Reviewed: anacubis Desktop v3.0
Company: anacubis, a division of the i2 Group http://www.anacubis.com/
Purpose: Uses a visual interface to retrieve, consolidate, and analyze information from business information vendors, Web sites, internal databases, and spreadsheets.
Starting Price: $2,750 per license, per machine. anacubis Desktop Intellectual Property Analysis Add-In $750 per machine.
Reviewer's View: anacubis Desktop is a novel and sophisticated tool for analyzing data—and connections between data and data sources—imported from a wide range of sources and file types, both textual and quantitative. The price is not trivial, and developing anacubis expertise will require climbing a substantial learning curve. But the potential rewards are great for analysts and companies willing to take the long view.

We're now well into the third decade of the Digital Revolution. Despite daily grumbling—often justified—about annoyances with hardware and software, the glass really is half full in terms of unprecedented digital analysis tools. anacubis is among the more ambitious products in the third generation and third tier of sophisticated tools—applications like geographic information systems (GIS) and social network analysis applications—that retrieve and analyze data and enhance communication of the results of that process.

anacubis digests data from a variety of sources: an analyst's PC, a corporation mainframe, multiple Web sites, or hand-rolled Excel or database files. That data is then sliced and diced with the objective of producing a spider-web network diagram illustrating connections between various entities. Those diagramed entities might be corporations, individuals, patents, political campaign organizations and any combination of the above.

Simultaneously, anacubis displays the underlying, highly specific data tied to the graphic icons. Highlighting an icon shows—in adjacent windows—the fine-grained details attached to that icon, everything from a company's annual sales and number of employees to hot-linked corporate home pages or a stock-ticker abbreviation.

anacubis has partnerships with a handful of large, online databases including Amazon, Google, and Hoover's, Inc. The easiest way to experience anacubis is to visit www.anacubis.com/amazondemo/ amazon. This real-time demo will show connections between any subject, title, or author keyword in Amazon's book database. If the Google or Amazon sample is intriguing, invest the time in a full-blown test drive.

Firing It Up
The minimum hardware required for anacubis is a Pentium III processor, 256MB RAM (512 is better), and 40MB of hard disk space. It runs on Windows 2000 or later, but works only with Internet Explorer or Netscape 7.1 because it makes heavy use of the Microsoft .NET Framework linkages. (Attempts to use Mozilla were maddening.)

I strongly recommend setting the monitor to 1280x1064 screen resolution; anacubis is a graphically intensive application that needs a lot of screen acreage.

anacubis can be downloaded from the Web and easily installed for a 10-day free trial. Submitting the registration form triggers an email to you with the evaluation license key. Note that the 10-day window begins with the issuance of the license key, not the installation of the program, so don't download the program until you've blocked out some time for its evaluation.

The downloaded files include a 37-page PDF tutorial—the closest thing to a traditional manual the company offers—along with the usual Help files on the toolbar. The tutorial is a fairly good walk-through of the program's capabilities, but because many of the concepts are novel and the screen layout somewhat unique, plan on devoting five or six deliberate hours to the tutorial to fully appreciate the program's potential.

The Learning Curve
There are some similarities in the look and feel of a spreadsheet application—especially Excel—and anacubis. The program has standard toolbars and menu buttons and tabs (that can be added or deleted) for different sets of data. A central zone—size adjustable—dominates the screen. anacubis calls this the "View Area," and this real estate is where all of the visual and much of the analytic action is revealed.

The View Area is bounded on the left by a "Property Browser" window that displays the metadata underlying highlighted content icons. Beneath the View Area is the "List Area," which looks like a spreadsheet. The rows display variables and data reflecting all the documents being tapped for a particular analysis project.

To the immediate right of the View Area is the "Filter Area," a column of toggle button icons that permit the user to view or not view specific types of entities. Just as a GIS analyst can toggle the display of maps for rivers or roads, so can an anacubis user toggle the presence of persons, companies, or types of relationships (i.e., competitor, subsidiary, and/or licensee) between entities.

On the far right of the screen is the anacubis "Task Pane." Here resides a wizard that assists the user in integrating customized data into the analytic profile. That data might be drawn from add-ins specific to anacubis partners such as Hoover's and LexisNexis, and it can import data from Excel-specific spreadsheets or databases you create.

As with any analytic tool, the anacubis process begins with establishing one's knowledge-need objectives. In this case, that means literally sharpening a pencil and getting out some paper to sketch what your unique network diagram should look like. Which variables are available and will be linked? Essentially, what do you want to know and how can you know it?

That step is followed by selecting appropriate data variables that will offer insights into those questions and selecting appropriate methods to make the invisible visible. Political debate is in the air, and I had some questions to ask about who was making contributions to federal elections from my state of New Mexico.

