Large-scale Digital Libraries Research Articles

Toward meaningful notions of similarity in NLP embedding models

Finding similar words with the help of word embedding models, such as Word2Vec or GloVe, computed on large-scale digital libraries has yielded meaningful results in many cases. However, the underlying notion of similarity has remained ambiguous. In this paper, we examine when exactly similarity values in word embedding models are meaningful. To do so, we analyze the statistical distribution of similarity values systematically, conducting two series of experiments. The first one examines how the distribution of similarity values depends on the different embedding model algorithms and parameters. The second one starts by showing that intuitive similarity thresholds do not exist. We then propose a method stating which similarity values and thresholds actually are meaningful for a given embedding model. Based on these results, we calculate how these thresholds, when taken into account during evaluation, change the evaluation scores of the models in similarity test sets. In more abstract terms, our insights give way to a better understanding of the notion of similarity in embedding models and to more reliable evaluations of such models.

Distribution of date elements and its relationship to the types of digital libraries: Analysing aggregated metadata of a Polish large-scale distributed system

Large-scale distributed digital library systems with aggregated metadata provide platforms for resource discovery and retrieval. For researchers, aggregated metadata offers a potential for big data analysis and exploration of digital knowledge growth. The paper reports the findings of the study that investigated the distribution of the date elements in the metadata aggregated in the Polish Federation of Digital Libraries and related it to the types of libraries. The purpose of this study was to address the gap in research about heterogeneous digital libraries and explore the dynamics of their growth. The authors included timeline characteristics of the development of Polish digital libraries and proposed a new dynamics parameter – resource release interval. They used histograms, which have been grouped according to the organizational and thematic criteria, developed for this study. All charts are characterized by two similar maximum points. Their shapes and ratio have been analysed by both statistical and visual methods. The shape of resource release interval charts revealed characteristic differences for libraries types. The proposed approach, based on time characteristics, is an important step in the development of systematic classification of digital libraries and digitizing institutions. It can be also considered as a new tool in monitoring the dynamics of digital knowledge growth.

User Navigation in Large-Scale Distributed Digital Libraries: The Case of the Digital Public Library of America

ABSTRACTThis article presents the findings of a case study that examined user navigation in a large-scale digital library in the context of academic use. Using the Digital Public Library of America (DPLA) as a case, the study explored user navigation and understanding of a distributed model of large-scale digital libraries. The DPLA model involves two- or three-step navigation pathways. Most study participants could navigate the DPLA distributed, multilayered system effectively. This study found some confusion among the participants when they must move through a three-step process to locate digital objects provided by a metadata aggregator service hub. The study participants also pointed out the advantages of a distributed model in collocating digital resources and connecting users to a wide range of digital libraries.

Digital Libraries that Demonstrate High Levels of Mutual Complementarity in Collection-level Metadata Give a Richer Representation of their Content and Improve Subject Access for Users

A Review of:
 Zavalina, O. L. (2013). Complementarity in subject metadata in large-scale digital libraries: A comparative analysis. Cataloging & Classification Quarterly, 52(1), 77-89. http://dx.doi.org/10.1080/01639374.2013.848316
 
 Abstract
 
 Objective – To determine how well digital library content is represented through free-text and subject headings. Specifically to examine whether a combination of free-text description data and controlled vocabulary is more comprehensive than free-text description data alone in describing digital collections.
 
 Design – Qualitative content analysis and complementarity comparison.
 
 Setting – Three large scale cultural heritage digital libraries: one in Europe and two in the United States of America. 
 
 Methods – The researcher retrieved XML files of complete metadata records for two of the digital libraries, while the third library openly exposed its full metadata. The systematic samples obtained for all three libraries enabled qualitative content analysis to uncover how metadata values relate to each other at the collection level. The researcher retrieved 99 collection-level metadata records in total for analysis. The breakdown was 39, 33, and 27 records per digital library. When comparing metadata in the free-text Description metadata element with data in four controlled vocabulary elements, Subject, Geographic Coverage, Temporal Coverage and Object Type, the researcher observed three types of complementarity: one-way, two-way and multiple-complementarity. The author refers to complementarity as “describing a collection’s subject matter with mutually complementary data values in controlled vocabulary and free-text subject metadata elements” (Zavalina, 2013, p. 77). For example, within a Temporal Coverage metadata element the term “19th century” would complement a Description metadata element “1850–1899” in the same record.
 
 Main Results – The researcher found a high level of one-way complementarity in the metadata of all three digital libraries. This was mostly demonstrated by free-text data in the Description element complemented by data in the controlled vocabulary elements of Subject, Geographic Coverage, Temporal Coverage, and Object Type. Only one library demonstrated a significant proportion (19%) of redundancy between free-text and controlled vocabulary metadata. An example of redundancy found included a repetition of geographic information in both a Description and Geographic Coverage metadata elements.
 
