Common sense internet

Original author: John Markoff
  • Transfer
While the World Wide Web consists of billions of documents and links that link them together, developers and hundreds of companies creating ever new startups are looking for new ways to understand user needs.

Their goal is to put these needs “at the head of the Internet” so that it becomes less like a catalog and more like a guide. Some of them even develop systems that understand human behavior. The problem of artificial intelligence, when machines can think instead of just executing commands, has occupied researchers for more than a dozen years.

“Web 3.0” is a “project” that is at the very initial stage of its development and which has already been called by the skeptics “unreal”. But the relevant technologies and their development have already found their adherents both in large corporations such as IBM or Google , and in small firms. Their projects now most often focus on practical possibilities, such as recommendations for planning a vacation or forecasting future music hits.

But in the future, more powerful systems will be able to work as personal advisers in the same heterogeneous and complex areas as financial planning: for example, drawing up a retirement plan for a couple or educational consulting, when a particular service chooses the best university for you to enter. For all these projects, technological progress, more powerful computers are only at hand.

“It can be called the World Wide Database,” says Nova Spivak, the founder of a company that develops technology that defines the relationship between pieces of information on the network. “We want to go from Internet related documents to Internet related information.”

"Web 2.0", which suggests the possibility of integrating web applications (for example, electronic geographic maps) and services (for example, photo hosting), has recently been in the focus of attention of Silicon Valley companies. But the commercial interest in Web 3.0 (or, as some call it, the semantic Web) with its idea of ​​smart services has already appeared.

A classic example of the "era of Web 2.0" is "mash-up", a term that implies the integration of various services. For example, a real estate search site with integrated Google Maps in the end is a new, more convenient service with which every user can immediately see all the houses for sale on the map.

And the developers of the “semantic Web” are concerned with the question of how to create a system that can give a clear and maximally complete answer to a simple request like this: “I am looking for a warm resort to relax during my holidays; I have $ 3,000 for it. And by the way, I will have an 11-year-old child with me. ” Under current conditions, the search for such information may take more than one hour: you will have to look through the lists of flights, hotels, and car rental companies. In the conditions of "Web 3.0", the user, ideally, should immediately receive a complete package of information as professionally and efficiently as if an agent of a travel company did it.

How exactly such systems will be built, how long they will be developed, and how soon they will begin to provide correct answers to requests - these are the main topics for the fierce debate of scientists and developers. Some of them focus on creating a new structure that will replace the current Internet, others on developing new applications that can “extract meaning” from existing information flows. But everyone agrees that these systems will bring more revenue than current search engines that issue thousands and even millions of documents, but do not directly answer questions.

To understand the potential of the “understanding human desires” technologies, we can give an example of Page Rank: this technology allows Google to use the potential of human knowledge and solutions to rank search results. (She interprets the link from one web page to another as a “voice”, with the “voices” on more popular pages carrying more weight.)

But researchers are moving on. The company of the already mentioned Mr. Spivak - Radar Networks- explores the content of social sites that allow users to post, collaborate and discuss various types of content, from travel descriptions to movies. The technology of this company is based on a new generation database system that defines and maintains associative connections between pieces of information in the same way as connections between certain people (colleagues, friends, relatives).

An example of the use of this technology is KnowItAll, a project from the University of Washington research team.funded by Google. Within its framework, the Opine system was created, collecting and sorting the opinions of users from various thematic sites. The hotel demonstration project “understands” such parameters as room temperature, comfortable beds and prices, and also distinguishes what is “great”, “good” and “come down” to provide useful answers to requests. On modern sites, the user will have to browse through huge lists of comments and reviews from other users, and the “webtrinol” system will weigh and rank all comments to find the optimal, reasonable solution and help the average user quickly find the right hotel.

“The system will understand that“ perfectly clean ”is better than just“ clean, ”says Oren Etzioni, a researcher at the University of Washington, and the project manager. “We're trying to make it clear that text on the Internet is just a huge source of information.”

The current Internet, it can be said, is in the “Lego phase”, a constructor in which parts of the information are mechanically joined together. Adherents of Web 3.0 see the “future version of the Internet” as an era in which machines will do intelligent work.

Researchers are already celebrating the advent of intelligent technology. For example, special webcams record an illegal invasion of any territory, and programs identify the date, time, place of the invasion, and sometimes even the identity of the criminals, transmitting data to the security panel. Researchers claim that this is the beginnings of Web 3.0.

“This is an important topic: people still don’t understand how much they depend on artificial intelligence,” says Daniel Hillis, specialist in this field, founder of Metaweb Technologies . His company still does not disclose what kind of products it is developing, although it says on their website that "Metaweb plans to build a new Internet infrastructure."

Both Radar Networks and Metaweb are partly associated with military and intelligence structures. Initial research at these companies was funded by the CIA, the NSA, and other US government agencies; work began shortly after the emergence of the concept of the "semantic Web" created by Tim Berners-Lee in 1999.

Intelligence services helped connect researcher Doug Lanat, whose company Cycorpsells systems and services to government and private corporations. The main development of the company is the Cyc artificial intelligence system, which, as promised by Lenat, will one day be able to answer any questions, both written and spoken. Initially, the system was built on the millions of facts introduced into it that it should “learn”, but at a lecture given last year at Google’s office, Lenat said that Cyc is now “teaching” network content. This process, according to researchers, should demonstrate the construction methods of Web 3.0.

Lenat argues that at the moment, the system can already answer questions posed in ordinary human language, for example, “Which city in the US may be affected by anthrax in the summer?”

At the same time, IBM employees say they regularly use a “digital cast” of 6 billion online documents to conduct research and answer various questions from corporate clients. The company used its system in marketing research for television networks, collecting and processing information from online communities. And with the help of information about the popularity of certain musical tracks on various websites of university communities, researchers were able to predict the leaders of the charts of the next two weeks.

Disputes are already underway about whether systems like Cyc can create a “new version of the Internet” or whether human intelligence can independently develop in a new manner using these technologies. Proponents of the second point of view say that this is already happening, and examples of this are sites such and Flickr (the bookmarking system and photo hosting acquired by Yahoo ), as well as Digg , a news service whose content is created by users.

For example, on Flickr, users tag photos with tags, helping others quickly find images of interest to them. “With Flickr, you can find images that a computer would never have found,” says Prabhakar Ragnavan, research director at Yahoo. “What we struggled with over the years suddenly became so simple. And that would not have been so simple without the World Wide Web. ”

Also popular now: