Knowledge mpping. (Data Management).

Information Builders 

Now that computer speeds and storage are capable c/storing and retrieving vast amounts of data, emphasis is moving towards its intelligent and meaningful exploitation, including the filtering of the vast resources of the knowledge available from the World Wide Web.

Extending the Web Paradigm

One of the great advantages of the World Wide Web is its ability to move from page to page with a single click of the mouse on a "hot spot." Knowledge Mapping follows this same paradigm but with more options. Instead of taking a user to merely one destination, each hot spot can suggest many possible destinations.

For example, if you have created a report about quarterly sales of particular products, and you want to determine why one particular product has inordinately low sales, you can follow a link to Quality Assurance, Production Scheduling, or Product Defects, according to the way the data has been mapped. The time period and product type will be preserved in each successive query, minimizing the number of clicks to obtain the desired information.

While standard Web queries follow single or "uni" links, Knowledge Mapping within the WebFOCuS environment enables multi-link reporting.

Reports created with Knowledge Mapping employ advanced functionality such as dynamic pop-up menus, drag-and-drop capabilities, and intelligent drill-downs. The path of inquiry is not limited to data in WebFOCUS databases; it can encompass any Web data that can be reached via a URL. Once Knowledge Mapping is enabled, users can simply "drop" a field or variable on any valid URL. Even Web searches can be incorporated into reports to allow speculative investigations and a truly dynamic return of information.

Traditional Business Intelligence

Researchers at Gartner Group define business intelligence as a user-centered process of exploring data, data relationships, and trends, thereby helping to improve overall decision-making. This involves an interactive process of accessing data and analyzing it to draw conclusions, derive insights, and communicate findings with the purpose of affecting positive change within an enterprise.

Jane Griffen, director of the Analytical Solutions Practice at Arthur Andersen Business Consulting, gives a more succinct definition, saying business intelligence strategies "Leverage knowledge to close the loop between information and action."

Henry Monis, Vice President for Applications and Information Access Research at IDC, takes a user-centric view, defining business intelligence software as technology that enables nontechnicaL personnel to view reports, run queries, and conduct analytic exploration in real time. While these analysts might differ in their definitions, they all agree that business intelligence involves sorting, summarizing, and drilling down into successive layers of information to gain insight. This "slicing and dicing," as it is commonly called, is valuable for answering specific questions and discerning trends in the data. But taken as a whole, it works by rearranging or magnifying the user's view of the data at hand. It is difficult to obtain information that is not in an initial report unless you launch a new report that takes you down a different path of inquiry.

A buyer for a department store, for example, knows she wants to analyze inventory data by location, so developers will optimize the database to sort and summarize product information by region, state, city, and store. The buyer can then query the monthly sales data to determine how well certain products are selling in each store, or run reports against the customer database to verify whether last month's promotions influenced customer behaviour. But in each case, she is following a linear path of analysis that was defined in advance and which the buyer can never deviate from.

Similarly, a sales manager could use business intelligence software to create a summary report that lists total sales by region, then drill down to determine sales by office, sales by rep, and so forth. He could also submit queries to analyze the data, How is the Boston office doing within the North Atlantic region as a whole? Who is the top performing sales rep at that office? How does that rep's performance this year compare with last year's? These are the types of questions that business intelligence systems can answer, as users generate reports and submit queries to analyze data based on any variable within a predefined data set.

As powerful and useful as these tools have become, they have their limitations. Traditional business intelligence involves lots of flexibility, and even the ability to access multi pie databases. But it is limited to the information in the data mart and defined by the pathways established by the developers.

For example, a sales report might reveal all activity within a designated period, including a cancellation by ACME Corporation. But there is generally no direct way to sidestep into related information to try and figure out why the order was cancelled - unless the information is already in the reporting database, and unless the user re-enters all the query parameters.

