Collection-Level Subject Access in Aggregations of Digital Collections: Metadata Application and Use

Problems in subject access to information organization systems have been under investigation for a long time. Focusing on item-level information discovery and access, researchers have identified a range of subject access problems, including quality and application of metadata, as well as the complexity of user knowledge required for successful subject exploration. While aggregations of digital collections built in the United States and abroad generate collection-level metadata of various levels of granularity and richness, no research has yet focused on the role of collection-level metadata in user interaction with these aggregations. This dissertation research sought to bridge this gap by answering the question “How does collection-level metadata mediate scholarly subject access to aggregated digital collections?” This goal was achieved using three research methods: • in-depth comparative content analysis of collection-level metadata in three large-scale aggregations of cultural heritage digital collections: Opening History, American Memory, and The European Library • transaction log analysis of user interactions, with Opening History, and • interview and observation data on academic historians interacting with two aggregations: Opening History and American Memory. It was found that subject-based resource discovery is significantly influenced by collection-level metadata richness. The richness includes such components as: 1) describing collection’s subject matter with mutually-complementary values in different metadata fields, and 2) a variety of collection properties/characteristics encoded in the free-text Description field, including types and genres of objects in a digital collection, as well as topical, geographic and temporal coverage are the most consistently represented collection characteristics in free-text Description fields. Analysis of user interactions with aggregations of digital collections yields a number of interesting findings. Item-level user interactions were found to occur more often than collection-level interactions. Collection browse is initiated more often than search, while subject browse (topical and geographic) is used most often. Majority of collection search queries fall within FRBR Group 3 categories: object, concept, and place. Significantly more object, concept, and corporate body searches and less individual person, event and class of persons searches were observed in collection searches than in item searches. While collection search is most often satisfied by Description and/or Subjects collection metadata fields, it would not retrieve a significant proportion of collection records without controlled-vocabulary subject metadata (Temporal Coverage, Geographic Coverage, Subjects, and Objects), and free-text metadata (the Description field). Observation data shows that collection metadata records in Opening History and American Memory aggregations are often viewed. Transaction log data show a high level of engagement with collection metadata records in Opening History, with the total page views for collections more than 4 times greater than item page views. Scholars observed viewing collection records valued descriptive information on provenance, collection size, types of objects, subjects, geographic coverage, and temporal coverage information. They also considered the structured display of collection metadata in Opening History more useful than the alternative approach taken by other aggregations, such as American Memory, which displays only the free-text Description field to the end-user. The results extend the understanding of the value of collection-level subject metadata, particularly free-text metadata, for the scholarly users of aggregations of digital collections. The analysis of the collection metadata created by three large-scale aggregations provides a better understanding of collection-level metadata application patterns and suggests best practices. This dissertation is also the first empirical research contribution to test the FRBR model as a conceptual and analytic framework for studying collection-level subject access.

GPRM: a high performance programming framework for manycore processors

Processors with large numbers of cores are becoming commonplace. In order to utilise the available resources in such systems, the programming paradigm has to move towards increased parallelism. However, increased parallelism does not necessarily lead to better performance. Parallel programming models have to provide not only flexible ways of defining parallel tasks, but also efficient methods to manage the created tasks. Moreover, in a general-purpose system, applications residing in the system compete for the shared resources. Thread and task scheduling in such a multiprogrammed multithreaded environment is a significant challenge. In this thesis, we introduce a new task-based parallel reduction model, called the Glasgow Parallel Reduction Machine (GPRM). Our main objective is to provide high performance while maintaining ease of programming. GPRM supports native parallelism; it provides a modular way of expressing parallel tasks and the communication patterns between them. Compiling a GPRM program results in an Intermediate Representation (IR) containing useful information about tasks, their dependencies, as well as the initial mapping information. This compile-time information helps reduce the overhead of runtime task scheduling and is key to high performance. Generally speaking, the granularity and the number of tasks are major factors in achieving high performance. These factors are even more important in the case of GPRM, as it is highly dependent on tasks, rather than threads. We use three basic benchmarks to provide a detailed comparison of GPRM with Intel OpenMP, Cilk Plus, and Threading Building Blocks (TBB) on the Intel Xeon Phi, and with GNU OpenMP on the Tilera TILEPro64. GPRM shows superior performance in almost all cases, only by controlling the number of tasks. GPRM also provides a low-overhead mechanism, called “Global Sharing”, which improves performance in multiprogramming situations. We use OpenMP, as the most popular model for shared-memory parallel programming as the main GPRM competitor for solving three well-known problems on both platforms: LU factorisation of Sparse Matrices, Image Convolution, and Linked List Processing. We focus on proposing solutions that best fit into the GPRM’s model of execution. GPRM outperforms OpenMP in all cases on the TILEPro64. On the Xeon Phi, our solution for the LU Factorisation results in notable performance improvement for sparse matrices with large numbers of small blocks. We investigate the overhead of GPRM’s task creation and distribution for very short computations using the Image Convolution benchmark. We show that this overhead can be mitigated by combining smaller tasks into larger ones. As a result, GPRM can outperform OpenMP for convolving large 2D matrices on the Xeon Phi. Finally, we demonstrate that our parallel worksharing construct provides an efficient solution for Linked List processing and performs better than OpenMP implementations on the Xeon Phi. The results are very promising, as they verify that our parallel programming framework for manycore processors is flexible and scalable, and can provide high performance without sacrificing productivity.