 Conclusion – The author reports high levels of mutual complementarity in the three cultural heritage digital libraries studied. The findings demonstrate that collection-level metadata which includes both free-text and controlled vocabulary is more representative of the intellectual content of the collections and improves subject access for users. The author maintains that there is no standard for collection-level metadata descriptions, and that this research may contribute to best practice guidelines in this area. It is unclear whether the digital libraries studied had written policies in place on how to describe collections and if those policies were adhered to in practice. The author expresses a need for further research to be conducted on collection-level metadata in other domains, such as science and interdisciplinary digital libraries, and on other scales (e.g., regional or state collections) and geographic regions beyond Europe and the United States.

Understanding the Information Needs of Large-Scale Digital Library Users

This paper reports on the results of a comparative study of user search logs in two large-scale, domain-specific digital libraries functioning in the United States: the National Science Digital Library and Opening History. Results demonstrate varying levels of use of advanced search options and substantial differences in the search query lengths, search query frequencies, and distribution of search categories in queries. The empirical data on how the members of the target communities search can be used in deriving important information for domain-specific digital library developers’ decisions regarding both the details of information organization and support of various search features.

Exploring the role of scale: Comparative analysis of digital library user searching

This poster reports preliminary results of a comparative study of user searching in two large-scale digital libraries with history focus in the United States: the federal-level and the state-level digital library. Similarities were observed in search query lengths and average numbers of search categories per search query. At the same time, the study reveals significant differences in the level of use of advanced search options and in search query frequencies, as well as in distribution of most search categories in user queries. The empirical data obtained in this exploratory study will inform further research and will be useful for professionals making decisions on resource description and user interfaces for digital libraries.

Scholar‐built collections: A study of user requirements for research in large‐scale digital libraries

ABSTRACTTo realize the great potential value of large‐scale digital libraries, we need a fuller understanding of the range of ways in which scholarly communities conduct research, or want to conduct research within them. Scholars build collections in the course of their work. How can we anticipate and support various kinds of collection‐building and ‐use, in order to support the diversity of researchers who work in libraries of digital books? This paper reports selected results of a study of how potential user groups of the HathiTrust Digital Library create and use collections in their research. This study aims to contribute to our broader understanding of scholarly practice, particularly of humanities scholars’ collecting activities. The results of the study inform ongoing work to develop a workset‐creation tool for the HathiTrust Research Center.

Author gender metadata augmentation of hathitrust digital library

Bibliographic metadata is essential for digital library resource description. Especially as the size and number of bibliographic entities grows, high-quality metadata enables richer forms of digital library access, search, and use. Metadata records can be enriched through automated techniques. For example, a digital humanities scholar might use the gender of a set of authors during their literature analysis. In this study, we undertook to enrich the metadata description of a large-scale digital library, the HathiTrust (HT) digital library, specifically by determining the gender of authors of the public domain portion of the collection. The results are stored to a separate Solr index accessible through the HathiTrust Research Center services. This study, which successfully resolved in 78.9% of the cases the gender of authors in the HT public domain corpus, suggests future research directions in capturing and representing the provenance of the contributing sources to enhance trust, and in machine learning to resolve the remaining names.

Complementarity in Subject Metadata in Large-Scale Digital Libraries: A Comparative Analysis

Provision of high-quality subject metadata is crucial for organizing adequate subject access to rich content aggregated by digital libraries. A number of large-scale digital libraries worldwide are now generating subject metadata to describe not only individual objects but entire digital collections as an integral whole. However, little research to date has been conducted to empirically evaluate the quality of this collection-level subject metadata. The study presented in this article compares free-text and controlled-vocabulary collection-level subject metadata in three large-scale cultural heritage digital libraries in the United States and the European Union. As revealed by this study, the emerging best practices for creating rich collection-level subject metadata includes describing a collection's subject matter with mutually complementary data values in controlled-vocabulary and free-text subject metadata elements. Three kinds of complementarity were observed in this study: one-way complementarity, two-way complementarity, and multiple complementarity.

