Extracting Process-Performance Interrelationships from IT Service Contracts

Abstract

Service Level Agreements (SLA) for multi-service IT outsourcing contracts contain voluminous text that often makes it difficult for the customer organization to comprehend the complex interplay of various process elements (e.g. activities, resources, and events) with Key Performance Indicators (KPI). We employ an Ontology-Based Information Extraction (OBIE) approach combined with a rule-base in developing a decision support framework that derives analytic insights about hidden process-performance relationships from SLAs. The framework is instantiated in a prototype system - 'SLA-Miner.' The system aids in decision making by encapsulating formal and semi-formal domain knowledge through an IT service ontology and associated rule-base. Advantages of the system include compact knowledge representation, maintainability, and adaptability to changing business environment. A set of real life SLAs are analyzed to demonstrate the feasibility of SLA-Miner. Also, the quality of information extracted is evaluated by relevant metrics. Results demonstrate the utility of the decision support framework to empower service analysts/managers in the customer organization in identifying KPI interrelationships as well as deficiencies and potential conflicts in the current contract. This knowledge is useful to the customer organization in contract negotiation, process management, and subsequent incentive decisions.