Laying out the future of final-form digital documents

It is just over 20 years since Adobe's PostScript opened a new era in digital documents. PostScript allows most details of rendering to be hidden within the imaging device itself, while providing a rich set of primitives enabling document engineers to think of final-form rendering as being just a sophisticated exercise in computer graphics. The refinement of the PostScript model into PDF has been amazingly successful in creating a near-universal interchange format for complex and graphically rich digital documents but the PDF format itself is neither easy to create nor to amend. In the meantime a whole new world of digital documents has sprung up centred around XML-based technologies. The most widespread example is XHTML (with optional CSS styling) but more recently we have seen Scalable Vector Graphics (SVG) emerge as an XML-based, low-level, rendering language with PostScript-compatible rendering semantics. This paper surveys graphically-rich final-form rendering technologies and asks how flexible they can be in allowing adjustments to be made to final appearance without the need for regenerating a whole page or an entire document. Particular attention is focused on the relative merits of SVG and PDF in this regard and on the desirability, in any document layout language, of being able to manipulate the graphic properties of document components parametrically, and at a level of granularity smaller than an entire page.

Custom Analytics with Google Tag Manager: Assessing Usage Statistics on the MD-SOAR Platform

As usage metrics continue to attain an increasingly central role in library system assessment and analysis, librarians tasked with system selection, implementation, and support are driven to identify metric approaches that simultaneously require less technical complexity and greater levels of data granularity. Such approaches allow systems librarians to present evidence-based claims of platform usage behaviors while reducing the resources necessary to collect such information, thereby representing a novel approach to real-time user analysis as well as dual benefit in active and preventative cost reduction. As part of the DSpace implementation for the MD SOAR initiative, the Consortial Library Application Support (CLAS) division has begun test implementation of the Google Tag Manager analytic system in an attempt to collect custom analytical dimensions to track author- and university-specific download behaviors. Building on the work of Conrad , CLAS seeks to demonstrate that the GTM approach to custom analytics provides both granular metadata-based usage statistics in an approach that will prove extensible for additional statistical gathering in the future. This poster will discuss the methodology used to develop these custom tag approaches, the benefits of using the GTM model, and the risks and benefits associated with further implementation.

Behavioral and neuroimaging research of reading: A case of Japanese

Behavioral studies showed that AS, an English-Japanese bilingual was a skilled reader in Japanese but was a phonological dyslexic in English. This behavioral dissociation was accounted for by the Hypothesis of Transparency and Granularity postulated by Wydell & Butterworth. However, a neuroimaging study using MEG (magnetoencephalography) revealed that AS has the same functional deficit in the left superior temporal gyrus (STG). This paper therefore offers an answer to this intriguing discrepancy between the behavioral dissociation and the neural unity in AS by reviewing existing behavioral and neuroimaging studies in alphabetic languages such as English, Finnish, French, and Italian, and nonalphabetic languages such as Japanese and Chinese.

