Saturday, September 25, 2004

allValuesFrom in iTQL

I like this query because:
* It shows what NOT is (basically allows you to say things like "give me all the states that are not Alabama").
* Uses the new HAVING clause.
* Seems to be a nice use case for count and/or subqueries i.e. the ability to execute an outer query and plug the results into the inner query finally makes sense.
select $s $p $x count (

select $type
from <rmi://.../server1#foo>
where $s $p $type
and not($type <rdf:type> $x ))
from <rmi://.../server1#foo>
where $r <rdf:type> <owl:Restriction>
and $r <owl:onProperty> $p
and $r <owl:allValuesFrom> $x
and $s $p $o2
having $k0 <tucana:occursMoreThan> '0'^^<xmlschema:nonNegativeInteger>
order by $s ;

$k0 is the variable binding for the COUNT - we should offer a way to explicitly bind aggregates to a variable.

Another new feature that Paul's been working on, querying only URIs, makes use of virtual graphs and magical predicates like our data typing support.

Improving Guided Search

Site-search tools expand scope ""It is all about creating the metadata surrounding your content and using it effectively. With automated metadata creation, we can go through a site and pick out portions to expose," he said.

Whereas other search tools go through SQL for relational data queries, Atomz Commerce does relational metadata queries through site metadata.

"It is easier than SQL queries because that [metadata] is readily available on the Web site," Kusmer said."

"Guided search and self-service tools aim to improve on basic keyword site searching by pointing users to what they really want, said Eric Peterson, site operation and technology analyst at Jupiter Research.

"We see search vendors each looking at [their] core technology to see how they can better leverage that into other areas," such as applying natural language processing to self-service or guided commerce, Peterson said. "

Google's XUL

Old but Google's XUL is something for new Mozilla/Firefox users (2 million in 10 days). Via Xuggle.

XUL, something for your generic RDF browser. Not one UI for data but a framework to render your data with. Mozilla/Firefox is using RDF, and a query language, templates and bindings. There's also Kowari's descriptors which allow nesting of templates (vCard example).

Friday, September 24, 2004

Think Local, Search Global

Simple search lightens Net load " Researchers from the University of California at Los Angeles have devised a fast search algorithm that uses local rules to find nodes and content in randomly-formed, scale-free networks such as the Internet."

"The researchers' algorithm is based on the bond percolation threshold, or the smallest probability that a message is guaranteed to reach a core sub-network of highly-connected nodes, said Roychowdhury.

As connections randomly percolate through the network at a low rate, only small, isolated islands form. Once the bond percolation threshold is passed, the core of the network becomes connected. The threshold is an abrupt phase transition like the quick transformation of would in that takes place when water boils or freezes.

The algorithm involves three basic steps: content caching, query implantation, and bond percolation, said Roychowdhury.

Content caching happens when a node joins a peer-to-peer network and performs a one-time short random survey, or walk of nearby nodes and adds its content directory to each of these neighboring nodes.

The query implementation step is similar, but happens at the beginning of every query. When a node has a query, it performs a short random walk and passes the query along to each node it encounters.

These random walks are long enough that any given node will almost surely encounter at least one highly-connected node, said Roychowdhury. "So after these two steps one of the high-degree nodes has a copy of a node's directory, and a query is implanted at one of the high-degree nodes.""

"The key step in working out the algorithm was making sure that each query could reach a high-degree node starting from any node in the network. The conceptual breakthrough was the realization that the whole search process could be implemented locally, through messages passed among neighbors only, said Nima Sarshar, a researcher at the University of California at Los Angeles."

For some background see: Linked.

Matter's has a follow up with a link to: Scalable Percolation Search in Power Law Networks .

Adobe's use of RDF

XMP Lowdown "Adobe looks very strongly committed to XMP. Their decision to make it an RDF-based format, the high profile that Adobe products have in the commercial publishing world, and big print media publishers' growing interest in efficiently tracking their metadata are three factors that combine to make XMP a golden opportunity for the business world to appreciate the value of RDF. This business community doesn't care much about the Semantic Web, and its use of XMP (and hence RDF) will be behind firewalls, but an increased use of XMP in PDF, JPEG, and other formats will eventually mean more files with RDF-based metadata sitting on publicly accessible web servers, and hence a greater extension of the Semantic Web."

Verity Entity Extractor

Verity extracts meaning from content "Extractor scans documents of all types, and identifies the expected "information entities" such as names, addresses, dates and Social Security numbers, but also less obvious ones such as places, measurements and holidays."

Wednesday, September 22, 2004

Does Google's browser have name?

gbrowser "I would just love if the following rumor was true: Google builds a browser based upon Mozilla. Of course, there is some evidence Google is going to build a browser, just whois gbrowser.com"

Google Browser Opens Up Gates For Web Services "This would take Google on a head-to-head strategy with Microsoft. Gmail or something like it could even very easily serve as a personal online server and archive.

The only core product that Google couldn’t challenge with that is the operating software and the office suite."

"In fact, a Google browser move is a strong example of what David Stutz warned Microsoft of; networked computing will overtake PC-based computing and Microsoft doesn’t have a competitive answer.

Google would have a lot to gain from a central position in Internet-based communications. Microsoft would have everything to lose.

Give me one good reason why Google wouldn’t do it."

IBM and its Semantic Web Technologies

Introduction to semantics technology "New semantic information-management schemes enable companies to make better use of their information. What exactly is semantics? And how can semantics technology help your development efforts?"

"Semantics technologies are important in a variety of industries such as life sciences, in which biomedical text plays a fundamental role in knowledge discovery. It can also be applied to numerous solutions, for example autonomic computing, where data correlation and inference technologies can be used as core components for building autonomic computing systems and to perform automated and continuous analysis that can result in actions that protect and heal a system."

"...State-of-the-art ontology management systems either are memory-based or use ad hoc solutions for persisting data...To overcome these limitations, researchers work on database-centered architectures for storing and manipulating ontological data. Such solutions will take advantage of existing standards for data management and the DBMS features that have been optimized over the years (robustness, concurrency control, recovery, and scalability). There is also much activity in the application domain. Two prominent applications that come to mind are the efforts involved in semantic Web services for capability-based discovery and composition of business processes, and the use of semantics with social networks for better collaboration."

Apart from the previous releases (IBM Integrated Ontology Development Toolkit and Ontology Management System (SNOBASE)) there's the more recent Ontology-based Web Services for Business Integration. Only for Windows (screenshots).

Tuesday, September 21, 2004

Early days for the Semantic Web Browser

Google's Browser Project...And Ours... "If true, the folks at Google should get in touch with me... without disclosing too much (yet), we are working on a project (for SRI and DARPA) to build a Java-based fully-semantic open-sourced PIM that grafts Mozilla onto my company's Semantic Applications Platform. The result is an integrated cross-platform PIM suite comprised of an OWL-ontology-based Web browser, e-mail, calendar, to-do manager, and chat... and that's just the beginning..."

FOAF based social network

Clique "Clique is a social networking site helping to drive the next evolution of the Web.

Clique is based on FOAF, an RDF vocabulary which helps to describe people and the relationships between them. These are the technologies that are driving the Semantic Web, which is envisioned to be the next evolution of the Semantic Web, and is well under way."

Pfizer Goes Semantic Too

WD: Pfizer Position Paper From the attached Word document:"There are really two issues here where Semantic Web technologies could be used to great effect. The first of these is the communications issue. How can experimental protocols, descriptions of model systems, statistical criteria for data acceptability, and many other critical elements be effectively communicated between technology silos? The second issue is that of synthesizing results from the various technology silos into a holistic picture of physiology..."

"...there are gaps between discovery, development, and clinical groups as well. These gaps have to do with inter-group communications and with the fact that, while there are some similarities, the different groups often work with very different types of information. Here again it seems that Semantic Web technologies could be instrumental in solving some of these problems, although it is clear that the solutions are not trivial."

"...When presented with a large amount of data and other information, IT departments typically create databases to store the data and provide interfaces by which users can query the data. Unfortunately, there are some problems with this approach. Non-IT users are typically very poorly equipped to perform queries of even middling sophistication, especially when they require the use of a query language such as SQL to accomplish. The upshot is that the data gets into the database, but it doesn’t come out again..."

Via langreiter.com.

Monday, September 20, 2004

Scaling Redland

Impact of storing RDF triples with Redland "In my prototyping Archipel, my pet software configuration management system, I started to use Redland to store the version information as RDF triples. I quickly realised that the RDF storage (stored by Redland using Berkeley DB) was using a lot of space, compared to what I was doing. For instance, storing 143 files generated a database of 624Kb, plus a directory containing the actual file content (the RDF storage did only contain versioning information). This is something like 5 to 6 time the size I was expecting."

"It seems that unless I am not using Redland Python API properly, Redland has an important overhead on storing triples. I hoped to use it as a storage backend for Archipel, because I liked the idea of managing version information in RDF, but the overhead is disappointing, if not discouraging.

However, Redland scales really well, and obvisouly will not grow in an unexpected manner when you reach the million of triples, which makes it really robust."

Data is structured

Forget unstructured data "In other words, all data is structured - it is just that some data is more structured than others. Actually, our misconceptions go further than this. The truth is that a photograph has more structure than a sales order, say. In fact, it has so much structure that we cannot usefully encapsulate it.

Indeed, this has important consequences for the way we that we think about metadata. This usually described as data about data. Now, consider the metadata that you would require to describe the Mona Lisa. We could agree that it was a painting, a portrait of a woman, and that it was painted by Leonardo. But metadata is supposed to describe the data: could you get any two people to agree on how to describe the Mona Lisa? And, if you can't get agreement on your metadata then it ceases to be worthwhile.

Perhaps then that is the distinction we need: it is not a question of whether data is structured or not, but actually about whether the data can be fully described by the relevant metadata."

Sunday, September 19, 2004

So much for the truth

Iraq had no WMD "A draft of the Iraq Survey Group's final report circulating in Washington found no sign of the alleged illegal stockpiles that the US and Britain presented as the justification for going to war, nor did it find any evidence of efforts to reconstitute Iraq's nuclear weapons programme."

A similar NY Times article and the Guardian also has another article on how Iraq is Far graver than Vietnam.

But then, there's the article "Who Cares About the Truth?".
"At the end of the day, is it always better to believe and speak the truth? Does the truth itself really matter? While generalizing is always dangerous, the above responses to the Iraq affair indicate that many Americans would look at such questions with a jaundiced eye. We are rather cynical about the value of truth."

"Sure, we may say we want to believe the truth, but what we really desire is to believe what is useful. Good beliefs get us what we want, whether nicer suits, bigger tax cuts, or a steady source of oil for our SUV's. At the end of the day, the truth of what we believe and say is beside the point. What matters are the consequences."

Saturday, September 18, 2004

Own your Data

The Future of the Semantic Web is Here Today and is Evenly Distributed "In this essay I will argue that the tools to do this exist now, and that all factors point toward a future of owning and publishing all our contributions."

"The most obvious example where redundant work has been noticed recently has been the proliferation of social networking sites like Orkut and Friendster. You had to reproduce the same data across every site. This didn't work out so well. In fact people were sick of these sites pretty fast. Now imagine producing, by hand, an rss feed for every rss aggregator in existence. Worst. Idea. Ever."

"We contribute our thoughts and work to the web now. We all write these incredibly useful pieces of information now, just not for ourselves.

Own your data. This future is here and is evenly distributed."

Friday, September 17, 2004

Drops to make an Ocean

many little models "Like many others (e.g. lost boy and more news) I’m wondering about the scalability of RDF data stores. In particular, we have many very small models (under 25 statements for most models, the largest is unlikely to exceed 100 statements) which we don’t really need super fast access to, but store more for reflection. While at current we only have twelve thousand of these models, we are expecting to collect much closer to 100,000 each term, and want to keep data around for at least the last couple of years (si lets say a million or so models). The 12,000 or so has already kicked our database to its knees (MySQL, Jena 2.1, an old P3 700Mhz with minimal ram), which we didn’t really expect."

Sesame 1.1 RC1 Released

"1.1-RC1 is the first official Release Candidate for Sesame 1.1. This release has a number of major improvements:

* The Graph API, an extension of Sesame's access APIs, allows fine-grained manipulation of RDF models directly from Java.
* The Native Disk Store is a new storage backend that works directly on the file system, without need for a DBMS. It uses B-Tree indexing on binary files for fast, efficient and scalable storage.
* SeRQL revision 1.1 is a syntax revision that makes SeRQL queries even easier to read and write, and makes embedding in XML easier.
* Blank node handling has dramatically improved compared to 1.0.x.
* RDF Schema inferencing has been updated to be fully compliant with the W3C RDF Semantics Recommendation.
* Support for MS SQL Server as storage backend RDBMS. Thanks to Adam Skutt for providing fixes and suggestions for this.
* Partial OWL reasoning support through Sesame's custom inferencer."

From openRDF.org.

In A new Sail for Sesame thread: "1. Loading of multimedia metadata (includes metadata-extraction from 2777 MP3 files, transformation and load; so don't care for the absolute times...):

MSR: 85 sec (49089 explicit + 139894 inferred =188983 total statements)
NR: 87 sec (49089 explicit statements)

=> As NR does not generate inferred statements, I'm not sure what the performance will be with inferencing, but this looks good to me at the moment.

2. Renaming of a Class with 2777 associated instances:
MSR: 52 sec
NR: 3 sec
=> Wow! This is much better and depends on the number of triples involved (in contrast to the memoryrepository)."

Live Link Sharing

What are Live Bookmarks? "Services like http://del.icio.us let you publish your own bookmarks as RSS feeds, so that other Firefox users can subscribe to your bookmarks through Live Bookmarks. Live Bookmarks and del.icio.us makes it easy to share cool sites you like with your friends."

Thursday, September 16, 2004

More Military Intelligence

Uncle Sam's Semantic Web ""The beginning of the Iraqi operation was postponed for weeks because information systems couldn't be made interoperable in the time required," said Hendler. "Systems couldn't talk to one another." It was a problem, he says, that a Semantic Web framework could have solved."

"As more triples are created, checkbooks are coming out. TopQuadrant, one of the co-sponsors of the conference, has decided that the Semantic Web is a bankable technology, and predicts that a U.S. $63 billion market for "Semantic Technology" will emerge by 2010.

Miller, the W3C Semantic Web Activity lead, is not as confident in these numbers -- not from a lack of confidence in the technology, but from a suspicion of all predictions. "I am cautiously optimistic. I'm seeing an increase in tools and services," said Miller."

Wednesday, September 15, 2004

The Right Amount of Integration

Killing the "WinFS is About Making Search Better" Myth "At its core, WinFS was about storing strongly typed objects in the file system instead of opaque blobs of bits. The purpose of doing this was to make accessing and manipulating the content and metadata of these files simpler and more consistent. For example, instead of having to know how to manipulate JPEG, TIFF, GIF and BMP files there would just be a Photo item type that applications would have to deal with. Similarly one could imagine just interacting with a built in Music item instead of programming against MP3, WMA, OGG, AAC, and WAV files. In talking to Mike Deem a few months ago and recently seeing Bill Gates discuss his vision for WinFS to folks in our building a few weeks ago it is clear to me that the major benefits of WinFS to end users is the possibilities it creates in user interfaces for data organization."