KR20160071039A - Method for generating map based on musical preferences - Google Patents

Abstract

The present invention relates to a method for generating a map based on musical preferences, wherein regions are mapped onto the musical preferences. The method for generating a map based on musical preferences according to an embodiment of the present invention includes the steps of: collecting music playback information including artist information and regional information of played music; extracting a tag list including a tag about an artist included in the music playback information according to regions from a tag information database; calculating a preferred music association between a first region and a second region using the extracted tag list; and generating a map using the calculated preferred music association.

Description

METHOD FOR GENERATING MAP BASED ON MUSICAL PREFERENCES [0002]

The present invention relates to a map generation method, and more particularly, to a music preference-based map generation method in which regions are mapped according to music preferences.

There are differences in the types of music preferred by each country. For example, the popularity of music such as pop, k-pop, and reggae depends on the country. Music-related applications use local information to search for or recommend music.

However, geographic proximity is not always proportional to the preference of music. Because the culture, language, and other styles differ greatly depending on the boundaries of the country, there is a difference in the preference of music between adjacent regions. If there are cultural commonalities even in remote areas, the preferences of music may be similar. A related prior art document is Korean Patent No. 10-1054904.

To alleviate the difference between geographic proximity and music preferences, a new map can be created in which the country (or city) is located according to the preference of the music. Countries (cities) distributed around the world may be located adjacently or remotely, depending on how similar their preference music is. Information according to music preference by region can be used for music search and marketing. Therefore, it is necessary to study how to generate map based on music preference.

It is an object of the present invention to provide a music preference-based map generation method of generating a new map based on regional music preferences.

According to an embodiment of the present invention, there is provided a method of reproducing music information, the method comprising: collecting music reproduction information including artist information and local information of reproduction music; Extracting, from a tag information database, a tag list including a tag for an artist included in the music reproduction information for each region; Calculating a preference music association between the first region and the second region using the extracted tag list; And generating a map using the calculated preferred music association map.

The music preference-based map generation method according to an embodiment of the present invention can generate a new map based on a similarity relationship between music preferences of the region.

According to an embodiment of the present invention, it is possible to guess which countries have similar music preferences and which regions have unique music preferences through a music preference-based map.

According to an embodiment of the present invention, a music preference-based map can be utilized for searching for new music or marketing of a music.

1 is a block diagram of a map generating device in accordance with an embodiment of the present invention.
FIG. 2 is a flowchart illustrating a music preference-based map generation method according to an embodiment of the present invention.
FIG. 3 is a flowchart illustrating a data collection step of the music preference-based map generation method of FIG.
FIG. 4 is a diagram illustrating distribution of music reproduction information collected by regions according to an embodiment of the present invention.
Figure 5 shows a preferred music association matrix associated with an embodiment of the present invention.
Figure 6 shows an example where regions are mapped onto a two-dimensional space according to an iterative operation associated with an embodiment of the present invention.
Figure 7 shows a Gaussian mixer model associated with an embodiment of the present invention.
Figures 8-9 illustrate a music preference based map in accordance with one embodiment of the present invention.
Fig. 10 shows a map of the form in which the top tag is displayed in the map of Fig.

Hereinafter, a music preference-based map generation method according to an embodiment of the present invention will be described with reference to the drawings.

As used herein, the singular forms "a", "an" and "the" include plural referents unless the context clearly dictates otherwise. In this specification, the terms "comprising ", or" comprising "and the like should not be construed as necessarily including the various elements or steps described in the specification, Or may be further comprised of additional components or steps.

1 is a block diagram of a map generating device in accordance with an embodiment of the present invention.

The map generating apparatus 100 may include a receiving unit 110, a collecting unit 120, an association calculating unit 130, a map generating unit 140, and a controller 150.

The receiving unit 110 can receive the novel network data from various web servers (not shown). The term "social network data" refers to data used or created by users of social network services such as Twitter and Facebook.

The collection unit 120 may analyze only the received social network data and collect only music playback information. The music reproduction information may include a music reproduction history and area information on which the music is reproduced. The collecting unit 120 may extract a tag list for each artist included in the music reproduction information from a tag information database (not shown).

The association calculation unit 130 can calculate the preference music association degree between regions using the extracted tag list of each region. The preference music association between the regions is an index indicating how similar is the preference music between regions.

The map generation unit 140 can generate a map based on music preference by converting the inter-region favorite music association map into distance.

The control unit 150 may control the receiving unit 110, the collecting unit 120, the association calculating unit 130, and the map generating unit 140 as a whole.

FIG. 2 is a flowchart illustrating a music preference-based map generation method according to an embodiment of the present invention.

In order to express the music preference by region, information of many pieces of music reproduced by each region is necessary. The collection unit 120 may analyze the social network data received by the reception unit 110 in order to collect music reproduction information (S210).

FIG. 3 is a flowchart illustrating a data collection step of the music preference-based map generation method of FIG.

The collecting unit 120 may extract only music reproduction information from a large number of social network data (e.g., a Twitter message) (S310). Hereinafter, a message of the tweeter will be described as an example of social network data in the embodiment.