Automated morphometry provides accurate and reproducible virtual staging of liver fibrosis in chronic hepatitis C

International audience

Big Code Applications and Approaches

The availability of a huge amount of source code from code archives and open-source projects opens up the possibility to merge machine learning, programming languages, and software engineering research fields. This area is often referred to as Big Code where programming languages are treated instead of natural languages while different features and patterns of code can be exploited to perform many useful tasks and build supportive tools. Among all the possible applications which can be developed within the area of Big Code, the work presented in this research thesis mainly focuses on two particular tasks: the Programming Language Identification (PLI) and the Software Defect Prediction (SDP) for source codes. Programming language identification is commonly needed in program comprehension and it is usually performed directly by developers. However, when it comes at big scales, such as in widely used archives (GitHub, Software Heritage), automation of this task is desirable. To accomplish this aim, the problem is analyzed from different points of view (text and image-based learning approaches) and different models are created paying particular attention to their scalability. Software defect prediction is a fundamental step in software development for improving quality and assuring the reliability of software products. In the past, defects were searched by manual inspection or using automatic static and dynamic analyzers. Now, the automation of this task can be tackled using learning approaches that can speed up and improve related procedures. Here, two models have been built and analyzed to detect some of the commonest bugs and errors at different code granularity levels (file and method levels). Exploited data and models’ architectures are analyzed and described in detail. Quantitative and qualitative results are reported for both PLI and SDP tasks while differences and similarities concerning other related works are discussed.

Optimizing AI at the Edge: from network topology design to MCU deployment

The first topic analyzed in the thesis will be Neural Architecture Search (NAS). I will focus on two different tools that I developed, one to optimize the architecture of Temporal Convolutional Networks (TCNs), a convolutional model for time-series processing that has recently emerged, and one to optimize the data precision of tensors inside CNNs. The first NAS proposed explicitly targets the optimization of the most peculiar architectural parameters of TCNs, namely dilation, receptive field, and the number of features in each layer. Note that this is the first NAS that explicitly targets these networks. The second NAS proposed instead focuses on finding the most efficient data format for a target CNN, with the granularity of the layer filter. Note that applying these two NASes in sequence allows an "application designer" to minimize the structure of the neural network employed, minimizing the number of operations or the memory usage of the network. After that, the second topic described is the optimization of neural network deployment on edge devices. Importantly, exploiting edge platforms' scarce resources is critical for NN efficient execution on MCUs. To do so, I will introduce DORY (Deployment Oriented to memoRY) -- an automatic tool to deploy CNNs on low-cost MCUs. DORY, in different steps, can manage different levels of memory inside the MCU automatically, offload the computation workload (i.e., the different layers of a neural network) to dedicated hardware accelerators, and automatically generates ANSI C code that orchestrates off- and on-chip transfers with the computation phases. On top of this, I will introduce two optimized computation libraries that DORY can exploit to deploy TCNs and Transformers on edge efficiently. I conclude the thesis with two different applications on bio-signal analysis, i.e., heart rate tracking and sEMG-based gesture recognition.

The cosmic ray tagger of the ICARUS detector: construction, commissioning and performances in the first physics run

The Short Baseline Neutrino Program at Fermilab aims to confirm or definitely rule out the existence of sterile neutrinos at the eV mass scale. The program will perform the most sensitive search in both the nue appearance and numu disappearance channels along the Booster Neutrino Beamline. The far detector, ICARUS-T600, is a high-granularity Liquid Argon Time Projection Chamber located at 600 m from the Booster neutrino source and at shallow depth, thus exposed to a large flux of cosmic particles. Additionally, ICARUS is located 6 degrees off axis with respect to the Neutrino beam from the Main Injector. This thesis presents the construction, installation and commissioning of the ICARUS Cosmic Ray Tagger system, providing a 4 pi coverage of the active liquid argon volume. By exploiting only the precise nanosecond scale synchronization of the cosmic tagger and the PMT optical flashes it is possible to determine if an event was likely triggered by a cosmic particle. The results show that using the Top Cosmic Ray Tagger alone a conservative rejection larger than 65% of the cosmic induced background can be achieved. Additionally, by requiring the absence of hits in the whole cosmic tagger system it is possible to perform a pre-selection of contained neutrino events ahead of the full event reconstruction.