K-Praxis - Topic: Unstructured Information Management - Decision Support Systems and Unstructured Information-Print

Decision Support Systems and Unstructured Information
June 24, 2003

A quick survey of Decision Support tools and solutions vendors reveal that majority of them can be easily confused with Business Intelligence solutions. K-Praxis looks at the constituents of a cutting-edge decision support system and finds out how unstructured textual data and information can play a major role in developing such solutions...read on....

Introduction to Decision Support Systems

This is how a classic AI-based Decision Support System could be defined:

"Decision support is a broad class of applications for artificial intelligence software. There are many situations when humans would prefer machines, particularly computers, to either automatically assist them in making decisions, or actually make and act on a decision. There are a wide range of non-AI decision support systems such as most of the process control systems successfully running chemical plants and power plants and the like under steady state conditions. However, whenever situations become more complex for example, in chemical plants that don't run under steady state, or in businesses when both humans and equipment are interacting intelligent decision support is required. That can only be provided by automatic decision support software using artificial intelligence techniques. SHAI has created a wide range of decision support applications that provide examples of such situations. Synonym:intelligent decision support." (http://www.shai.com/ai_general/glossary.htm)

So a decision support system should be able to process various information inputs, tag them if required, classify, analyze and "understand" these inputs and make intelligent predictions or decisions which would help humans in diagnosis, planning, and design. Logically, decision support systems will match the "understood" actionable information to other strata of inputs it has acquired through past past experiences (abstracted and encoded in rules and heuristic rules), statistical computational models, or rules generated by experts in the field.

Decision Support Systems and Unstructured data

Decision support systems use various methodologies such as decision trees; Bayesian belief networks; Monte Carlo simulation methods; analytic hierarchy process; case-based reasoning etc. Many of the methods are dependent on structured numerical data and do not take into account textual unstructured data, hence the decision analysis provided by these systems is limited to quantifiable business data and any attempts to build a complete and comprehensive decision analysis systems eventually will require incorporating text-base approaches and methodologies to offer realistic and broader decision analysis solutions.

Decision Support: Untapped Potential

As enterprise software vendors try to consolidate their offerings (seen recently in Oracle-Peoplesoft tussle) and businesses, on other hand, opt for more niche, need-driven software to revitalize their business performance and competitive positioning, decision-analysis and business intelligence solutions could be looked at as the next generation of applications catering to that "unfulfilled promise" offered by enterprise software vendors in B2B days! But yet again, the promise of decision analysis will remain unfulfilled if technology vendors over look the need to integrate with unstructured data.

Decision Support: Vendors integrating both structured and unstructured data

K-Praxis survey revealed that among the decision support and analytics vendors (Crystal Decisions, SAS, Business Objects, Cognos, MicroStrategy, Decisioneering, Matrix Cognition, Fair, Isaac & Company and Megaputer), a few vendors like Fair Iasaac and Megaputer have already started looking at the ways of integrating structured and unstructured data.

	Making sense of online textual information and information management technologies

	Decision Support Systems and Unstructured Information June 24, 2003 A quick survey of Decision Support tools and solutions vendors reveal that majority of them can be easily confused with Business Intelligence solutions. K-Praxis looks at the constituents of a cutting-edge decision support system and finds out how unstructured textual data and information can play a major role in developing such solutions...read on.... Introduction to Decision Support Systems This is how a classic AI-based Decision Support System could be defined: "Decision support is a broad class of applications for artificial intelligence software. There are many situations when humans would prefer machines, particularly computers, to either automatically assist them in making decisions, or actually make and act on a decision. There are a wide range of non-AI decision support systems such as most of the process control systems successfully running chemical plants and power plants and the like under steady state conditions. However, whenever situations become more complex for example, in chemical plants that don't run under steady state, or in businesses when both humans and equipment are interacting intelligent decision support is required. That can only be provided by automatic decision support software using artificial intelligence techniques. SHAI has created a wide range of decision support applications that provide examples of such situations. Synonym:intelligent decision support." (http://www.shai.com/ai_general/glossary.htm) So a decision support system should be able to process various information inputs, tag them if required, classify, analyze and "understand" these inputs and make intelligent predictions or decisions which would help humans in diagnosis, planning, and design. Logically, decision support systems will match the "understood" actionable information to other strata of inputs it has acquired through past past experiences (abstracted and encoded in rules and heuristic rules), statistical computational models, or rules generated by experts in the field. Decision Support Systems and Unstructured data Decision support systems use various methodologies such as decision trees; Bayesian belief networks; Monte Carlo simulation methods; analytic hierarchy process; case-based reasoning etc. Many of the methods are dependent on structured numerical data and do not take into account textual unstructured data, hence the decision analysis provided by these systems is limited to quantifiable business data and any attempts to build a complete and comprehensive decision analysis systems eventually will require incorporating text-base approaches and methodologies to offer realistic and broader decision analysis solutions. Decision Support: Untapped Potential As enterprise software vendors try to consolidate their offerings (seen recently in Oracle-Peoplesoft tussle) and businesses, on other hand, opt for more niche, need-driven software to revitalize their business performance and competitive positioning, decision-analysis and business intelligence solutions could be looked at as the next generation of applications catering to that "unfulfilled promise" offered by enterprise software vendors in B2B days! But yet again, the promise of decision analysis will remain unfulfilled if technology vendors over look the need to integrate with unstructured data. Decision Support: Vendors integrating both structured and unstructured data K-Praxis survey revealed that among the decision support and analytics vendors (Crystal Decisions, SAS, Business Objects, Cognos, MicroStrategy, Decisioneering, Matrix Cognition, Fair, Isaac & Company and Megaputer), a few vendors like Fair Iasaac and Megaputer have already started looking at the ways of integrating structured and unstructured data.