Subject Access: Conceptual Models, Functional Requirements, and Empirical Data

One of the central functions of bibliographic control is providing subject access. However, numerous studies conducted over decades have shown that users routinely experience problems with subject access in library catalogs and databases. These problems are often due to inadequate quality of subject metadata, which is greatly influenced by complexity of subject representation. The fact that major cataloging standards (e.g., AACR2 and its predecessors) have been overlooking the importance of subject access and have not addressed subject cataloging is arguably one of the reasons behind problems in organization of subject access. The new cataloging code—Resource Description and Access (RDA)—attempts to fill this gap. Upon examination of how subject access is addressed in RDA and its underlying conceptual models that specify functional requirements for bibliographic control—FRBR, FRAD, and FRSAD—this article presents results of a study that used FRBR model as an analytical framework in examination of user search queries in a large-scale digital library. The findings of this study provide empirical data to inform the development of RDA sections that cover subject access, particularly subject entities and relationships.

Exploring the richness of collection-level subject metadata in three large-scale digital libraries

A growing number of digital libraries worldwide are now generating collection-level metadata to describe entire digital collections as integral wholes. This paper reports results of the study that used an in-depth comparative content analysis to assess free-text and controlled-vocabulary collection-level subject metadata in three large-scale digital libraries in the European Union and the USA. As observed by this study, the emerging best practices in creating collection-level subject metadata include: (a) describing collection's subject matter with mutually complementary values in controlled-vocabulary and free-text metadata; (b) encoding a variety of collection properties in free-text metadata, including both subject properties (topical, geographic and temporal coverage, and types/genres of objects) and non-subject information: title, size, provenance, collection development, copyright, audience, navigation and functionality, language of items in a digital collection, frequency of additions, institutions that host a collection or contribute to it, funding sources, item creators, importance, uniqueness and comprehensiveness of a digital collection.

The ‘universal library’ returns in digital form

This paper begins with two contemporary technological developments both of which present a serious challenge to the dominance in literature and learning of the book in its codex form. While their earlier manifestations were not commercially successful, the most recent e-book readers (portable technological contraptions that have the capacity to store thousands of books that can be read electronically) have been praised not only for their functionality but also for their aesthetic appeal. A relateddevelopment has been the growth of large-scale digital libraries, the most prominent of which is the Google Books Library Project, launched in 2004 and now committed to the digitisation of around 15 million volumes or 4.5 billion pages in the following six years from some of the world’s leading academic libraries.The purpose of this paper is to explore these developments within the context of ancient and Enlightenment ideas about the ‘universal library’ which assert that the construction of such an institution is the most effective way of promoting universal knowledge. Rather than employing a kind of technological determinism that renders these technologies as merely points along an inexorable continuum of progress, it will be argued that they are the latest manifestation of an idea that long pre-dated digital technology. Over two millennia ago, the Ptolemies attempted to collect the entire corpus of literature in the Greek language as well as significant works in other languages. Many have argued that the institution that held these huge collections, the library at Alexandria, was effectively the world’s first universal library. Even though this library was eventually destroyed, the idea of universalism survived and flourished again during the European Enlightenment, through Diderot’s Encyclopédie project and the construction of national libraries and archives. Latterly, the creation of the World Wide Web is conceived of by some as the apotheosis of the universalism of knowledge.Not everyone is convinced, though, that the attainment of universal knowledge is possible. And, ironically, the most powerful arguments against universalism emanate from a medium whose existential status is most threatened by this idea and its accompanying technologies: the codex book. Despite our technological sophistication, it seems that literature is still able to give us much richer insights into the nature of contemporary society than other forms of media. This paper will illustrate this through the writings of Jorge Luis Borges, Jonathan Swift, Jean Paul Sartre, Washington Irving and Gustave Flaubert, who in their various ways demonstrate the epistemological impossibility of obtaining total knowledge in any intellectual discipline, parody the insane search for it and muse on the tension between canonicity and universalism. These issues are particularly pertinent in an age where the virtualization of text has seemingly rendered obsolete some of the practical obstacles to the universal library and offer a sophisticated rejoinder to those cyber-utopians who have heralded the imminent arrival of universal knowledge via the World Wide Web

Digital Libraries Are Taking Form

Large-scale digital libraries and book digitization projects are poised to go beyond prototypes into the mass market. This is generally accepted as a universal boon. However, whether end users anywhere in the world will be able to easily access materials across these projects is another matter. Much of digitization's promise might diminish unless commercial, nonprofit, and publicly funded resource organizations can agree to cooperate.

Digital Libraries Are Taking Form

Large-scale digital libraries and book digitization projects are poised to go beyond prototypes into the mass market. This is generally accepted as a universal boon. However, whether end users anywhere in the world will be able to easily access materials across these projects is another matter. Much of digitization's promise might diminish unless commercial, nonprofit, and publicly funded resource organizations can agree to cooperate.

