Strategic Developer | Martin Heller » TAG: Semantic Web

August 23, 2008 | Comments: (0)

Freebase Parallax a promising search tool

TAGS: JavaScript, RIAs, Semantic Web, Technologies, Web

Freebase is a semantic Web site from Metaweb that I've mentioned before (here and here), and is one of the technologies used by Powerset, which was recently acquired by Microsoft. The Freebase data set has been growing by leaps and bounds, and its user interface has improved on a monthly basis.

Recently, David Huynh joined Freebase from the SIMILE project at MIT, and he's already brought insight and technology from SIMILE to bear on Freebase: witness his Parallax prototype (see the screen shot at left; click on it for a full-size view).

A normal search like Google returns a set of results and allows you to look at them one at a time. A normal information site like Wikipedia has individual articles on subjects, loosely joined by hyperlinks.

Freebase adds a layer of ontology and semantic relations to information gleaned from sites like Wikipedia. So, for instance, an article about Jon Udell at Wikipedia says that he's a U.S. journalist who used to work for InfoWorld and now works for Microsoft. A similar article about Jon at Freebase adds structured ontological classification information for Jon as a person and an author. So, for instance, you can find Jon in Freebase not only by name, but also by searching for authors or people born in Philadelphia in 1956, and you can also follow the Publishing relation to his book.

Parallax takes advantage of that additional information to search Freebase with sets, examining many-to-many relationships. So, to use the example in David's video, you can ask Parallax to find the collection of U.S. Presidents, filter that by Republicans, find all of their children, find all the places their children went to school, and create a map of those schools, using one search per step.

Even better, David has open-sourced Parallax on Google Code. I pulled down the trunk source code from Subversion on Friday, and had a good look at it. It's one honking sophisticated JavaScript application, which (if I read it right) uses SIMILE technology and jQuery to call the Freebase Metaweb API.

Yes, there are bugs: it is, after all, a new prototype. One major bug is that every keystroke raises an error in IE 7 if debugging is enabled. It works fine on Firefox 3, however, although the error console does list a few CSS issues. There are also some issues that crop up when you do a search that returns a large set.

Despite that, color me impressed. It's nicely conceived, and very promising.

Posted by Martin Heller on August 23, 2008 08:22 AM

August 23, 2007 | Comments: (0)

Freebase is open

TAGS: Semantic Web

Freebase, which I discussed just the other day, has now opened to anyone, at least for reading and searching. Registering so that you can contribute to the site still requires an invitation. I have 20 invitations, and I can probably get more if they run out. Email me at martin_heller@infoworld.com if you need one.

One of the recent improvements in Freebase is that you can now construct the URL for a common topic by guesswork. For example, try:

http://www.freebase.com/view/bob_dylan, or

http://www.freebase.com/view/jane_austen, or

http://www.freebase.com/view/pride_and_prejudice.

The power of Freebase still lies in its data typing, but from a user's point of view that is a little more transparent than it was, because they've improved the algorithm for auto-completing searches within the site. You'll see that when you try it. For example, go to http://www.freebase.com/view/ and start typing "new york" into the search box.

Enjoy!

Posted by Martin Heller on August 23, 2007 08:24 AM

August 20, 2007 | Comments: (0)

Recovering from a Senior Moment

TAGS: Observations, Semantic Web

The other day around the dinner table, we were discussing movies adapted from books. I wanted to add something specific to the conversation, but the memory wouldn't come.

I remembered that the movie was an adaptation of Emma, by Jane Austen. I remembered that it was made in the 1990s, and that I had watched it with my daughters. I remembered that the movie was set in modern L.A., and that the young actress playing the Emma analog was blonde. I could see her face, and the faces of her character's best friend, father, step-brother, and "project" (the analog of the Harriet Smith character in the book.)

And there I stopped. I couldn't remember the name of the movie, or the name of the ingénue, or of any of the other actors. Searching Google and IMDB for "Emma" and "Jane Austen" got me nowhere. Neither my wife nor my younger children could fill in the blanks for me. I sent off emails to my older daughters, and then tried looking this up in Freebase, figuring that data typing would help me out.