I downloaded state-specific data from the Federal Election Commission related to all 2004 federal races, presidential and congressional. As always, the data was dirty: misspelled names or the same person with slightly different names; some middle initials present, others missing; some names all in upper case, others with standard capitalization; and so on.

Because it was a relatively small data set with just over 500 records, I used MS Word to clean up some of these potential problems and to eliminate stray null characters. I then formatted the data fields into unique columns and inserted the necessary end-of-record characters. The cleaned data was pasted into Excel and saved as a worksheet. The entire set was then highlighted and copied to the clipboard. (Version 3.0, just out in September, allows the analyst to dynamically link to an Excel, Access, Oracle, or SQL Server file.)

Switching back to anacubis, I used the Task Pane wizard to guide me through the import process to bring the data to the View Area. My initial attempt was not successful: Only the first two records would import. I'm pretty sure the problem was caused by the novice trying to use the program without first maximizing his free RAM: I rebooted with minimal applications open, and lo and behold, with my next attempt all the records easily slipped into anacubis.

Soon I was highlighting entities, and defining and viewing subsets of campaign contributions. One of the interesting features of the current version is what the company calls the "aggregated view."

In the case of my data, single individuals often made multiple donations to a single political action committee or party. Taken at its most raw form, the network diagram drew links from the donor to the Public Affairs Council for each contribution. Invoking the aggregated view, however, summed the amounts of all those donations and entered that amount in the link label, but only drew one line. That makes for a much cleaner diagram.

• Entities—real-world objects such as a company, person, document, or location.
• Links—real-world relationships such as when a person may be a company shareholder or director and the graphic connection between them.
• Properties—attributes associated with Entities or Links such as revenues, shares held, and DUNS numbers. A Properties list may be edited, expanded or abridged by the user.

That said, a key concept underlying anacubis' ability to winnow valuable data from the infinity of the datasphere is its "Consolidation Threshold." Remember that an anacubis document can be filled with data from a variety of sources? Conclusions reached by the program about how entities are linked are only as good as the algorithms that determine which data being boiled down all pertain to the same company or individual. anacubis sets—but the user can vary—that desired or necessary Consolidation Threshold.

The likelihood that any particular data linkage pertains to the same company is reflected as a positive or negative percentage quickly seen by the analyst. Here's how Paula Wells of anacubis describes it:

"If you compare Nokia and Symbian, then you get a very high negative percentage (-92.4%) as it is extremely unlikely they represent the same real-world company. However, if you compare ‘Sony Ericsson Mobile Communications AB' and ‘Sony Ericsson' you get a positive percentage (42.1%) as it is possible that they do represent the same real-world company as they share a number of common properties."

Highlighting the icon and right clicking can open the company-specific consolidation report describing the rationale for the percentage score.

At the End of the Day
So what did anacubis and the political contribution data tell me? I was able to see concentrations of "transactions"—and secondary and tertiary contribution patterns—to a degree that I wouldn't have seen as easily using filtering tools in a spreadsheet or a database. And I was able to easily drill up or down in all those entities.

That does not mean anacubis is an all-in-one utility: I think the experienced analyst will use it in conjunction with spreadsheet and database programs to see things the other applications cannot display.

anacubis also proves the truism that there are direct correlations between the potential power of an application, the amount of time required to fully understand and apply its features, and the potential to screw up big time if one is not careful with assumptions, commands, and conclusions.

This is not a tool to pick up the night before a big presentation and expect results. But it is a tool worth dedicating some serious time to learning because the potential rewards are great. In fact, anacubis is so rich that even a small organization dependent on absolutely current and deep data analysis might well have a person working full time using this application alone. Just as project managers spend much of their workday living in that specialized application universe, so too will an experienced anacubis driver return a respectable ROI.

Sidebar: Key Features at a Glance

Integrates data from various sources	High-speed algorithms merge Web data and customized database, spreadsheet, or documents and their variables.
Partnerships with licensed data suppliers	anacubis customers have access to news and business data from Dun & Bradstreet, Hoover's, LexisNexis, and Questel•Orbit, a suite of patent and trademark services.
See the connections	Algorithms measure probability of similarity of merged entities—corporations or people or documents—and draw network diagrams of the connections.
Ask the right questions	Query a single or multiple databases, based on your search criteria, and see the results in an Intranet Web Viewer and/or the anacubis Desktop.
Drill down/drill up in the nodes	Highlight a node or connection and the data underlying those entities is immediately apparent. A click brings the analyst back to network diagram.

Sidebar: Business Profile

anacubis, based in Cambridge, England and with offices in Springfield, Virginia, is a three-year-old technology spin-off of the i2 Group. The i2 software is used by more than 2,000 organizations including international, federal, and national intelligence; law enforcement; and Fortune 500 companies. anacubis software is a more general—and business-oriented—application of those same concepts of association and network analysis, and the visual display of those relationships.