The social web bookmarking service Delicious.com [2] is a popular social tagging network with millions of registered users and over one hundred million unique URLs bookmarked (as of September 2007 [1], end of 2008 [8]).
By registering at Delicious.com, users can store and share own bookmarks, i.e.
links to URLs of web pages, or discover bookmarked web pages of other users in the network. For this, each user has got a homepage in the social network, listing all her bookmarks. When saving a bookmark on her homepage in Delicious.com, a user can make notes on each web page, allowing for a content description or other related comments. Notes on web pages are shown in addition to the bookmark on the user’s homepage. An example of a user homepage in Delicious.com (as of October 2011) is given in Figure 3a.
While bookmarks are shown in an non-hierarchical and unstructured flat list, the organisation of bookmarks is achieved by tags, individually chosen by each user. Tags
(a) Example for user homepage in Delicious.com
(b) Bookmarks in Delicious.com about "Social Tagging Network"
Figure 3: Example screenshots of Delicious.com
are usually assigned when a bookmark is saved in Delicious.com but also can be added at a later time. They are used to categorise or describe the content of an associated web page with single keywords. On the right hand side of a homepage, there is a list of all tags assigned to web pages which allows users or visitors of the homepage to navigate through all saved bookmarks and to locate web pages of interest. By clicking on the tags in this list, only the bookmarks annotated with the respective tag are displayed. In general, all bookmarks listed on a user’s homepage are public and visible to everyone if not explicitly declared otherwise as private (on a per bookmark granularity).
Apart from the tags and notes used to describe a web page, the list of bookmarks additionally shows for each URL—and this is true for all URLs discovered in De-licious.com—how many users in Delicious.com saved the same web page on their homepages (but not necessarily annotated it with the same tags). Users are offered many ways for discovering bookmarks. For example, they can browse through most recent or popular bookmarked URLs or through most recent or popular used tags. For one or more given tags, Delicious.com lists all bookmarks annotated by any users with these tags. The latter allows to search for web pages of certain topics. An example is given in Figure 3b which is a screenshot of the results when searching for the tags
"social", "tagging", and "networks".
Moreover, by clicking on the number showing how many users have saved the same bookmark, it is possible to navigate to the homepages of those users and browse through their list of bookmarks. In this way, it is easily possible to find other users with preferences matching personal interests. Finally, it is possible to remember those users on the own homepage to quickly have got access to their bookmarks.
At the time when we crawled the Delicious.com website, users could subscribe to another user’s homepage or to only one or more single tags used by another user. In this way, either all bookmarks or only those annotated with a certain tag were listed (separately) on the own homepage, too, being updated when new bookmarks were saved by the subscribed user.
However, just recently, on 26th of September 2011, Delicious.com changed their terminology and slightly modified their functionality [7]. Instead of subscribing to a user’s bookmark list, a user now can follow another user and instead of subscribing to single tags, a user can follow another user’s stack. A stack allows for thematically grouping bookmarks by creating a stack for an arbitrary topic and assigning URLs to this stack. To give an example, a user can create a stack about "healthy recipes" and, in her sole discretion, then adds web pages to this stack, matching the given topic.
Previously, a user achieved the same by adding the tag "healthy recipes" or ("healthy"
and "recipes") to each URL belonging to this topic. Thus, the principle is unchanged:
A user can choose “friends” to easily have got access to all or a group of bookmarks on a friend’s homepage in Delicious.com.
Another change in Delicious.com is, that the bookmarks from users being followed are not listed anymore on the own homepage (like in the former case when a user subscribed to another user’s homepage). Instead, one has to navigate to her homepage by clicking on her name in the list of followed users in order to see her bookmarks.
Matching the SENSE Data Model
Obviously, users and tags in Delicious.com match our data model as given and most relations between these entities apply in a straight forward way, too. Therefore, we only refer to mappings being non-obvious in Delicious.com.
• Documents. The user provided contents in Delicious.com are bookmarks and, thus, correspond to the documents in our Data Model.
• (social) Friendship Relation. Naturally, the subscription or following activity of users coincide with the friendship relation as defined in our data model: If a user U subscribes to / follows another user Uf or subscribes to one of her tags / follows another user’s stack, Ufis considered a social friend of U . At the time we crawled Delicious.com, only the subscription functionality ex-isted. Hence, in our dataset two users participate in a friendship relation when one user subscribed to the other user or to at least one of the other user’s tags.
• Document-Document Relation. Bookmarks are links to web pages which in-clude hyperlinks and, thus, the document-document relation is inherently defined on the hyperlink graph of web pages. However, as the set of web pages harvested by following the bookmarks from Delicious.com did not exhibit a reasonable amount of hyperlinks among them, we did not consider a document-document relation in our studies on Delicious.com.
The dataset from Delicious.com was retrieved by subsequent crawls and, thus, varies with different experiments. Therefore, the details about the employed dataset are given together with the description of the respective experimental setup in Section 5.4 and 6.4.