, a
title theme classification unit 520, a classification result
merging unit 530, a message theme classification unit 540, and a
music selection unit 550.
[0044] The lyrics theme classification unit 510 classifies lyrics
of music files according to a theme. The lyrics theme
classification unit 510 may omit operation of theme classification
with respect to the music files when there is no information of the
lyrics from the music files. The title theme classification unit
520 classifies titles of music files. Specifically, the title theme
classification unit 520 extracts the title of a music file from an
identification3 (ID3) tag of the music files or music file names of
the music files, and classifies the title of the music files using
the extracted title of the music files.
[0045] The classification results merging unit 530 merges a result
of the classifications of the lyrics with a result of the title of
music files, and outputs a theme of the music files.
[0046] The message theme classification unit 540 classifies a
message service according to a theme, and outputs the classified
theme of the message service.
[0047] The themes of the music files and the message service may be
variously defined depending on categories such as sorrow,
happiness, love, a breakup, yearning, spring, summer, autumn,
winter, and a journey. When a specific music does not fall into the
defined categories, the specific music may be classified into the
others.
[0048] The music selection unit 550 selects a music file based on a
the music files classified according to theme and the message
service classified according to the theme. Specifically, the music
selection unit 550 may select the music file which corresponds to
the classified theme of the message service.
[0049] FIG. 6 is a diagram illustrating an example of a
configuration of the lyrics theme classification unit 510, the
title theme classification unit 520, and the message theme
classification unit 540 of FIG. 5.
[0050] Referring to FIG. 6, the lyrics theme classification unit
510, the title theme classification unit 520, and the message theme
classification unit 540 includes a theme classification learning
unit 610 and a theme classification unit 620.
[0051] The theme classification learning unit 520 performs learning
for theme classification of music files stored in the database 330
of FIG. 3 and a message service, and includes a feature selection
unit 611 and a category index unit 612. The database 330 may be
previously built for the theme classification learning, and themes
corresponding to the title of music, lyrics of the music and a
message service are classified in the database 330. The music
recommendation system 100 of FIG. 1 may use a lyrics database when
it is difficult to build a database with respect to the
message.
[0052] The feature selection unit 611 extracts feature candidates
from the title of the music, the lyrics of the music stored in the
database 330, the message service, and selects one feature to be
used for the theme classification using the extracted feature
candidates. The feature candidates may include a morpheme n-gram, a
word n-gram, and a syllable n-gram. When a processing capability of
the feature selection unit 611 is sufficient, the feature selection
unit 611 uses the morpheme n-gram as the feature candidates, and
when the processing capability of the feature selection unit 611 is
insufficient for using the morpheme n-gram, the feature selection
unit 611 may use the word n-gram or the syllable n-gram. The
feature selection unit 611 may select the feature from the feature
candidates using a mutual information scale, an information
acquisition quantity, and Chi-square statistic. In the message
service, an emoticon has important information, and is used for the
feature since the emoticon is used for expressing emotions of a
user. Accordingly, the emoticon is additionally collected to be
used for the feature.
[0053] The category index unit 612 indexes a category of each theme
using the selected feature. Namely, the category index unit 612 may
express the category of each theme as a category vector including
features and values. Each theme category vector has all of the
selected features as the feature, and a feature of the category
vector for each theme has `1` as a feature value when a
corresponding feature is selected, the feature of the category
vector for each theme has `0` as the feature value when the
corresponding feature is not selected. The category index unit 612
outputs a result of the category index as the category vector.
[0054] The theme classification unit 620 classifies the theme of
music files stored in the user terminal, or provided from a web
server, and the theme of the message service inputted by the user,
and includes a pre-processing unit 621, an index unit 622, and a
category allocation unit 623.
[0055] The pre-processing unit 621 performs pre-processing on the
music files and the message service in order to classify the music
and the message service according to the theme. Namely, the
pre-processing unit 621 extracts a title and lyrics from the music
files, and acquires text information or emoticon information from
the music files and the message service. The title may be extracted
from an ID3 tag of the music files or the music file name.
[0056] The index unit 622 expresses the title, the lyrics of the
music files, or the message service as a vector to index the
expressed vector. Namely, the index unit 622 determines whether
each feature of the selected feature is included in the title, the
lyrics of the music files, or the message service, and indexes a
value according to a result of the determination. Since words,
directly associated with the theme of the music, are compressively
shown in the title, the index unit 622 may allocate to index `1`
when the feature is shown in the title, and may allocate to index
`0` when the feature is not shown in the title after applying a
binary weight to each of the feature values. The index unit 622 may
allocate to index a frequency number, i.e. how many times a
corresponding word occurs in the text, to the feature value by
applying a frequency weight to the lyrics and the message service.