Detection and analysis of table of contents based on content association

As a special type of table understanding, the detection and analysis of tables of contents (TOCs) play an important role in the digitization of multi-page documents. Most previous TOC analysis methods only concentrate on the TOC itself without taking into account the other pages in the same document. Besides, they often require manual coding or at least machine learning of document-specific models. This paper introduces a new method to detect and analyze TOCs based on content association. It fully leverages the text information throughout the whole multi-page document and can be directly applied to a wide range of documents without the need to build or learn the models for individual documents. In addition, the associations of general text and page numbers are combined to make the TOC analysis more accurate. Natural language processing and layout analysis are integrated to improve the TOC functional tagging. The applications of the proposed method in a large-scale digital library project are also discussed.

A Fault-Tolerant Architecture for Supporting Large Scale Digital Libraries

At the Research Library of Los Alamos National Laboratory (LANL), we have developed an affordable multi-terabyte redundant RAID system -- running on Linux -- that addresses the needs of a constantly growing demand for electronic journals, conferences and standards. Although there are several practical issues associated with systems supporting electronic content -- such as acquisition, archiving, access and cataloging -- this paper will focus on archiving and providing access to electronic journals. This article explains the design and development of a scalable, fault-tolerant and affordable system, presents an overview of the architecture, summarizes the problems and architectural issues encountered, and describes the technical solutions adopted.

Usage Analysis for the Identification of Research Trends in Digital Libraries

The analysis of user logs from large-scale digital libraries offers new opportunities to assess research trends in an institution's user communities. We describe the application of a methodology to derive weighted journal relationship networks from reader logs at the Los Alamos National Laboratory's Research Library during 1998 and 2001. A journal impact metric is defined that derives journal impact from the structural features of the generated journal relationship networks, much in the same manner Google's PageRank evaluates the impact of web pages for a given subject on the basis of its context of hyperlinks to other pages. A comparison of this reader impact metric to the ISI Impact Factor values for the same journals in 1998 and 2001 allows us to detect and interpret community-specific research trends where the LANL community deviates from Usage Analysis for the Identification of Research Trends in Digit... http://www.dlib.org/dlib/may03/bollen/05bollen.html 1 of 21 12/9/13 4:02 PM more general trends as indicated by changes in the Institute for Scientific Indexing (ISI) Impact Factors during those same years. Such analysis yields information to aid digital library managers to improve the evaluation of not only which parts of the collection are most highly valued by their local community, but it also detects research trends in user communities as they evolve over time.

Formal ontology for subject

Subject-based classification is an important part of information retrieval, and has a long history in libraries, where a subject taxonomy was used to determine the location of books on the shelves. We have been studying the notion of subject itself, in order to determine a formal ontology of subjects for a large-scale digital library card catalog system. Deep analysis reveals a lot of ambiguity regarding the usage of subjects in existing systems and terminology, and we attempt to formalize these notions into a single framework for representing it.

Creating a Large-Scale Digital Library for Georeferenced Information

Creating a Large-Scale Digital Library for Georeferenced Information

Exploratory analysis of concept and document spaces with connectionist networks

Exploratory analysis is an area of increasing interest in the computational linguistics arena. Pragmatically speaking, exploratory analysis may be paraphrased as natural language processing by means of analyzing large corpora of text. Concerning the analysis, appropriate means are statistics, on the one hand, and artificial neural networks, on the other hand. As a challenging application area for exploratory analysis of text corpora we may certainly identify text databases, be it information retrieval or information filtering systems. With this paper we present recent findings of exploratory analysis based on both statistical and neural models applied to legal text corpora. Concerning the artificial neural networks, we rely on a model adhering to the unsupervised learning paradigm. This choice appears naturally when taking into account the specific properties of large text corpora where one is faced with the fact that input-output-mappings as required by supervised learning models cannot be provided beforehand to a satisfying extent. This is due to the fact of the highly changing contents of text archives. In a nutshell, artificial neural networks count for their highly robust behavior regarding the parameters for model optimization. In particular, we found statistical classification techniques much more susceptible to minor parameter variations than unsupervised artificial neural networks. In this paper we describe two different lines of research in exploratory analysis. First, we use the classification methods for concept analysis. The general goal is to uncover different meanings of one and the same natural language concept. A task that, obviously, is of specific importance during the creation of thesauri. As a convenient environment to present the results we selected the legal term of "neutrality", which is a perfect representative of a concept having a number of highly divergent meanings. Second, we describe the classification methods in the setting of document classification. The ultimate goal in such an application is to uncover semantic similarities of various text documents in order to increase the efficiency of an information retrieval system. In this sense, document classification has its fixed position in information retrieval research from the very beginning. Nowadays renewed massive interest in document classification may be witnessed due to the appearance of large-scale digital libraries.