BibSonomy is a web-based social resource sharing system which allows users to organise and share bookmarks and publications in a collaborative manner. Apart from standard folksonomy features such as an intuitive user interface, navigation along all dimensions, or browser integration via RSS feeds, BibSonomy provides tag hierarchies, group management and privacy features, and numerous import and export functions.
Social resource sharing systems are web-based systems used to manage resources on the web in a collaborative way. Users can describe the resources with arbitrary words, so-called tags. The systems can be distinguished according to what kind of resources are supported. Flickr, for instance, allows the sharing of photos, del.icio.us the sharing of bookmarks, CiteULike1 and Connotea2 the sharing of bibliographic references, and 43Things3 even the sharing of personal goals and resolutions. Our own system, BibSonomy,4 can be used for sharing bookmarks and BIBTEX entries simultaneously. In their core, these systems are all very similar. Once a user is logged in, he can add a resource to the system and assign arbitrary tags to it.
The collection of all users’ tag assignments is called a
folksonomy. The word ‘folksonomy’ (coined by Vander Wal in [
2http://www.connotea.org
3http://www.43things.com
4http://www.bibsonomy.org
a blend of the words ‘taxonomy’ and ‘folk’, and stands for
conceptual structures created by the people. Folksonomies are thus
a bottom-up complement to more formalized Semantic Web
technologies, as they rely on emergent semantics [
A typical user interface allows for exploration of the folksonomy in all dimensions: for a given user one can see all resources he has uploaded, together with the tags he has assigned to them; when clicking on a resource one sees which other users have uploaded this resource and how they tagged it; and when clicking on a tag one sees who assigned it to which resources (see Figure 1).
After an introduction to the user interface and architecture of
BibSonomy, we give an overview about some of its advanced
features. For the underlying data structure of the system and more
details, we refer to [
BibSonomy with its more than 5,000 registered users allows to
share bookmarks (i.e., URLs) as well as publication references.
The data model of the publication part is based on BIBTEX [
A detailed view of one bookmark post from the list in Figure 1 can be seen in Figure 2. The first line shows in bold the title of the bookmark which has the URL of the bookmark as underlying hyperlink. The second line shows an optional description the user can assign to every post. The last two lines belong together and show detailed information: first, all the tags the user has assigned to this post (web, service, tutorial, guidelines and api), second, the user name of that user (hotho) followed by a note, how many users tagged that specific resource. These parts have underlying hyperlinks, leading to the corresponding tag pages of the user (/user/hotho/web 5, /user/hotho/service, . . . ), 5All paths given in parentheses are relative to http://www. the users page (/user/hotho) and a page showing all four posts (i. e., the one of user hotho and those of the 3 other people) of this resource (/url/r, where r is a hashed representation of the resource). The last part shows the posting date and time followed by links for actions the user can do with this post – depending on whether this is his own post (edit, delete) or another user’s post (copy).
The structure of a publication post displayed in BibSonomy is very similar, as shown in Figure 3. The first line shows again the title of the post, which equals the title of the publication in BIBTEX. It has an underlying link leading to a page which shows detailed information on that post. This line is followed by the authors or bibsonomy.org. editors of the publication, the journal or book title and the year. The following lines show the tags assigned to this post by the user, whose user name comes next, followed by a note on how many people tagged this publication. As described for bookmark posts, these parts link to the respective pages. After date and time of the posting follow the actions the user can do, which in this case include picking the entry for later download, copying it, accessing the URL of the entry or viewing the BIBTEX source code. 1.2
This section briefly describes some extensions of BibSonomy which go beyond the basic folksonomy model and have evolved during the practical use of the system. 1.2.1
Tagging gained so much popularity in the past two years because
it is simple and no specific skills are needed for it. Nevertheless the
longer people use systems like BibSonomy, the more often they ask
for options to structure their tags. A user specific binary relation ≺
between tags as described in our model of a folksonomy (cf. [
Therefore we included this possibility in BibSonomy and extended it further to use it for conceptual navigation. For instance, it is possible, given a tag, to show all posts with one of the subtags of the given tag.
Figure 4 shows details of the relation management in BibSonomy. Relations can be edited manually, but usually they will be created in the normal tagging process by tagging a resource, e. g. as eclipse->java, expressing that the tags eclipse and java should be assigned to a particular resource, and that the pair (eclipse, java) should be inserted into the user’s ≺ relation. When browsing the folksonomy, a user can decide if she wants to make use of the tag hierarchy, e. g. when querying for the tag java, resources that are tagged with eclipse can be included in the result set, even if they do not have the tag java themselves. In order to distinguish between simple tag queries and those involving subtags, we call the latter one a query for java as a concept.