The index unit 622 may output a result of the index as a
vector.
[0057] The category allocation unit 623 determines the theme of the
title, the lyric, and the message service using the category
vector, the music vector, or the message vector, and allocates each
category which corresponds to the determined theme. Specifically,
the category allocation unit 623 determines the theme by measuring
a similarity between the theme category vector, obtained by the
category index unit 612, and the title, the lyrics, and the message
service vectors, and allocates the each category which corresponds
to the determined theme. As an example, a vector dot product
(abcos.theta.) or a cosine similarity may be used for the vector
similarity.
[0058] The first filtering unit 220 of FIG. 2 analyzes the
plurality of music files according to a mood, and filters out the
plurality of music files, selected by the selection unit 210 of
FIG. 2, based on the analyzed mood. Specifically, the first
filtering unit 220 deletes music files whose theme of the music
files and mood of the music files are not matched with each other.
A matching relation between the mood of the music files and the
theme of the music files may be understood by referring to a
matching table in FIG. 7.
[0059] FIG. 7 is a diagram illustrating an example of a theme
matching table and a mood matching table according to an exemplary
embodiment of the present invention.
[0060] Referring to FIG. 7, a theme of a message service and a mood
of a music file are matched with each other in the matching table.
The theme of the message service includes happiness, sorrow, a
journey, a yearning, etc., the mood of the music file includes
pleasant, sad, calm, exciting, etc., and various types of
classification may be used. The matching table may be stored in the
database 330 of FIG. 3. There is a probability that a theme of
music and a mood of the music do not match each other, from among
the music files which correspond to the theme of the message
service. As an example, when the theme of the music is a `breakup`,
and the mood of the music is `pleasant`, this indicates the theme
of the music and the mood of the music are not matched with each
other.
[0061] Accordingly, the music recommendation system 100 of FIG. 1
deletes the music file whose mood of music does not match a theme
of music. As an example, when a theme of the message service is
`breakup` while selecting the music, the music recommendation
system 100 deletes a music file having an inappropriate mood for
the theme of the music using the mood of the music, from among the
selected music files, in order to prevent selecting a music file
whose mood is pleasant, and selects a music file having an
appropriate mood for the theme of the music. In this case, the
music recommendation system 100 uses the mapping table in order to
filter out the selected music file.
[0062] As an example, when a theme of the music file or a theme of
the message service is `happiness`, and a mood of music
corresponding to the theme is `pleasant`, a first filtering unit
220 of FIG. 2 may select music files whose mood of the music is
`pleasant` by filtering out music files from a plurality of music
files whose mood of the music is not `pleasant`.
[0063] As an another example, when a theme of the music file or a
theme of the message service is `sorrow`, a mood of music
corresponding to the theme is `sorrow` or `calm`, a first filtering
unit 220 may select music files whose mood of the music is `sorrow`
by filtering out music files whose mood of the music is not
`sorrow` or `calm` from the plurality of music files.
[0064] As still another example, when a theme of the music file or
a theme of the message service is `journey`, a mood of music
corresponding to the theme is `exciting`, a first filtering unit
220 may select music files whose mood of the music is `exciting` by
filtering out music files whose mood of the music is not `exciting`
of the plurality from music files.
[0065] As yet another example, when a theme of the music file or a
theme of the message service is `yearning`, a mood of music
corresponding to the theme is `calm`, a first filtering unit 220
may select music files whose mood of the music is `calm` by
filtering out music files whose mood of the music is not `calm`
from the plurality of music files.
[0066] As described above, the music recommendation system 100 may
provide a user with an appropriate number of music files when the
appropriate number of music files is selected by the first
filtering unit 220. However, when a number of a selected music
files is great, there is a probability that there are music having
an identical theme or mood, or the user wants to show something
different via a message service even if music has an identical
theme. Accordingly, the music recommendation system 100 according
to the present invention measures a similarity between content of
lyrics of music and content of a message service via a second
filtering unit 230 of FIG. 2, and filters out the music according
to the similarity.
[0067] The second filtering unit 230 calculates the similarity
between the content of the message and the content of the music
file, and filters out the music files according to the
similarity.
[0068] The second filtering unit 230 may be omitted in order to
increase a speed of the music recommendation system 100, or the
second filtering unit 230 may be omitted when the number of music
files selected by the first filtering unit 220 is relatively few.
Hereinafter, a configuration and operations of the second filtering
unit 230 will be described in detail by referring to FIG. 4.
[0069] FIG. 4 is a diagram illustrating an example of a
configuration of a second filtering unit 230 of FIG. 2.
[0070] Referring to FIG. 4, the second filtering unit 230 includes
a message index unit 410, a lyrics index unit 420, a similarity
calculation unit 430, a comparison unit 440, and a selection unit
450.
[0071] The message index unit 410 indexes a message service
inputted by a user. Specifically, the message index unit 410
extracts character strings of a text from the message service
inputted by the user, and indexes the message service by expressing
the character strings of the text as a vector. A morpheme n-gram
and a word n-gram may be used for the character strings of the
text. The message index unit 410 may reduce a number of features by
selecting the morpheme n-gram or the word n-gram, including
morphemes having a substantial meaning such as a noun and an
inflected word. The feature, having been used in the index unit
622, may be used for the feature of the vector.
[0072] The lyrics index unit 420 indexes the music files by
expressing the lyrics and the title of the music files as a vector.
In this case, features, which are indexed by the message index unit
410, may be used a feature for the vector. A feature value of the
vector of the message service and features of the vectors of the
lyrics and the title may be indexed by calculating a number of
features occurring in the message service or the lyrics using a
frequency weight.
[0073] The similarity calculation unit 430 calculates a similarity
between content of the music files and content of the message
service. Specifically, the similarity calculation unit 430
calculates the similarity between the content of the music files
and the content of the message service using a vector of the
indexed music files and a vector of the indexed message service. A
cosine similarity may be used for the similarity between the
content of the music files and the content of the message
service.
[0074] The comparison unit 440 compares the similarity with a
threshold value. The threshold value is defined via a predetermined
experiment. Specifically, the comparison unit 440 compares the
similarity with the threshold value to determine whether the
similarity between the content of the music files and the content
of the message service is greater than or equal to the threshold
value, or less than the threshold value. When the similarity is
less than the threshold value, operation process returns to the
message index unit 410.
[0075] The selection unit 450 selects, from among the indexed music
files, the music files when a similarity of a specific music file
is greater than the threshold value. Specifically, the selection
unit 450 may select music files whose similarity between content of
music files and content of message service is greater than or equal
to the threshold value.
[0076] The recommendation unit 120 of FIG. 1 recommends a music
file corresponding to the theme of the message service, based on a
result of the analysis. Specifically, the recommendation unit 120
may recommend the user a music list whose selected music files are
arranged in a descending order according to a similarity. The
recommendation unit 120 may include a music arrangement unit and a
music list providing unit.
[0077] The music arrangement unit arranges the selected music files
based on the similarity. The music arrangement unit may arrange the
selected music files using a category, having been calculated by
the category allocation unit 623 of FIG. 6, and the similarity
between music themes. Specifically, the music arrangement unit may
arrange the selected music files in an order similarity between the
music themes and a theme category vector corresponding to the theme
of the message service. Also, the music arrangement unit may
arrange the selected music files based on the content similarity,
having been calculated by the similarity calculation unit 430 of
the second filtering unit 230. Specifically, the music arrangement
unit may select music files in an order similarity with respect to
the theme and the content of the indexed music files and the
message service. The music arrangement unit may not operate when a
number of the indexed music files is insufficient to arrange the
music files, or the music arrangement unit may not operate by
considering a speed of selecting the music files.
[0078] The music list providing unit provides a music list of the
arranged music files. The music list provides information of music
files with respect to the arranged music files, and the information
of the music files may include information of the title, lyrics, a
singer of the music. The user may check the information of the
title, lyrics, the singer of the music on the music list, and may
select required music or may select after listening to the
music.
[0079] FIG. 8 is a flowchart illustrating a music recommendation
method with respect to a message service according to another
embodiment of the present invention.
[0080] Referring to FIG. 8, in operation 810, a music
recommendation system selects music files according to a theme of
the message service and music, a mood of the music, a similarity
between content of the message service and content of the music.
Herein after, the selecting of the music files will be described in
detail by referring to FIG. 9.
[0081] FIG. 9 is a flowchart illustrating an example of operations
of analyzing themes of FIG. 8.
