Geospatial Semantic Web Blog

Reading Martin Soutschek’s question to my previous post triggered me to think of a new way to do photo geotagging. The idea is to annotate photos with machine tags that point to geonames features.

For example, you can tag Golden Gate Bridge photos with

geonames:feature=5352844

How does it work?

Read the rest of this entry »

I was invited to give a talk at UMBC. I decided to talk about Geospatial Semantic Web – things that I’ve blogged about on GSWB.

An introduction to Geospatial Semantic Web technology. An invited talk given by Dr. Harry Chen at UMBC (CMSC 491/691M Spring 2007 Special Topic Course on the Semantic Web).

This presentation covers key issues related to Geospatial Semantic Web: (1) extracting hidden knowledge from unstructured and structured data, (2) knowledge fusion over heterogeneous data sources, (3) ontology sharing and (4) building user-friendly Semantic Web applications.

It also describes state-of-the-art technologies that attempt to solve these problems. This discussion covers upper-case Semantic Web vs. lower-case semantic web, GeoRSS, W3C Geo ontology, GML, Flickr machine tags, geonames and Google Maps mashups.

Since this is a combined undergraduate and graduate course, I tried to stay fun with my presentation and didn’t dive deep into the technical details. The slides (PDF) are available for download on ebiquity. You are free to use the slides in anyway you want. If you want a PowerPoint version of the slides, let me know.

Google Maps now supports data files that are described in GeoRSS. GeoRSS is a set of vocabularies for encoding location objects in RSS feeds. The Goolge Maps API Official Blog has few demos that highlight this new feature.

This is a great news to the GeoRSS community. For awhile, people criticize GeoRSS for being too complex and lacking strong industry backing. Many others think KML, Google’s own location markup language, will eventually overtake GeoRSS because of the popular usage of Google Maps and Google Earth.

The situation is now completely different. With Google Maps embracing GeoRSS, we should see greater demand and usage of GeoRSS on the Web. Could this be the beginning of a new geospatial (semantic) web trend?

I hope so. GeoRSS and its alike should encourage more people to express location information on the Web with explicit description. This may eventually lead to the habit of publishing other types of semantic information on the Web (e.g., SKOS, FOAF and Dublin Core).

Other interesting GeoRSS + Google Maps demos:

• Pleiades — study of ancient geography
• Charles Chuang’s MyMap — maps of different locations in Taiwan.
• Irish Blogs on a Map – RSS -> GeonamesRSS -> Google Maps

According to BusinessWeek, AOL is offering users a new instant messaging capability that allows them to see the physical location of their buddies. Currently this new location-aware feature is available only to IM clients on desktop or laptop computers. Supports for cellphones are coming soon.

Central to this new feature is a plug-in called Skyhook. It doesn’t rely on GPS to provide location data, but rather it uses Wi-Fi signals.

Read the rest of this entry »

In my previous post, I wrote about Metalink, a new markup language and technology for unifying Internet download. The current Metalink language is defined in XML. Publishers use this language to describe meta-data about their download files on the Web.

Today Anthony Bryan email me an interesting example that maps an existing Metalink description into RDF and uses SPARQL to query this description. In his message, Anthony didn’t mention the author of this example. But I suspect it’s the creation of Dan Brickley (since couple URL points to spypixel.com) — maybe Dan had blogged about this before, but I missed his post.

Here is how the example goes.

1. Given a Metalink file (e.g., OpenOffice Metalink), use some transformation technology (e.g., XSLT) to map the XML content into RDF — here is an RDF document output.
2. To check if this RDF document is well-formed, use the W3C RDF validator.
3. Let’s assume that this document and many others similar documents are stored in a Metalink directory server or in Swoogle. We can build applications to query those servers for Metalink data.
4. Here is a SPARQL SELECT that generates information about different OpenOffice download URL and their respective downloading preference score (see also a PDF screenshot of the output result).

This is what we can learn from the above exercise.

Read the rest of this entry »