Perhaps the cancellation was the result of difficulty with another product, or a customer support issue that was not satisfactorily resolved. But because the sales manager is working with the sales data, not the customer support data, there is no direct path to thc answers he needs. In order to research these related issues, he would have to launch a new report and re-enter all of the variables and sort criteria to locate the customer, product set, and time period.

Exploring Overlapping Dimensions

For decades, the 20th-century Dutch artist M.C. Escher has fascinated viewers with his unique perspective on spaces and dimensions. In Escher's graphical world, common geometric shapes give rise to horses, swans, and other creatures. Fish transmute into birds and fly off of the page, stretching the boundaries of reality. Mud puddles reflect the grandeur of an overarching forest, and buildings fold in on themselves to reveal new perspectives on familiar objects. Knowledge Mapping brings a similar perspective to reporting, freeing users from static, predetermined, or linear paths of analysis and opening doors to cross-domain analysis to obtain information. The boundaries are not limited only to the data within a given report, but to topics and data sets that are logically related even if they are not physically linked--like Escher's reflections on the surface of a pond.

The sales manager, for example, knows that ACME canceled its order at the last minute. But in order to figure out why, using traditional business intelligence tools, he would have to launch a different set of reports to investigate open issues with the customer support team. What if the sales manager could simply select the detail line that reveals the ACME cancellation and drag it over to another report that looks at customer support information?

This is precisely what Knowledge Mapping enables. Rather than firing up a completely different report or path of inquiry, the sales manager could direct a query into the customer support domain, all while preserving the context about the customer, product mix, and time period. He would not have to re-enter all those variables; the drilling process carries its own form of intelligence, allowing the resulting information to be determined based on preselected variables, scripts, and the report or URL that the variable is "dropped" on.

Expanding Circles--The Ripple Effect 

With Knowledge Mapping, reporting and analysis activities are not limited to data within a previously defined data mart, but to topics and data sets that are logically related. This saves a significant amount of time in generating new reports and new information. Users can preserve the context and move it intact into new types of decision-making processes. Let's return to our sales example. Perhaps, as the sales manager's thought process develops, he decides he wants to know more about the financial status of ACME Corporation. He wonders if the cancellation was not merely caused by a customer support snafu, but by unrelated business dynamics at the company. Simply by dragging the customer name from the report onto a hot link, he can use the Internet to pass a new search string to the Securities and Exchange Commission, or to ACME's own Web site, gathering data from a public information portal.

Once URLs are enabled within a Knowledge Mapping environment, users can use the context of a report to qualify a search for corporate information on Yahool, for example, or for jobs on Monster.com. The potential is limitless, with the entire mosaic of predefined parameters and variables carried into each new context. Perhaps you have created a report on second-quarter 2001 sales for all companies in the aircraft engine business. General Electric is further down the list than you expected, and you want to know more about GE'S financials during that period. You can simply click on General Electric, then drag that item to Yahoo! to discover GE's stock price or request a 10k report. You would not have to requalify the search for the company and time period; that context would be automatically maintained in the Internet search.

Setting Up the Knowledge Mapping Environment

For existing WebFOCUS customers, Knowledge Mapping is easy to enable. Developers simply pick out common data elements that users might wish to see as links, and WebFOCUS does the rest. In the simplest instance, Knowledge Mapping can be used as a way to branch logic.. for example, giving users two options when they click on a hot spot rather than just one. Because WebFOCUS was designed from the outset for browser-driven, Web-based computing, all search variables and formatting options are generated at runtime. This makes it simple to specify navigation options on the fly and to present users with a list of subjects to drill through at each designated hot spot, an shown in Figure 1.

Source reports give users a wide array of drilling options that they never had before. All fields in the source report are "hot," which means they can be dragged to a target. When the user's cursor hovers over each field, a context menu containing drill-down options appears. These hot fields can be dragged either to another report or to a URL. Dragging a source value on top of a target passes the dragged value to the target, minimizing the number of clicks to obtain information.

