NetReality

Dateline: 06/15/99

We've all been witnesses of a marketing hype surrounding intelligent search agents, bots or other "smart" creatures that should help users to harness the vast amount of information available on the Web. First reality check shows that many of them are ordinary software packages that automate and simplify Web-related tasks, but do not show any other behaviour that could be considered intelligent. These terms are today often used to lend marketing appeal to traditional software products that would otherwise not be labeled as "intelligent" - seems that software sells better with "AI inside" label sticked on it.

I have recently downloaded and installed new software product that seems to fulfill these promises. NetReality from Bittco Solutions reads and organizes information for users, presenting Web sites, searches and bookmarks files in a searchable and meaningful virtual reality landscape according to their content and meaning. This powerful form of presentation enables users to quickly locate, correlate, and apply information on the Web to tasks at hand.

NetReality reads and organizes information for users, presenting Web sites, searches and bookmarks files in a searchable and meaningful virtual reality landscape.

The first step will be the hardest for most users: the current Beta 1 version requires Linux OS, so all Win95/98/NT/MacOS... users will have to wait for a few more weeks or install Linux on their boxes. Apart from that, you should have no problems with installation or use of this program. This technology could quickly find its way to the most major browser and OS platforms. It is still too early to predict its success on the market - we'll have to see versions for the the most common operating systems first. My feeling is that people at Bittco have a winner here...

Technically speaking, NetReality is a collection of modules built on top of the Bittco Application Framework (BAF) for analyzing, organizing and presenting collections of Web documents in virtual reality. Bittco views NetReality as an example of this foundation technology which has broad application in personalizing and automating how people and companies interact with information and documents. Although this article focuses on their most recent products, there are others (SmartDocs, Wormhole, etc). The graphical user interface of NetReality consists of two layers - the base functionality of the interface is programmed in C, while the surface layer is scripted in Tcl/Tk, a popular and cross-platform scripting language.

NetReality reads, understands and organizes documents much like a person do. The exact process is described in detail at Bittco's site, but here are just a few basic steps:
NetReality obtains its content from the Web into a cache so that subsequent acquisitions need only fetch new and updated documents. Document attributes and processing information are parsed right from the documents. This currently includes the document title as represented in HTML as well as document size and dates. Processing information includes identification of a suitable cleaner module (currently available for HTML, MS Office, SGML, XML, ps, pdf and ASCII text) that can prepare the document for processing by NetReality down the line. During comprehension, NetReality builds an understanding of all the documents acquired, cleaned and added to the document space. It understands the whole document space by identifying significant word combinations, then using AI techniques to organize a semantic map which represents an understanding of the key words and concepts in the document space. This semantic map, called a genome, maps significant words into a large grid where they are positioned according to similarity. NetReality implements a Self-Organizing Map (SOM) neural nets which uses built-in algorithms as a guide for self-directed learning. Linear Vector Quantizer (LVQ) is also used for the categorization (bucket sorting) of complex data. The resulting genetic sequence of each document provides a very subtle and accurate description of its meaning which is mapped to the virtual reality landscape. These personalized and relevancy-ranked views of the Web, called Worlds, enable users to locate documents of interest quickly.

For more details, including architectura and conceptual block diagrams, visit the sites listed below.

   Site(s) of interest: