Jennifer Zaino
SemanticWeb.com Contributor

During a webinar conducted by Stephen Wolfram of WolframAlpha, he piqued the audience's interest in the technology's upcoming APIs, which were also a subject during his last webinar update. At that time, in June, he had expected a preliminary version of the developers' API to be available soon. Wolfram said on this webinar that within a few weeks the company expects to deliver a core API that will provide a data exposure layer that gives access to every level of detail around the way Wolfram Alpha data is presented and the way lumps of data are exposed to it. Another level of API is designed to help users build widgets they can install on their web site that can interact with WolframAlpha.

Web sites aren't the only places such widgets can be deployed -- they may find homes on mobile devices or within applications themselves. CAD, spreadsheet and other such software can take advantage of this to pull computable knowledge right back into their results, he expects. Wolfram Research's own Mathematica technical and scientific software, in its next version, should support the ability to send queries right to WolframAlpha. The final result can be turning freeform input into precise mathematical input that can be combined with other input to build precise and complex programs, he says.

Wolfram Alpha is also in the process of conducting a big experiment that leverages the technology's underlying linguistic processing to try to understand multi-lingual queries, even if those queries are somewhat broken (for instance, written partially in English and partially in another language). The experiment involves 30 languages and involves trying to turn such queries into something cleaner that can be processed in Wolfram Alpha. Wolfram hopes that technology will roll out in a few weeks. Synthesizing output in different languages remains more of a challenge.

Wolfram Alpha has also started a group to work with enterprises on taking companies' internal data and mashing it up with Wolfram Alpha's knowledge; it's developing a pipeline for such data curation. The company can use Wolfram Alpha itself to handle some of the technology challenges of the data curation process (such as understanding data within a database that relates to regions and currency), but the harder part is understanding what has to be computed from that data to make it useful, how to set up linguistics to access different aspects of that data, and what is important to present from that data. He called this work a "very active effort" and expects that there will soon be some case studies of organizations that are doing this to some interesting ends.

Also, users may want to be on the lookout for how Wolfram Alpha may be leveraging search engines as a distribution channel. "Search is an interesting distribution channel for computable knowledge," he says, and Wolfram Alpha is certainly interested in getting its technology into the hands of as many people as possible. "That means distribution through as many channels as possible," and search is a complementary one to Wolfram Alpha's own mission, he believes.
Also hitting the web last week were two YouTube screen casts on the upcoming Twine 2.0, hosted by Radar Networks founder and CEO Nova Spivack. Twine has been taking a beating in the blogosphere lately, which has covered significant fall-off in its traffic and its need to get 2.0 right to regain traction. Spivack has stated that the company is in fact focusing all its resources on bringing Twine 2 to the market by year's end, and in the meantime is not doing any further work or marketing on version 1 beyond supporting it from a technical and user perspective.

The two YouTube screen casts provide a consumer preview and an ontology editor preview. The Consumer Preview shows Twine in a very different light, focusing less on the world of internal Twines and more on wide-scale web search and navigation. "Twine 2.0 is building a giant semantic index of things, things of various types all around the web," Spivack says. It's turning the web into a "giant database," indexing not just plain text but structured and complex data, and providing database-like filtering capabilities to more precisely hone results with just a few clicks.

This is a web-scale system, Spivack says.

"We're doing semantic extraction on the entire web," he says. "It doesn't require pages to already have semantic web metadata or RDF or anything like that in their pages ... We go out and we analyze the pages and we generate this semantic metadata ourselves. That's because we've really figured out that most people aren't going to learn RDF. It's too complicated."

It can leverage RDF or microformats embedded in pages if they're there, but the idea is why bother if Twine will do it for you? Twine indicates it plans to make this available as linked data available for others to use as well. Spivack says the company is working with publishers now to provide this capability on their web sites on their data, and for their categories of interest, to deliver new kinds of vertical searches on structured data that are more precise than keyword searches.

The Ontology Editor is one of the developer tools Twine is planning to launch later this year in conjunction with Twine 2.0. The editor includes many different existing ontologies, and also lets developers collaboratively author their own ontologies. What Spivack says what sets this editor apart from other tools for creating and sharing ontologies is its ease of use, so that it's readily adopted by the average webmaster or domain expert: You don't have to know RDF, OWL or other "fancy semantic web concepts -- it's all point and click and it's hosted on the web."

They can link their web sites to one of the ontologies in the editor, or create a new one, to make the data within their sites understandable by any application that understands RDF or OWL, including Twine. "If you want to make the web function like a big database you need a standard schema language to define what the data means," he says. "Not just how it's structured but what it means."