Browsing by Subject "quality control"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
Item Harvesting Quality: Evaluating Metadata for Digital Collections(2014-03-25) Biswas, Paromita; Western Carolina UniversityMetadata creation practices for digital library projects vary widely amongst libraries. Digital library projects often have to deal with multiple metadata creators, new formats and resources, and dynamic metadata standards for different communities (Park & Tosaka, 2010). As a result while accuracy and consistency in metadata are prioritized by field practitioners, metadata records created for specific digital projects may lack the quality needed to support successful end-user resource discovery and access. Park and Tosaka’s survey of metadata quality control in digital repositories and collections reveal that digital repositories often rely on periodic sampling or peer review of original metadata records as mechanisms for quality assurance (Park & Tosaka, 2010). This poster proposal presents another means of running quality checks on metadata created for digital projects based on Hunter Library’s experience with the WorldCat Digital Collection Gateway tool used for harvesting metadata for digital collections into WorldCat. Hunter Library’s digital collections are described using Dublin Core in Contentdm and the Library has recently started harvesting its collections into WorldCat using Gateway. During harvesting the Gateway, by default, places the names of “creators” and “contributors” recorded in separate fields in the local metadata environment into one broad “Author” field for WorldCat users. A cursory review of this “Author” field in WorldCat for several harvested items from one of the library’s collections revealed an unexpected presence of corporate body names alongside personal names. Consequently this led to an evaluation of how the “creator” and “contributor” fields had been used in that collection. The “Frequency Analysis” feature in Gateway proved to be particularly useful in this evaluation since it provided a breakdown of each field in a particular collection by the values used in that field and the number of times they had been used. For example, a high frequency usage of a particular name indicated that the usage had not been a random mistake but had been consistent. A subsequent analysis of the library’s digital collections’ metadata using “Frequency Analysis” revealed that for some collections, the “contributor” field had been used to record entities whose roles, in relation to the item described, spanned from publisher, printer, editor, or recipient of letter. However, the library’s then current metadata schema had limited the definition of the “contributor” field to entities who had a direct but secondary role in the creation of an item like editors or illustrators. This discrepancy between the library’s metadata schema and the usage of the “contributor” field led to a redefinition of the role of the “contributor.” The schema now incorporates the plethora of roles that “contributors” could have in relation to an item and recommends that the role of each “contributor” be explained in the “description” field to account for the diversity of roles. Updating of the schema has thus promoted consistency in recording the “contributor” field across the library’s digital collections while also possibly benefitting users searching for an item by the various names associated with it.Item Identifying nonlinear variaiton patterns in multivariate manufacturing processes(Texas A&M University, 2005-02-17) Zhang, FengThis dissertation develops a set of nonlinear variation pattern identification methods that are intended to aid in diagnosing the root causes of product variability in complex manufacturing processes, in which large amounts of high dimensional in-process measurement data are collected for quality control purposes. First, a nonlinear variation pattern model is presented to generically represent a single nonlinear variation pattern that results from a single underlying root cause, the nature of which is unknown a priori. We propose a modified version of a principal curve estimation algorithm for identifying the variation pattern. Principal curve analysis is a nonlinear generalization of principal components analysis (PCA) that lends itself well to interpretation and also has theoretically rich underpinnings. The principal curve modification involves a dimensionality reduction step that is intended to improve estimation accuracy by reducing noise and improving the robustness of the algorithm with the high-dimensional data typically encountered in manufacturing. An effective visualization technique is also developed to help interpret the identified nonlinear variation pattern and aid in root cause identification and elimination. To further improve estimation robustness and accuracy and reduce computational expense, we propose a local PCA based polygonal line algorithm to identify the nonlinear patterns. We also develop an approach for separating and identifying the effects of multiple nonlinear variation patterns that are present simultaneously in the measurement data. This approach utilizes higher order cumulants and pairwise distance based clustering to separate the patterns and borrows from techniques that are used in linear blind source separation. With the groundwork laid for a versatile flexible and powerful nonlinear variation pattern modeling and identification framework, applications in autobody assembly and stamping processes are investigated. The pattern identification algorithms, together with the proposed visualization approach, provides an effective tool to aid in understanding the nature of the root causes of variation that affect a manufacturing process.Item Tropical Cyclone Data Assimilation: Experiments with a Coupled Global-Limited-Area Analysis System(2014-04-22) Holt, ChristinaThis study investigates the benefits of employing a limited-area data assimilation (DA) system to enhance lower-resolution global analyses in the Northwest Pacific tropical cyclone (TC) basin. Numerical experiments are carried out with a global analysis system at horizontal resolution T62 and a limited-area analysis system at resolutions from 200 km to 36 km. The global and limited-area DA systems, which are both based on the Local Ensemble Transform Kalman Filter algorithm, are implemented using a unique configuration, in which the global DA system provides information about the large-scale analysis and background uncertainty to the limited-area DA system. In experiments that address the global-to-limited-area resolution ratio, the limited-area analyses of the storm locations for experiments in which the ratio is 1:2 are, on average, more accurate than those from the global analyses. Increasing the resolution of the limited-area system beyond 100 km adds little direct benefit to the analysis of position or intensity, although 48 km analyses reduce boundary effects of coupling the models and may benefit analyses in which observations with larger representativeness error are assimilated. Two factors contribute to the higher accuracy of the limited-area analyses. First, the limited-area system improves the accuracy of the location estimates for strong storms, which is introduced when the background is updated by the global assimilation. Second, it improves the accuracy of the background estimate of the storm locations for moderate and weak storms. Improvements in the steering flow analysis due to increased resolution are modest and short-lived in the forecasts. Limited-area track forecasts are more accurate, on average, than global forecasts, independently of the strength of the storms up to five days. This forecast improvement is due to the more accurate analysis of the initial position of storms and the better representation of the interactions between the storms and their immediate environment. Experiments that test the treatment and quality control (QC) methods of TC observations show that significant gainful improvements can be achieved in the analyses and forecasts of TCs when observations with large representativeness error are not discarded in the online QC procedure. These experiments examine the impact of assimilating TCVitals SLP, QuikSCAT 10 m wind components, and reconnaissance dropsondes alongside the conventional observations assimilated by NCEP in real time. Implementing a Combined method that clips the special TC observations via Huberization when multiple observation types are unavailable, and keeping the TCVital observation when other special observations are present, showed significant systematic improvements for strong and moderate storm analyses and forecasts.