It did: the entry for Jane Austen (author) had film writing credits for many movies adapted closely from and named for her books, plus Clueless, which was what I was seeking, and Bride and Prejudice, a Bollywood-style adaptation of Pride and Prejudice. In addition, the book entry for Emma listed two movie adaptations, the 1996 Emma starring Gwyneth Paltrow, and the 1995 Clueless starring Alicia Silverstone.

Later on, I discovered that there's an article on Freebase of type Film Collection called "Films based on Jane Austen novels," which has an associated discussion about whether Clueless and Bridget Jones's Diary should or should not have been included. (Both were, eventually.)

I also discovered later that the Wikipedia article on Jane Austen includes a Filmography section. In fact, it's more complete than either of the Freebase lists of Jane Austen adaptations, since it includes TV as well as film. By then, of course, I didn't care: I had already recovered from my little senior moment.

Freebase is still in Alpha. I have a few invites left; email me if you want one.

Posted by Martin Heller on August 20, 2007 02:24 PM

March 21, 2007 | Comments: (0)

Freebase, the Semantic Web, and the Metaweb Query API

TAGS: Databases, Semantic Web, Technologies, Web

As I discussed in my article on the Semantic Web for our Crackpot Tech feature on February 19th, the standard Web was originally designed for document distribution, and has yet to realize its full potential for distributing data. The Semantic Web is an effort to relate information by classifying it and linking the classifications.

Some of the efforts related to the Semantic Web concentrate on ontologies, or systems of classification. As useful as ontologies can be, they often seem dry and academic to me.

Once ontologies are turned into database schemas, they often make more sense, at least to me. And once the database is implemented and the application built, it all falls into place.

Imagine my delight, then, to find that the new Freebase site is something like a modifiable database already integrated with a Web application, or as the Freebase.com FAQ puts it:

Freebase.com is home to a global knowledge base: a structured, searchable, writeable and editable database built by a community of contributors, and open to everyone.  It could be described as a data commons.

How is that different from Wikipedia? There's a FAQ for that, too:

It's an apple versus an orange: each is deliciously different. Wikipedia is an encyclopedia with information arranged in the form of articles. Freebase is more of an almanac, organized like a database, and readable by people or software. Wikipedia and Freebase both appeal to people who love to use and organize information. In fact, many of the founding contributors to Freebase are also active in the Wikipedia community. Whenever Freebase and Wikipedia cover the same topic, Freebase will link to the Wikipedia article to make it easy for users to access the best of both sites.

The Freebase type system is basically a flexible, editable ontology. For example, in the computer domain are types about computer hardware, software, computer science and theory, for example Programming Language. If you're browsing the Programming Language type, you can filter the 94 currently listed languages by the properties of the type: Name, Parent Language, Language Paradigms, Influenced By, Influenced, Dialects, Language Designers.

If I type "Gui" into the Language Designers filter entry, I get a drop-down completion of Guido van Rossum, along with a pop-up entry about Guido of type Programming Language Designer. If I filter by his name, I of course get an entry for Python (in this case, a description based on a Wikipedia article), which has the type Software as well as the type Programming Language.

Get it?

There's more. Freebase has an open API, the Metaweb Query API. Here's a sample read query, broken into two lines so that you can see it all:

http://www.freebase.com/api/service/mqlread?queries={"albums":{"query":{
"type":"/music/artist","name":"The Police","album":[]}}}

If you have a Freebase account and have used it on the browser running the query, this will most likely return a JSON-format response giving a list of albums by The Police, which you can save as a text file and view with an editor. Otherwise, it will probably give you an Error 401. It will not give you back articles about police in law enforcement, because we have restricted the queried record type to musical artists and asked to be given a list of albums.

Now, getting back a text file is not exactly stimulating stuff, but this API can easily be turned into an application using some fairly simple code like the freely available JavaScript for parsing JSON format. It would not necessarily have to be an application that interrogates Freebase. And there, eventually, is how Metaweb, the company behind Freebase, expects to make its money: by licensing commercial applications using its technology.

Want to give it a try? It's currently still in Alpha test, and by invitation only. You can try entering your email address at Freebase.com; I don't know how quickly you'll get a response. On the other hand, if you know someone who has Freebase invitations to give out, you can probably get on in a matter of minutes.

At this point, I have five invitations to give out. They will go to people I actually know, so please don't ask me for one if we haven't already met.

Posted by Martin Heller on March 21, 2007 06:00 AM