While it is not enforced upon the user how the ≺ relation is to be used, we expect that most of the time, it will be used in order to express subsumptions in an is-a hierarchy. The actual use we are observing confirms this assumption. 1.2.2
In particular for literature references there is the problem of detecting duplicate entries, because there are large variations in how users enter fields such as journal name or author. On the one hand it is desirable to allow a user to have several entries which differ only slightly. On the other hand one might want to find other users’ entries which refer to the same paper or book even if they are not completely identical. Hence it is necessary to map these entries together to allow such browsing functionality.
To fulfill both goals we implemented two hashes to compare publication entries at different levels of granularity. One is used to warn the user if she posts very similar BIBTEX entries twice, possibly creating unwanted redundancy. The other one is used to aggregate BIBTEX entries that were posted by more than one user in a common view, providing an opportunity to pick the most complete one or copy over missing data to one’s own version of the publication entry. The implemented solution does not allow for a common creation of a single (correct) entry in a wiki style as we want to allow every user to store BIBTEX in the way she likes without changes from other users. 1.2.3
To encourage users to transition from other systems, we implemented an import functionality. For del.icio.us, this functionality also takes into account the del.icio.us bundles which are named sets of tags. We map bundles to relations. Furthermore it is possible to import bookmark files of the Firefox6 web browser, where the typical folder hierarchy of the bookmarks can be added to the user’s ≺ relation.
Existing BIBTEX entries can be imported by uploading files, past
ing BIBTEX snippets into BibSonomy, or just marking a BIBTEX entry on a web page and hitting the post button. Furthermore, for numerous digital library services including the ACM Digital Library, SpringerLink, arXiv, and CiteSeer, automatic screen scrapers for publication metadata are provided.
For unstructured publication metadata, such as the publication lists often found on researchers’ web pages, semi-automatic extraction using the Mallet7 information extraction tool is supported. 1.2.4
The content of any page in the BibSonomy user interface can be presented in different formats for export. Exporting BIBTEX is accomplished by preceding the path of an URL with the string /bib – this returns all publications shown on the respective page in BIBTEX format. For example the page http://www.bibsonomy.org/ bib/search/text+clustering returns a BIBTEX file containing all literature references which contain the words “text” and “clustering” in their fulltext.
HTML-formatted8 publication lists can be exported which can be easily included into personal homepages. For example, a user schmitz might want to tag his own publications with myown and the year of publication. This enables him to include the page http:// www.bibsonomy.org/publ/user/schmitz/myown+2006 into his personal web page to get an automatically updated list of his publications on the web.
Other exports such as XML, RSS and BURST9 feeds, RDF according to the SWRC ontology, BIBTEX and EndNote work similarly.
Furthermore, links can be provided to an OpenURL10 resolver. Such a resolver allows every user to find any publication presented in BibSonomy in her own local library. 1.2.5
Group Management and Privacy.
8A small heuristic is applied to handle special letters e.g. accents but LATEX commands are not used to format the output. 9http://www.cs.vu.nl/∼pmika/research/burst/BuRST.html 10http://www.exlibrisgroup.com/sfx openurl.htm with
In many situations it is desirable to share resources only among certain people. If the resources can be public, then one could agree to tag them with a special tag and use that tag to find the shared resources. The disadvantage is, that this could be undermined by other users (or spammers) by using the same tag. To solve this problem and also to allow resources to be visible only for certain users, we introduced groups in BibSonomy which gives users more options to decide with whom they share their resources.
BibSonomy is, to the best of our knowledge, the only folksonomy system currently online which combines bookmark and publication management in a common user interface.
In addition to standard folksonomy features, BibSonomy provides capabilities for structuring a user’s tag cloud, group and privacy management, and various import and export options, including screen scrapers for popular publication services as well as for unstructured publications lists. 2.2
There are several important topics which we will be address in
the near future. As stated in the introduction folksonomies can be
seen as a lightweight knowledge representation. One major goal
is therefore the convergence with the semantic web effort which
is also called Web 3.0. To reach this goal a more machine
understandable tagging is needed which can be reached be using so
called “machine tags”11 but also by developing new methods to
extract semantics from folksonomies. Our next steps in this direction
but also to enhance the usability are as follows:
Ranking The FolkRank [
Conceptual Clustering and Community Detection We are
currently investigating different approaches [
Tag Recommender and Ontology Learning As users are
providing tag-tag relations in BibSonomy, we are currently
investigating techniques which enable the semi-automatic learning
of the tag-tag relation. A first step in that direction is the
learning of subsumption relations using association rules on
the folksonomy data set [
API Experience has shown, that an Application Programming
Interface (API) is crucial for a folksonomy system to gain
success. It is something which has been requested by many
people and which allows for easy interaction of BibSonomy with
other systems. Hence we are currently investigating several
approaches to add an API to BibSonomy. Most systems use
lightweight APIs similar to the idea of REST [