Return levels of hydrologic droughts under climate change

Developments in the statistical extreme value theory, which allow non-stationary modeling of changes in the frequency and severity of extremes, are explored to analyze changes in return levels of droughts for the Colorado River. The transient future return levels (conditional quantiles) derived from regional drought projections using appropriate extreme value models, are compared with those from observed naturalized streamflows. The time of detection is computed as the time at which significant differences exist between the observed and future extreme drought levels, accounting for the uncertainties in their estimates. Projections from multiple climate model-scenario combinations are considered; no uniform pattern of changes in drought quantiles is observed across all the projections. While some projections indicate shifting to another stationary regime, for many projections which are found to be non-stationary, detection of change in tail quantiles of droughts occurs within the 21st century with no unanimity in the time of detection. Earlier detection is observed in droughts levels of higher probability of exceedance. (C) 2014 Elsevier Ltd. All rights reserved.

Active learning in keyword search-based data integration

The problem of scaling up data integration, such that new sources can be quickly utilized as they are discovered, remains elusive: Global schemas for integrated data are difficult to develop and expand, and schema and record matching techniques are limited by the fact that data and metadata are often under-specified and must be disambiguated by data experts. One promising approach is to avoid using a global schema, and instead to develop keyword search-based data integration-where the system lazily discovers associations enabling it to join together matches to keywords, and return ranked results. The user is expected to understand the data domain and provide feedback about answers' quality. The system generalizes such feedback to learn how to correctly integrate data. A major open challenge is that under this model, the user only sees and offers feedback on a few ``top-'' results: This result set must be carefully selected to include answers of high relevance and answers that are highly informative when feedback is given on them. Existing systems merely focus on predicting relevance, by composing the scores of various schema and record matching algorithms. In this paper, we show how to predict the uncertainty associated with a query result's score, as well as how informative feedback is on a given result. We build upon these foundations to develop an active learning approach to keyword search-based data integration, and we validate the effectiveness of our solution over real data from several very different domains.