There is a tremendous volume of data being generated in today’s world. As organizations around the globe realize the increased importance of their data as being a valuable asset in gaining a competitive edge in a fast-paced and a dynamic business world, more and more attention is being paid to the quality of the data. Advances in the fields of data mining, predictive modeling, text mining, web mining, business intelligence, health care analytics, etc. all depend on clean, accurate data. That one cannot effectively mine data, which is dirty, comes as no surprise. This research is an exploratory study of different domain data sets, addressing the data quality issues specific to each domain, identifying the challenges faced and arriving at techniques or methodologies for measuring and improving the data quality. The primary focus of the research is on the SAR or Search and Rescue dataset, identifying key issues related to data quality therein and developing an algorithm for improving the data quality. SAR missions which are routinely conducted all over the world show a trend of increasing mission costs. Retrospective studies of historic SAR data not only allow for a detailed analysis and understanding of SAR incidents and patterns, but also form the basis for generating probability maps, analytical data models, etc., which allow for an efficient use of valuable SAR resources and their distribution. One of the challenges with regards to the SAR dataset is that the collection process is not perfect. Often, the LKP or the Last Known Position is not known or cannot be arrived at. The goal is to fully or partially geocode the LKP for as many data points as possible, identify those data points where the LKP cannot be geocoded at all, and further highlight the underlying data quality issues. The SAR Algorithm has been developed, which makes use of partial or incomplete information, cleans and validates the data, and further extracts address information from relevant fields to successfully geocode the data. The algorithm improves the geocoding accuracy and has been validated by a set of approaches.

Indiana University-Purdue University Indianapolis (IUPUI) / Derived from the field of art curation, digital provenance is an unforgeable record of a digital object's chain of successive custody and sequence of operations performed on the object. It plays an important role in accessing the trustworthiness of the object, verifying its reliability and conducting audit trails of its lineage. Digital provenance forms an immutable directed acyclic graph (DAG) structure. Since history of an object cannot be changed, once a provenance chain has been created it must be protected in order to guarantee its reliability. Provenance can face attacks against the integrity of records and the confidentiality of user information, making security an important trait required for digital provenance. The digital object and its associated provenance can have different security requirements, and this makes the security of provenance different from that of traditional data. Research on digital provenance has primarily focused on provenance generation, storage and management frameworks in different fields. Security of digital provenance has also gained attention in recent years, particularly as more and more data is migrated in cloud environments which are distributed and are not under the complete control of data owners. However, there still lacks a viable secure digital provenance scheme which can provide comprehensive security for digital provenance, particularly for generic and dynamic ones. In this work, we address two important aspects of secure digital provenance that have not been investigated thoroughly in existing works: 1) capturing the DAG structure of provenance and 2) supporting dynamic information sharing. We propose a scheme that uses signature-based mutual agreements between successive users to clearly delineate the transition of responsibility of the digital object as it is passed along the chain of users. In addition to preserving the properties of confidentiality, immutability and availability for a digital provenance chain, it supports the representation of DAG structures of provenance. Our scheme supports dynamic information sharing scenarios where the sequence of users who have custody of the document is not predetermined. Security analysis and empirical results indicate that our scheme improves the security of the typical secure provenance schemes with comparable performance.