Teaching:TUW - UE InfoVis WS 2005/06 - Gruppe 10 - Aufgabe 3 - Design

Application Area and given Dataset

Application Area Analysis

The application area for this task is to visualize a music archive with the data provided by ID3/iTunes tags. These tags include information about songname, artist, year, album,... The challenge in visualizing this information is to select "good" attributes that can be brought into relation, because of the high number of discreet and "organizational" attributes.

Dataset Analysis

Attribute	Data type	Description
Name	Discreet	Song title
Artist	Discreet	Artist name
Album	Discreet	Album name
Genre	Nominal	Genre the song belongs to
Composer	Discreet	The composer of the song
Size	Ordinal	The file size
Total time	Ordinal	The total time of the song
Disc number	Ordinal	Number of the disc
Disc count	Ordinal	Total number of discs
Track number	Ordinal	Track number of this song on the disc
Track count	Ordinal	Total number of tracks on the disc
Year	Ordinal	Year of origin
Date Modified	Ordinal	Date of modification
Date added	Ordinal	Date when song was added to archive
Bit rate	Ordinal	Bit rate of song (e.g. 128kbit/s)
Sample rate	Ordinal	Sample rate of song (e.g. 44100Hz)
Play count	Ordinal	Number of time the song was played
Play date	Ordinal	Date of last play
Play date UTC	Ordinal	Date of last play in UTC
Location	Discreet	File location
File Folder Count	Ordinal	Number of file in the same folder as the current song
Library folder count	Ordinal	Number of files in library folder
Kind	Nominal	Kind of file (e.g. MPEG audio file)
Rating	Ordinal	Personal rating (1-5 stars)

One data record consists of a concatenation of attributes listed in the table above. Those attributes are all 1-dimensional. A data record contains only necessary attributes or attributes with data.

Target Group Analysis

Who should use this visualation technique?

The visualisation is mainly meant for people collecting mp3s or for musicians who like to see some nice information they won't find in the usual representations. Moreover it's for people who like to explore their music collection under different standpoints. This will fit the exploring character of information visualisation.

Our visualisation tool will make it possible to browse through the music collection in different ways than what iTunes offers. It's meant for people who want to find some music, they for example didn't listen to for a long time. They will see from wich period the songs are, which genre they fall into and how they have been rated. This is a different starting point than usual and will allow the user to probably find other music, than without the tool.

The people that will use our tool are familiar with the internet and use it quite often. Moreover they are using their computer a lot at home and have their pc or mac in the living room where it is not only used to work, but also as multimedia station. That means they play video, music and watch TV with their computer.

They are also interessted in music and they are buying the music from the internet and not buying the songs on a CD. If they buy a CD they rip the content on the pc to have it in their music library.

What are the special interests of our target group?

I think our target group is quite young, probably creative and I suppose they have some sort of interest in playing around with things a bit. As we are analysing iTunes the target group likes nice and neat design, cool tools, cool features and gimmicks. They should use the visualisation mainly for playing around with their music collection.

Our target group has more interesst in mp3s than they have in buying CDs and playing them from their CD player. They have all their music stored on their computer and want to have many possibilities to browse through their collection. They are looking for new ways of finding music so that they can listen to they music they like and the music they feel for in the moment. Instead of putting a CD into the HiFi set they want to see what they have in their collection that suits their mood.

Are there any known / often used Methods / Visualisation Techniques?

Up to now there are no real visualisation techniques in the area of music players like iTunes. The only visualisation techniques can be found at the playback time of a song, when the music itself is visualised.

For representing the Songs of a library ususally lists or tables are used. That means this is not a real graphic visualisation, it's just a representation of the data, that the user can read and use to make for example playlists. Especially in iTunes the possibilities of combining many ID3 attributes to get a playlist are vast. You can for example say you want only songs from 1980 to 1984, that are in the rock genre, that have a rating between 1 and 3 stars, and are not from Guns n' Roses. So you can do quite a lot of things with it, but there is no visualisation behind it. You can choose all that, but you can not choose it graphically which can make this much easier, faster, or just more interessting.

Purpose of the Visualisation

Which tasks should be solved?

Warum haben wir nur wenige Datensätze heraus genommen?

Designproposal

Which kinds of Visualisation should be used?

The actually existing music library systems mostly don't really visualize the contained music using its attributes but simply use lists of items (artists, albums, songs, playlists...). While this concept is easily understandable for the end-user it only allows for simple browsing using the following techniques:

• Searching for specific artists, albums, titles and so on.
• Filtering the list by choosing ranges or values for some attributes
• Sorting the list by different attributes
• Scrolling the list up and down

Some attributes of song items like artist, album and title are of hierarchical character and therefore suited mainly for being displayed in a list or being used for browsing in multi-step lists. One example of such a multi-step list is the media library in Winamp (TODO: cite): After selecting an artist in the first list, the second list gets filled with the available albums of this artist. Selecting an album fills the next list with the associated songs of this album and so on.

TODO: list the different types of attributes

While hierarchical attributes can be considered essential and are very useful for a multi-level selection process or direct textual searching/filtering, they only can be used if the user exactly knows what he or she wants to listen to. In opposite users often want to listen to randomly chosen or special pattern matching song lists like "My Top 10 Most Played Ones".

Used Techniques / Applied Principles

• Focus & Context: Tiled Multi-Level Browser (Slides 0, Page 16)
  • Overview Window
  • Zoomed Window
  • Details On Demand Window
• Scatterplot + Color and Shape concept
• Dynamic Queries: live choosable sliders for attribute ranges (Slides 4, Page 8)
• Linking & Brushing: Detail window containing rating distribution of selected songs (Slides 0, Page 25)
• Visual Encoding: volume, color

Mockup(s) / Fake Screenshot(s)

Entwurf 1