Home > Uncategorized > Paper: World Explorer

Paper: World Explorer

World Explorer: Visualizing Aggregate Data from
Unstructured Text in Geo-Referenced Collections

basisidee:

  • bron: 6 milioen geo-tagged Flickr foto’s
  • tags uitlezen
  • op basis van populaire tags, deze weergeven op de kaart
  • wanneer men over een tag beweegt, ziet men foto’s die met de tag te maken hebben
  • hoe populairder de tag, hoe groter de tag

applicatie: http://tagmaps.research.yahoo.com/worldexplorer.php

quotes uit de paper:

We use these tags to create a visualization tool, World Explorer, that can help expose the content of the data, using a map interface to display the derived tags and the original photo items. We perform a qualitative evaluation of World Explorer that outlines the visualization’s benefits in browsing this type of content.

Flickr supplies a map interface through which users can “drag” their photos to the map locations where the photos were taken.

The visualization exposes, for each map region and zoom level, the high-scoring tags for the generated clusters; these tags are shown as text over the map area where each cluster occurs. In addition, the tags are connected to the corresponding photographs such that when a user points the mouse over a tag, photographs associated with that tag, from that map area, appear in a dedicated pane. Thus, a user can get an idea of what is available in the visible map region via the tags, and explore the available individual items via photographs.

Several similar map-based photo browsing systems appeared on the Web in the last few years. Flickr now also employs a map-based interface to access photos. All of those systems face the problem of clutter in the map interface: as the number of photos available in each location grow, the full set of images cannot possibly be shown on the map at once.

The principal element of the visualization is the “primary tags”. The primary tags are overlaid by the application on top of the viewed map area. The primary tags are meant to give the user an idea regarding the landmarks, points of interest and other items available in the viewed area.

..for each area and zoom level, the system retrieves the appropriate clusters and returns the top-scoring tag for each cluster, the tag’s score, and the tag’s position (the cluster’s centroid).

The principal interaction method in the application is clicking, or simply hovering the mouse over one of the tags…
Clicking a primary tag will show “secondary tags” as well as relevant photos.

The secondary tags are displayed as a drop-down menu when the user clicks one of the primary tags, thus avoiding clutter while still allowing for more details to be exposed.

the needle and the haystack:

This need was most commonly expressed when participants were looking for specific locations or landmarks in a city that they were not familiar with (tasks 1 and 3). For example, while browsing through Barcelona for Task 3, Participant 6 said “I guess what I’m looking for are bull fighting pictures. Is there a way to actually [search]?”.

Participant 1 (“I personally am a very map person”) liked the integration for planning: “I can not only decide what I like and visit, I can even plot it out and first I do this and I don’t need anyone’s help…”.

Participants appeared to have difficulties when the visible tags did not match their mental model of a location. For instance, at certain zoom levels in the San Francisco area, the tag Chinatown was not visible, while Japantown was available; many of the users complained that Chinatown was more important and should be displayed before Japantown. This could be possibly remedied by giving the users a more personalized view of the tags where the tags are tailored to their interests. Participant 1 requested a filter to help plan a trip “If the tags were really interesting and could pick up what kind of person I am and what kind of trip I would do in New York which would be really different from someone like my dad then that filter would really help me visit New York in the way i would.”

About half the participants also wanted to have more information about the tags especially in the unknown locations. Participant 3 summarized it as “now that I think about it just looking at the pictures doesn’t really help, I need text to go with it. So for example ’opera’ all I see are buildings that look like other buildings so it doesn’t really give me a feel for what its like. If there is text behind it it will be more helpful because I don’t know if I’m there to see an opera or a building…”.

The evaluation of World Explorer brought up issues we hope to address in future work. In particular, it was clear that user needs often required both the “needle” and the “haystack” view of the data, even while performing a single task.

ideeën voor eigen thesis:

  • Flickr foto’s cross-referencen met dataset van bezienswaardigheden. Zodat enkel bezienswaadigheden gevisualiseerd worden.
Advertenties
Categorieën:Uncategorized
  1. Nog geen reacties
  1. No trackbacks yet.

Geef een reactie

Vul je gegevens in of klik op een icoon om in te loggen.

WordPress.com logo

Je reageert onder je WordPress.com account. Log uit / Bijwerken )

Twitter-afbeelding

Je reageert onder je Twitter account. Log uit / Bijwerken )

Facebook foto

Je reageert onder je Facebook account. Log uit / Bijwerken )

Google+ photo

Je reageert onder je Google+ account. Log uit / Bijwerken )

Verbinden met %s

%d bloggers liken dit: