VisiNav is a system to search and navigate web data. The conceptual model consists of objects (for example, a person). Objects can have attributes (a person has a name). Objects can also have connections to other objects (a person knows other people).
Users of VisiNav can explore information spaces comprising objects integrated from a large number of (web) sources publishing data in RDF.
VisiNav can be used to search and navigate web data. A screencast demonstrating some examples can be found on the project home page.
There is also an older screencast that demonstrates (among other things) how data from Drupal, a content management system, can be explored with VisiNav:
The data in VisiNav consists of objects (a person, a location, a project, etc.). Objects can have attributes (such as name, date, or latitude) or links to other objects (such as knows, based_near, or made).
VisiNav uses point-and-click and drag-and-drop operations. Items that arecan be dragged and dropped.
With VisiNav, you start by selecting keywords (e.g. "tim berners-lee") that yield an initial set of objects.
Next, you can restrict the set of objects further by dragging and dropping classes () over the query trail.
The detail view contains facets - attribute/value or link/object combinations. E.g. by dragging and dropping "Dan" grouped under the "knows" predicate, you can restricts the result set to all instances which "know" "Dan".
You can also traverse paths from the current result set. E.g. in the detail view showing Tim, dragging "knows" over the query trail yields all people Tim knows.
Depending on the result set, you have the option to visualise the result set in several ways. The Detail view shows all information pertaining to one object, the List view lists a few attributes of ten objects per page. The Table view allows you to view the result set in a spreadsheet-like form, with the ability to add attributes values to the view. If there has been a path traversal operation and you have jumped back to a previous result set in the query trial, you are offered a Graph view. If the result set contains dates, you are offered a Timeline view. If the result set contains geographical locations, you are offered a Map view.
The following lists the functionality depending on the browser used. We would like to support all browsers, so if you know a particular browser well and can fix issues there, please let us know.
The dataset is automatically collected from the Web -- a RDF crawl six degrees from Tim Berners-Lee's FOAF file. We plan to update the data at least once a week.
To get an overview of the entire dataset, click on "Content" in the top left corner. The resulting page lists the top-ten instances of popular classes.
The best way to get data into VisiNav is to put a link in your FOAF file.
rdfs:seeAlso for general links or more specific predicates, depending on the connection to the thing added..
If you come across interesting datasets in RDF, please put a link to a sample URI into your FOAF file. Linking is vital for the web, and helps crawlers to discover data!
That happens when multiple people assign URIs to the same thing.
You can help consolidate the data by adding
owl:sameAs statements to your file, pointing to all URIs that denote you.
In the next crawl, we will pick up the new data and consolidate all identifiers to one.
You can download a dump of the raw data (ex reasoning): current.nq.gz (164M, N-Quads format).
We know. The more people contribute to a dataset, the more chaotic the data becomes, and this site integrates data from tens of thousands of sources. Ranking assuages some of the noise in collaboratively-edited information spaces, but issues with the underlying data surface nonetheless.
While we would like to fix all these errors manually, or send emails to the delinquents, we do not have the time nor the resources to do so. Our hope is that these errors will be fixed over time via social processes (i.e., communciation between data providers and consumers).
You can help with making the web a better place by tracing the error. Mouse-over the predicate/object pair in question and a footnote number will appear that indicates the provenance of the statement in question. If you are sure the statement is wrong, contact the data provider with a suggestion for a fix. Once corrected at the source, wait for the next index update for the error to disappear in VisiNav. Thanks!
This particular error happened because people confuse the URI of a thing and the URI of a web page about it. I.e. the URI for Tim's FOAF file is http://www.w3.org/People/Berners-Lee/card, while the URI for Tim is http://www.w3.org/People/Berners-Lee/card#i. Tthe appendix #i makes a differnt identifier and therefore the difference between denoting a Person or a Document (even if the content looks the same when dereferenced).
Now, assume that somebody (e.g. the guy who maintains
http://www.eecs.harvard.edu/~chaki/card) uses the URI of Tim's RDF file with a
foaf:knows predicate, essentially stating he knows Tim's RDF file.
Doesn't make sense? Read on.
The range of the
foaf:knows predicate is
Thus the reasoner infers that the document is of type
foaf:Person, in-line with the stated facts and OWL Lite semantics.
Establishing that the document URI is an instance of both
foaf:Document is unfortunate and confusing.
However, these type of glitches are typically quickly fixed once detected by the data publisher.
This particular error happens when using the wrong predicate to link to a Ivan's person URI
http://www.ivan-herman.net/foaf.rdf says that the person
http://www.ivan-herman.net/me holds the online account
http://www.ivan-herman.net/me (note the same URI is used for both the
Online Account and
The range of the
holdsAccount predicate is
Now, the reasoner infers that
http://www.ivan-herman.net/me is of type
Online Account, in-line with the stated facts and OWL semantics.
In total, the dataset contains descriptions of about 400k objects (for example, people, locations, or news items). People and organisations across the globe interested in web-related matter maitain the database.
Most of the system components are an outcome of DERI's SWSE project. People working on the project have been supported by SFI and the EC. Andreas Harth designed the user interface and maintains the site. Data processing is done using Jürgen Umbrich's crawling component (MultiCrawler), Aidan Hogan's reasoner (SAOR), Sheila Kinsella's ranking component (IdRank). The source diagram has been inspired by Richard Cyganiak's LOD cloud and benefited from discussions with Jürgen Umbrich, Philipp Obermeier and Rene Schubotz. The screencast has been produced by Stephane Corlosquet and the Drupal community.
Andreas Harth, Paul Buitelaar. "Exploring Semantic Web Datasets with VisiNav". The 6th Annual European Semantic Web Conference (ESWC2009), 31 May - 4 June 2009, Heraklion, Greece. Poster Session (poster)
We have a list of open bugs at the forum.
If you encounter a bug which is not listed there we'd appreciate if you can add the bug. Thanks!
Great! We have a forum for suggestions and feature requests.