[0082] Referring to FIG. 9, in operation 910, a music
recommendation system selects target music files. Specifically, in
operation 910, the music recommendation system selects either a
music file stored in a user terminal or a music file which is
downloadable via a music download service on an Internet
connection, as the target music files.
[0083] In operation 920, the music recommendation system classifies
the target music files and the message service according to a
theme, and selects, from among the target music files, a plurality
of music files which corresponds to the theme of the message
according to a result of the classification. Hereinafter, the
classification of the target music files will be described in
detail by referring to FIG. 10.
[0084] FIG. 10 is a flowchart illustrating operations of the
classifying of the music files according to the theme of FIG.
9.
[0085] Referring to FIG. 10, in operation 1011, a music
recommendation system classifies a title of music of target music
files according to a theme.
[0086] In operation 1012, the music recommendation system
classifies lyrics of the music of the target music files.
[0087] In operation 1020, the music recommendation system merges a
result of the classification of the lyrics with a result of the
classification of the title.
[0088] In operation 1030, the music recommendation system
classifies the target music files according to a theme, based on
the merged results of the classification of the lyrics and the
title. A theme of the music files may be variously defined
depending on categories such as sorrow, happiness, love, a breakup,
yearning, spring, summer, autumn, winter, and a journey. When a
specific music does not fall into the defined categories, the
specific music may be classified into the others. Hereinafter, the
selecting of the target music files will be described in detail by
referring to FIG. 11.
[0089] FIG. 11 is a flowchart illustrating an example of operations
of selecting a music file according to the theme of FIG. 9.
[0090] Referring to FIG. 11, in operation 1110, a music
recommendation system classifies a message service according to a
theme. Similar to the theme of the target music files, the theme of
the message service may be variously defined depending on
categories such as sorrow, happiness, love, a breakup, yearning,
spring, summer, autumn, winter, and a journey. When a specific
music does not correspond to the defined categories, the specific
music may be classified into the others.
[0091] In operation 1120, the music recommendation system selects
music files appropriate for the theme of the message service based
on the classified theme of the music target music files.
Specifically, in operation 1120, the music recommendation system
may select a plurality of music files corresponding to the theme of
the classified message service, from a collection of the classified
target music files.
[0092] In operation 930, the music recommendation system classifies
the plurality of music files according to a mood, and filters out a
music file whose mood is inappropriate for the theme of the music
files.
[0093] In operation 940, the music recommendation system filters
the plurality of the music files according to a similarity between
content of the message service and content of the music files.
Hereinafter, the filtering of the plurality of the music files
according to the similarity between the content of the message
service and the content of the music files will be described in
detail by referring to FIG. 12.
[0094] FIG. 12 is a flowchart illustrating an example of the
filtering of the plurality of music files according to the
similarity of FIG. 9.
[0095] Referring to FIG. 12, in operation 1210, a music
recommendation system indexes a title and lyrics of music files,
stored in a database, and content of a message service.
Specifically, in operation 1210, the music recommendation system
extracts character strings of a text from the message service
inputted by the user, and indexes the message service by expressing
the character strings of the text as a vector. In this case, a
morpheme n-gram and a word n-gram may be used for the character
strings of the text, and features to be selected in operation 1310
may be used for the character strings of the text. Feature values
may be indexed by calculating a number of each features' occurring
in the message service or the lyrics using a frequency weight.
[0096] In operation 1220, the music recommendation system
calculates a similarity content of indexed music files and content
of indexed message service. Specifically, in operation 1220, the
music recommendation system calculates the similarity between the
content of the music files and the content of message service using
a vector of the content of the indexed music files and a vector of
the content of the indexed message service. In this case, a cosine
similarity may be used for the similarity between the content of
the music files and the content of the message service.
[0097] In operation 1230, the music recommendation system compares
the similarity with a threshold value. Specifically, the music
recommendation system compares the similarity with the threshold
value to determine whether the similarity between the content of
the music files and the content of the message service is greater
than or equal to the threshold value, or less than the threshold
value.
[0098] In operation 1240, the music recommendation system
determines whether the similarity is greater than the threshold
value. When the similarity is less than the threshold value,
operation process returns to operation 1210.
[0099] In operation 1250, the music recommendation system selects
music files whose similarity is greater than the threshold value.
Specifically, in operation 1250, the music recommendation system
may select the music files whose similarity between the content of
the music files and the content of the message service is equal to
or greater than the threshold value.