For example, in the Knowledge Map report pictured in Figure 1. we can look at DVD sales information. Dragging the DVD value under the Product category to the Sales by Store target bypasses both "eMart" and "St. Louis." The resulting target report displays sales for DVDs in all eMarts located in St. Louis. With comparatively little effort, users can suddenly combine information that - from a database-design standpoint--was formerly unrelated. For more information on the Knowledge Mapping development process, consult the manual "Creating Knowledge Map Reports."

Picking Up Where CLAP Leaves Off

How does Knowledge Mapping differ from online-analytical processing (OLAP) tools that conduct multi-dimensional data analysis? As with most business intelligence systems, OLAP tools work well within a carefully constructed domain, or cube. Typically, OLAP users conduct analysis against a subject-oriented data mart. There may be dimensions within that subject, but the context is established ahead of time by the developers. Once users try to traverse multiple cubes or send queries to data marts that were not designed to work together, the paradigm breaks down very quickly.

Multi-dimensional analysis primarily involves functions called roll-up and pivot. Roll-up is used to summarize data based on greater or lesser levels of detail. Pivot involves reordering and rearranging information to reveal correlations in the data. In both cases, users conduct their analyses against very specific data models and cubes. It is difficult, from a system-design standpoint, to relate variables and data sources that are not physically related.

Knowledge Mapping is analogous to simultaneously reporting against multiple data marts, multiple cubes, or multiple subjects- without the need for cubes or data marts. One of your paths of analysis could be a data mart of inventory information. Another could be a data mart concerned with supplier contracts. Yet a third could be an operational data source such as financial data from an ERP system, or a flat file on a mainframe computer. So long as developers have specified the pertinent sources and targets in the Knowledge Map, users can quickly hop from subject to subject, data mart to data mart, database to database, dragging contextual information from one query to the next.

The WebFOCUS Architecture - Designed for Contextual Analysis

Part of this flexibility stems from the inherent reach of the WebFOCUS environment, which is able to access more than 85 types of data on 35 computer platforms. Knowledge Mapping complements and extends the natural information access capabilities of WebFOCUS by making it easier to combine information from these disparate data sources in a natural and intuitive fashion. The WebFOCUS architecture is not built on preset databases and cubes. It has always been open to multiple data sources. Knowledge Mapping extends this broad deployment architecture even further.

As with all WebFOCUS technology, Knowledge Mapping works in a thin-client environment, in which no Java applets or other controls arc required on the client. It utilizes the latest Web-based technologies such as JavaScript, Cascading Style Sheets, XML, and XSLT, Thc Mapping tool is available in the Windows version of the WebFOCUS environment in conjunction with Web FOCUS User Administration Services.

Serendipitous Discovery -- An Infinite Perspective Real-time information delivery software is reaching a wider audience outside the organization as companies provide their suppliers, customers, shareholders, and other stakeholders with access to data through extranets. Due to its server-based architecture, WebFOCUS applications can be deployed to thousands of users and more, either via the public Internet, a virtual private network (VPN), or on a corporate intranet.

Reflecting Escher' s drive to extend the boundaries of visual domains, Knowledge Mapping reveals new worlds of perspective and discovery. Users are no longer hindered by rigid cubes and linear reporting structures. This allows them to easily pursue new relationships in the data and follow more creative avenues of inquiry, streamlining the reporting process while opening doors to serendipitous investigation. Knowledge Mapping supplies an exceptionally useful navigational tool, a quicker way to obtain real-time information. Rather than drilling down into a certain data set, then starting over to drill into a new data set, users can jump from one page or report to another very easily--keeping the context of their searches intact. They can follow their logical instincts to gather information, preserving the information they have gathered within each new search or avenue of inquiry.

As IDC's Henry Morris says, the issue isn't really how to turn raw data into information, but how to improve the quality of decision making. "The question is not simply how. to provide technology, but how to apply the technology to the business situations where it becomes actionable. That's really the only way that companies are going to be able to show a measurable return from investments in business intelligence technology."

www.informationbuilderrs.com