In social network services (such as Twitter), many users upload the music they are listening to as textual information. Although the Twitter message has the form of unstructured text data, it includes a hash tag (Hashtag) which is a user spontaneous classification system, so that only the tweets including the music reproduction contents can be filtered and collected. For this purpose, a hash tag used when a user uploads music information being played can be defined in advance.

A hash tag is a function that expresses a word about a specific word in the form of a '# specific word' on Twitter, a social network service (SNS). By putting a specific word after the '#', it can be expressed as a text on the subject. For example, if the text to be input is a text for music, the user can input '# MUSIC'. Hashtags are a feature introduced for the convenience of searching, but they are also used as a way or means to reveal interest and support for a specific topic.

The music reproduction information may include an artist name, a song title, a place where the reproduction music is disclosed, and the like. In the present specification, an artist may include a singer, a composer, a lyricist, and the like as a person who creates or expresses music. Hereinafter, in the embodiment, the artist will be described as an example of an artist.

A regular expression matching technique may be applied to extract music reproduction information (song title, artist name, etc.) from unstructured text data (social network data) collected in real time. For example, in the tweet message, music information usually exists in four text patterns (song title-singer, song title by singer, song title / singer, and song title-singer) It is judged as regular expression matching and data is extracted.

On the other hand, since the collected tweet message contains the place information of the uploaded message, the collected data can be classified according to the region.

FIG. 4 is a diagram illustrating distribution of music reproduction information collected by regions according to an embodiment of the present invention. Figure 4 shows the collection distribution of music related tweets from around the world.

The collection unit 120 may define group data for the area and the area using the collected area information (e.g., GPS information) (S320). The collection unit 120 groups by city name or country name so that the collected local information (e.g., GPS information) can be converted into the corresponding country or city name.

The collecting unit 120 may define group data for each region (city or country). The collecting unit 120 may extract a group data set for each region (a list of artists and a reproduction count of songs of singers) (S330).

The number of reproductions of the singers 'list and the singers' songs can be expressed by the following equations (1) and (2).

A _r is the definition of the singers list by region through the twitter playback information, and r represents the region. a is the artist (artist) name.

Equation (2) is the reproduction count number corresponding to the mantissa of Equation (1). That is, it can be seen from the collected twitter reproduction information data that the song of the singer in region r is reproduced a1 times ac1 times. In Equations (1) and (2), subscript n means the number of mantissa mentioned.

The collecting unit 120 may collect the tag list of the singer using the artist name and extract tag statistics (S340). The tag may be distinguished by a specific criterion (e.g. genre, mood, etc.).

The tag list (T _r ) can be expressed by the following equation (3).

The collection unit 120 may collect the tag list from the tag list database (e.g., Last.fm) for the singers collected in Equation (11). The tag t representing the region r can be defined through the union of these tags. Here refers to a tag of getTopTags (a _i) is a top predetermined number of (e. G., The top 10) on the basis of a singer ai of the tag tag number. m is the number of tags collected in region r.

The collection unit 120 can calculate the RTC (Regional Tag Count) using the following equation (4).

getTagCount is the number of times tag t was tagged for artist a. By multiplying the getTagcount of the mantissa a by the number of times reproduced in the region r, the number of times the tag t can be computed in the region r. For example, when t = 'rock', we compute the number of tags tagged with 'rock', and multiply the number of reproductions of each singing song by the number of rocks produced by the singers, .

Equation (5) is the RTC for the tag set, not the individual tag.

The association calculating unit 130 may calculate an inter-region favorite music association index (S220), which is an index indicating how similar regions are similar using the extracted tag statistics (e.g., RTC).

The preference music associations of the two regions r1 and r2 can be expressed through Tag Similarity Measurement (TSM) between the two regions, which can be calculated using Equation (6) and Equation (7) .

Here, RTC (r1, Tu) xTC (r2, Tu) denotes the inner product of two vectors.

The preference music association between all possible pairs of regions among regions defined using the TSM can be expressed as a matrix.

Figure 5 shows a preferred music association matrix associated with an embodiment of the present invention.

For example, if the defined region is m, a TSM matrix of m x m can be generated. Figure 5 shows the TSM matrix for 34 countries.

The map generating unit 140 may generate a map based on the music preference using the TSM indicating the preference music association (S230).

The map generation unit 140 may place the TSM matrix on the two-dimensional plane. The map generation unit 140 can convert TSM (a value between 0 and 1) into distances between regions. Equation (8) is a formula for converting the TSM into a tag distance (TD) between two regions.

The map generating unit 140 randomly places the regions ri to be mapped into the two-dimensional space. The map generation unit 140 selects one region r _s at random among the located regions. Next, the map generation unit 140 calculates tag-based distances between r _s and all other regions.

And Euclidean distances between two regions on the two-dimensional plane are calculated as shown in Equation (9).

The map generating unit 140 may perform a process of modifying the location of the area so that the Euclidean distance ED and the tag distance TD are equal to each other. This is to generate a map that takes the tagged distance (TD) as much as possible.

Equations (10) and (11) are equations for correcting the position of the region so that the Euclidean distance (ED) on the map and the tag distance (TD) coincide.

Modifications can be performed on the x-axis and the y-axis, since the space in which the regions are mapped is two-dimensional. This modification can be performed for all regions r _i except for the selected region r _s .

The distances on the two-dimensional plane and the TSM may be inversely proportional to each other through the iterative calculation of Equation (10) and Equation (11). As a result of the iterative operation of about 10,000 times, regions are mapped (mapped) to a new two-dimensional plane according to the preference relation.

FIG. 6 shows an example where regions are mapped on a two-dimensional space according to the iterative operation of Equation 10 and Equation 11 relating to an embodiment of the present invention.

Here, the learning rate can be defined as Equation (12) so that the width of correction becomes smaller as the iterative operation is performed more and more.

λ ₀ is the initial learning rate value.

A process of calculating a contour such as a contour line is required to make a map through a distribution of points representing an area mapped on a two-dimensional plane through the above-described operation. In this embodiment, a GMM (Gaussian Mixture Model) is used. GMM is a probabilistic modeling technique that creates a terrain (a three-dimensional value) with a high degree of clustering of regions and a low level of clustering of regions. The GMM uses the following parameters to create the terrain.

FIG. 7 shows an example of representing a region using a Gaussian mixer model related to an embodiment of the present invention.

는 지역의 위치이다.

는 지역의 봉우리 넓이가 된다.

는 지역의 봉우리 높이인데, 인접한 지역nn(r_i)이 많을수록 높이가 작아진다.As shown in FIG. 7, using the Gaussian mixer model, contours with multiple peaks are generated. These peaks are the result of the summation of several peaks (probability distributions) according to their adjacency. A single peak (probability distribution) corresponds to a Gaussian of a region, for a region i (eg Korea)

Is the location of the area.

Is the area of the peak of the area.

Is the height of the area, and the height of the neighboring area nn (r _i ) decreases.

That is, a Gaussian mixture model can be used to represent the mapped two-dimensional plane distribution as a map-like drawing. Through this, the two - dimensional distribution is represented by a three - dimensional probability density function, and thresholding is performed according to the probability value, so that land and sea can be distinguished.

FIG. 8 shows a music preference-based map of 34 countries created using the Gaussian mixture model. 9 also shows a music preference-based map of 249 countries where the region of the reproduced music is expanded.

According to an embodiment of the present invention, the map generating unit 140 generates a tag identifier (for example, an identifier for a predetermined number of tags on the basis of the number of tags representing the region) Can be displayed.

Fig. 10 shows a map of the form in which the top tag is displayed in the map of Fig.

10, it is possible to guess which countries have similar music preferences and which regions have unique music preferences.

As described above, the music preference-based map generation method according to an embodiment of the present invention can generate a new map based on a similarity relationship between music preferences of the region.

According to an embodiment of the present invention, it is possible to guess which countries have similar music preferences and which regions have unique music preferences through a music preference-based map.

According to an embodiment of the present invention, a music preference-based map can be utilized for searching for new music or marketing of a music.

The above-described music preference-based map generation method may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable recording medium. At this time, the computer-readable recording medium may include program commands, data files, data structures, and the like, alone or in combination. On the other hand, the program instructions recorded on the recording medium may be those specially designed and configured for the present invention or may be available to those skilled in the art of computer software.

The computer-readable recording medium includes a magnetic recording medium such as a magnetic medium such as a hard disk, a floppy disk and a magnetic tape, an optical medium such as a CD-ROM and a DVD, a magnetic disk such as a floppy disk, A magneto-optical media, and a hardware device specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like.

The recording medium may be a transmission medium such as a light or metal line, a wave guide, or the like including a carrier wave for transmitting a signal designating a program command, a data structure, and the like.

The program instructions also include machine language code, such as those generated by the compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

The above-described music preference-based map generation method and apparatus can be applied to all or some of the embodiments so that various modifications may be made to the embodiments and methods of the embodiments described above. Or may be selectively combined.

100: map generating device
110:
120: collecting section
130: association calculation unit
140: map generating unit
150:

Claims (5)

Collecting music reproduction information including artist information and local information of the reproduction music;
Extracting, from a tag information database, a tag list including a tag for an artist included in the music reproduction information for each region;
Calculating a preference music association between the first region and the second region using the extracted tag list; And
And generating a map using the calculated preferred music association map. 2. The method of claim 1,
And calculating a frequency of appearance of the tag in each region. 3. The method of claim 2, wherein the step
Further comprising using a tag list commonly used in the first region and the second region. 4. The method of claim 3, wherein the occurrence frequency of the tag in each region is
Wherein the music preference-based map generation step is performed using the number of music reproduction times of the artist and the number of times the music of the artist is used in the tag. 2. The method according to claim 1,
And converting the calculated preferred music associations into a distance between the first region and the second region.

Images