Business Performance Measurement and Management
Measuring and managing the performance of a business is one of the main requirements of the management of any organization. This book introduces new contexts and themes of application and presents emerging research areas related to business performance measurement and management. It draws authors from all around the globe from a variety of functional disciplines, all of whom are working in the field of business performance measurement and management, thus resulting in a variety of perspectives on performance measurement from various functional areas – accounting, finance, economics, marketing, and operations management – in a single volume.
In this chapter, the results of the application of a model enterprise expert search system applied to the tasks introduced at the text retrieval conference (TREC) are presented. Two specific indicators are used in order to treat the lexicon statistically. Calculating lexicon-candidate connection power enables one to reveal definite terms, which are characteristic for a candidate, so this candidate can be found by such terms. Calculating the weight of the lexicon allows the extraction from the whole collection of a small portion of vocabulary, which is identified as significant. The significant lexicon enables one to perform an effective search in thematically specialised knowledge fields. Thus, the search engine minimises the lexicon necessary for answering a query by extracting the most important part from it. The ranking function takes into account term-usage statistics among candidates to raise the role of significant terms in comparison to others, and more noisy ones. In describing the application of the model presented, the possibility of effective expertise retrieval by merging several heuristic ranking metrics into a single weighting model is demonstrated. To enhance the search efficiency, the model is optimised by its free parameters. The shown efficiency is better than that of most TREC participant models. A further efficiency improvement by means of query classification is proposed.