[0100] FIG. 13 is a flowchart illustrating an example of operations
of classification of a theme of lyrics, a theme of a title, and a
theme of a message service of FIG. 11.
[0101] Referring to FIG. 13, in operation 1310, a music
recommendation system selects a feature via learning, from music
files stored in a database or a collection of a message service.
Specifically, in operation 1310, the music recommendation system
may select the feature to be used for theme classification, from a
title and lyrics of the music files stored in the database or the
message service.
[0102] In operation 1320, the music recommendation system indexes
one of a category of the music files or the message service using
the selected feature.
[0103] In operation 1330, the music recommendation system
pre-processes the music files stored in a user terminal and the
message service inputted by a user.
[0104] In operation 1340, the music recommendation system indexes a
music file and a message service according to a result of the
pre-processing.
[0105] In operation 1350, the music recommendation system allocates
each category to the indexed music files and the message service
based on a result of the category index.
[0106] Operations 1310 and 1320 are performed using a music file
collection stored in the database, the database being previously
built, and the message service. Also, operations 1310 and 1320 may
be performed in advance when building a system, not at a point of
when a user uses the system, since operations 1310 and 1320
correspond to operation of the learning. Operations 1330, 1340, and
1350 are performed during operation of classification of music and
a message, and a target of operations 1330, 1340, and 1350 is a
music file in a user terminal or a message service to be
transmitted by the user.
[0107] In operation 820, the music recommendation system recommends
appropriate music files for the message service to the user based
on the themes calculated in operation 810 and a similarity value
between the contents of the music files and the message service.
Specifically, in operation 820, the music recommendation system
arranges the music files based on themes calculated in operation
810 and the similarity value between the contents of the music
files and the message service to recommend the arranged music files
to the user.
[0108] As an example, in operation 820, the music recommendation
system may recommend to the user a music list whose selected music
files are arranged in a descending order by using the similarity
value. Specifically, in operation 820, the music recommendation
system may include the arranging of the selected music files, and
the providing of the music list of the arranged selected music
files. The music list may include a title, lyrics, and a singer of
music. The user may check information of the title, the lyrics, and
the singer of the music on the music list, and select required
music or select after listening to the music.
[0109] The music recommendation method according to the
above-described embodiment of the present invention may be recorded
in computer-readable media including program instructions to
implement various operations embodied by a computer. The media may
also include, alone or in combination with the program
instructions, data files, data structures, and the like. Examples
of computer-readable media include magnetic media such as hard
disks, floppy disks, and magnetic tape; optical media such as CD
ROM disks and DVD; magneto-optical media such as optical disks; and
hardware devices that are specially configured to store and perform
program instructions, such as read-only memory (ROM), random access
memory (RAM), flash memory, and the like. The media may also be a
transmission medium such as optical or metallic lines, wave guides,
and the like, including a carrier wave transmitting signals
specifying the program instructions, data structures, and the like.
Examples of program instructions include both machine code, such as
produced by a compiler, and files containing higher level code that
may be executed by the computer using an interpreter. The described
hardware devices may be configured to act as one or more software
modules in order to perform the operations of the above-described
embodiments of the present invention.
[0110] According to the present invention, there are provided a
music recommendation method which can select a music file according
to a theme of the music, a mood of the music, and a similarity
between content of a message service and content of the music, and
automatically recommend to a user the selected music file, and a
music recommendation system using the method.
[0111] Also, according to the present invention, there are provided
a music recommendation method which can classify a title of music,
lyrics of the music, and a text of a message service according to a
theme, compare the classified theme, and select the music as a
result of the comparison, and a music recommendation system using
the method.
[0112] Also, according to the present invention, there are provided
a music recommendation method which can recommend music, which is
matched with a theme of a message service, by classifying music
according to a theme, and also classifying the music according to a
mood in a personal communication terminal, and a music
recommendation system using the method.
[0113] Also, according to the present invention, there are provided
a music recommendation method which can accurately select music,
which is matched with a message service, by calculating a
similarity between content of lyrics and content of the message
service, and a music recommendation system using the method.
[0114] Although a few exemplary embodiments of the present
invention have been shown and described, the present invention is
not limited to the described exemplary embodiments. Instead, it
would be appreciated by those skilled in the art that changes may
be made to these exemplary embodiments without departing from the
principles and spirit of the invention, the scope of which is
defined by the claims and their equivalents.
* * * * *