Mass digitization of historical documents is a challenging problem for optical character recognition (OCR) tools. Issues include noisy backgrounds and faded text due to aging, border/marginal noise, bleed-through, skewing, warping, as well as irregular fonts and page layouts. As a result, OCR tools often produce a large number of spurious bounding boxes (BBs) in addition to those that correspond to words in the document. This paper presents an iterative classification algorithm to automatically label BBs (i.e., as text or noise) based on their spatial distribution and geometry. The approach uses a rule-base classifier to generate initial text/noise labels for each BB, followed by an iterative classifier that refines the initial labels by incorporating local information to each BB, its spatial location, shape and size. When evaluated on a dataset containing over 72,000 manually-labeled BBs from 159 historical documents, the algorithm can classify BBs with 0.95 precision and 0.96 recall. Further evaluation on a collection of 6,775 documents with ground-truth transcriptions shows that the algorithm can also be used to predict document quality (0.7 correlation) and improve OCR transcriptions in 85% of the cases.

To mine large digital libraries in humanistically meaningful ways, we need to divide them by genre. This is a task that classification algorithms are well suited to assist, but they need adjustment to address the specific challenges of this domain. Digital libraries pose two problems of scale not usually found in the article datasets used to test these algorithms. 1) Because libraries span several centuries, the genres being identified may change gradually across the time axis. 2) Because volumes are much longer than articles, they tend to be internally heterogeneous, and the classification task also requires segmentation. We describe a multilayered solution that trains hidden Markov models to segment volumes, and uses ensembles of overlapping classifers to address historical change. We demonstrate this on a collection of 469,200 volumes drawn from HathiTrust Digital Library.

Data-intensive flow computing allows efficient processing of large volumes of data otherwise unapproachable. This paper introduces a new semantic-driven data-intensive flow infrastructure which: (1) provides a robust and transparent scalable solution from a laptop to large-scale clusters,(2) creates an unified solution for batch and interactive tasks in high-performance computing environments, and (3) encourages reusing and sharing components. Banking on virtualization and cloud computing techniques the Meandre infrastructure is able to create and dispose Meandre clusters on demand, being transparent to the final user. This paper also presents a prototype of such clustered infrastructure and some results obtained using it.

With the development of petascale computing systems, a long-term effort is needed to educate and train the next generation of researchers. As part of its graduate education component, the Virtual School of Computational Science and Engineering held a summer school in August 2008 entitled "Accelerators for Science and Engineering Applications," providing participants with knowledge and hands-on experience with graphics processing units (GPUs). In this paper, we present visualizations exploring the broad spectrum of summer school applicants and participants. We examine demographic information of the overall applicant pool, accepted and attending applicants, and remote participants, as well as apply hierarchical clustering and rule association techniques to all applicant data. These statistical and data mining analyses demonstrate the wide range of fields of study where research applications can be readily accelerated through the use of massively parallel computing resources.

This paper addresses the problem of making text mining results more comprehensible to humanities scholars, journalists, intelligence analysts, and other researchers, in order to support the analysis of text collections. Our system, FeatureLens1, visualizes a text collection at several levels of granularity and enables users to explore interesting text patterns. The current implementation focuses on frequent itemsets of n-grams, as they capture the repetition of exact or similar expressions in the collection. Users can find meaningful co-occurrences of text patterns by visualizing them within and across documents in the collection. This also permits users to identify the temporal evolution of usage such as increasing, decreasing or sudden appearance of text patterns. The interface could be used to explore other text features as well. Initial studies suggest that FeatureLens helped a literary scholar and 8 users generate new hypotheses and interesting insights using 2 text collections.

This paper presents a demonstration of our recent research and development of the MAIDS system, which mines alarming incidents from data streams, with the following major analysis functions: (1) multi-resolution modeling using a tilted time window framework, (2) multi-dimensional analysis using a stream “data cube” model, (3) online stream classification, (4) online frequent pattern mining, (5) online clustering of data streams, and (6) stream mining visualization.