Tapati Bandopadhyay

A Generic Model of Self-Incrementing Knowledge-Based Decision Support System Using the Bolzmann Machine

Knowledge-driven decision support systems (DSS) rely significantly on the currency of it’s back-end knowledge-base for improved quality of support that it can provide to the decision making process. In this paper, a model has been developed and presented to support this purpose i.e. making the knowledge base of a knowledge-driven DSS as self-incrementing. First, the necessity of such a model for maintaining knowledge currency and the importance of knowledge currency in the context of DSS supporting operational decision processes, are discussed. Then, a generic model is presented using customer-email as input knowledge sources, frames as a knowledge representation scheme, and Bolzmann machine as the seif-incrementing mechanism. The model can be further extended or fine-tuned both in terms of the input options and process options. The input options can include other types of knowledge inputs with varying degrees of structuredness. The process options may include other algorithms and machine learning processes.

A process model for preliminary classification of knowledge sources using dependency graphs and incremental computing concepts

Knowledge extraction from semi-structured or unstructured documents and texts can be handled by various approaches and viewpoints, including theoretical or mathematical constructions and matrix algebra applications to various ontologies and implementation techniques.

A Data-Driven View of the Evolving IT Infrastructure Technologies and Options

For effective service delivery on any infrastructure and SOA-based environment, data is an essential building block both as inputs as well as in monitoring the output. Data is an integral part of any formation delivery and/or service delivery in an IT-enabled business environment. For the new-age IT infrastructure models with cloud and virtualization technology options and web-based service-delivery models, recording of the characteristic data parameters that define or govern the run-time service delivery environment is crucial both for client businesses and provider organizations to be cost-viable and quality-effective. In this paper, we create a data-driven framework of the new-age IT Infrastructure. Operationally, this framework can be used for IT-user organizations as well as service providers, to track their delivery concerns. Strategically, this framework may serve as a baseline template to plan for a minimal set of parameters recoded as operational data-points from the real-time run-time IT infrastructure operations environment.

Reusable knowledge pattern extraction from peer-to-peer communication elements

Tacit knowledge embedded in various communication elements are highly unstructured, embedded in short message texts, and can be highly contextual thereby rendering generic English language thesaurus-based pattern recognition and extraction mechanisms relatively less useful. However, this embedded tacit knowledge is a significant source of context-specific, technical, problem-solving knowledge. Peer-to-peer communication threads have highly valuable problem-solution knowledge elements embedded in them. If these knowledge elements can be extracted, experiential knowledge in specific domains or technology areas can be significantly enhanced. In this paper, a process model has been designed, with algorithms developed and validated using appropriate examples, for extracting these reusable knowledge patterns from various peer-to-peer communication elements in any organisation or in the social network environment. Using this process model, the knowledge extraction can be partially or completely automated, depending on the context-specificity of the knowledge elements and therefore the requirements of specific thesaurus or term-dictionaries.