IT Performance: Overview and Impetus
|
|
Information Technology Revolution
|
Commerce at the beginning of the twenty-first century is seeing
an information technology revolution driven by software
innovation, by network systems, and by hardware advances.
Assessing quality in each of these areas draws on statistical
models that are developed expressly for this purpose. At NIST,
statistical modeling and research addresses these areas with
major collaborative efforts in software testing, in information
retrieval, in network performance assessment, and in ongoing
standards for electronics.
|
|
Testing Computer Software
|
Testing software often requires development of stochastic models
for failures, where failure points are determined by the specs
for the software, whether this be computer graphics meta file,
for Java, for computer-aided design software, or for statistical
computation software. Then statistical methods for inference
must be derived for tests of conformance, measures of
reliability, estimates of failure rates and of variability
according to failure mode.
|
|
Image Recognition
|
In the case of testing algorithms for feature recognition, such
as the identification of an individual person based on a poor or
partial image, the first statistical issue is the design of the
testing procedure using an accumulated data set of images. The
goal of assessing the performance of an algorithm requires
development of an overall peformance measure and also
construction of indicators of specific algorithm weaknesses
together with statistically sound measures of uncertainty
associated with each measure.
|
|
Information Retrieval Systems
|
To understand performance measurements on information retrieval
systems, several statistical methods must be applied to discover
whether differences among the retrieval systems are real, and
then to characterize these differences with respect to such
factors as topics searched or collections of documents.
Similarly, network performance testing requires statistical
designs for the utilization of an accumulating database,
although the database characteristics differ sharply.
|
|
Date created: 2/7/2002
Last updated: 2/7/2002
Please email comments on this WWW page to
sedwww@nist.gov.
|
