Research and Advanced Technology for Digital Libraries

This paper describes the Social Networks and Archival Context project, built on a database of merged Encoded Archival Context - Corporate Bodies, Persons, and Families (EAC-CPF) records derived from Encoded Archival Description (EAD) records held by the Library of Congress, the California Digital Library, the Northwest Digital Archives, and Virginia Heritage, combined with information from name authority files from the Library of Congress (Library of Congress Name Authority File), OCLC Research (The Virtual International Authority File), and the Getty Vocabulary Program (Union List of Artist Names). The database merges information from each instance of an individual name found in the EAD resources, along with variant names, biographical notes and their topical descriptions. The SNAC prototype interface makes this information searchable and browseable while retaining links to the various data sources.

Bibliographic databases are a prosperous field for data mining research and social network analysis. The representation and visualization of bibliographic databases as graphs and the application of data mining techniques can help us uncover interesting knowledge regarding how the publication records of authors evolve over time. In this paper we propose a novel methodology to model bibliographical databases as

Power Graphs

, and mine them in an unsupervised manner, in order to learn basic author types and their properties through clustering. The methodology takes into account the evolution of the co-authorship information, the volume of published papers over time, as well as the impact factors of the venues hosting the respective publications. As a proof of concept of the applicability and scalability of our approach, we present experimental results in the

DBLP

data.

The World Wide Web has a very dynamic character with resources constantly disappearing and (re-)surfacing. A ubiquitous result is the “404 Page not Found” error as the request for missing web pages. We investigate tags obtained from Delicious for the purpose of rediscovering such missing web pages with the help of search engines. We determine the best performing tag based query length, quantify the relevance of the results and compare tags to retrieval methods based on a page’s content. We find that tags are only useful in addition to content based methods. We further introduce the notion of “ghost tags”, terms used as tags that do not occur in the current but did occur in a previous version of the web page. One third of these ghost tags are ranked high in Delicious and also occurred frequently in the document which indicates their importance to both the user and the content of the document.

The MPEG-7 is the dominant standard for multimedia content description; thus, the audiovisual Digital Library contents should be described in terms of MPEG-7. Since there exists a huge amount of audiovisual content in the cultural heritage domain, it is expected that several cultural heritage objects, as well as entities related with them (i.e. people, places, events etc.), have been described using MPEG-7. On the other hand, the dominant standard in the cultural heritage domain is the CIDOC/CRM; consequently, the MPEG-7 descriptions cannot be directly integrated in the cultural heritage digital libraries.

We present in this paper a mapping model and a system that allow the transformation of the MPEG-7 descriptions to CIDOC/CRM descriptions, thus allowing the exploitation of multimedia content annotations in the cultural heritage digital libraries. In addition, the proposed mapping model allows linking MPEG-7 descriptions to CIDOC/CRM descriptions in a Linked Data scenario.

Subject headings systems are tools for organization of knowledge that have been developed over the years by libraries. The SKOS Simple Knowledge Organization System has provided a practical way to represent subject headings systems using the Resource Description Framework, and several libraries have taken the initiative to make subject headings systems widely available as open linked data. Each individual subject heading describes a concept, however, in the majority of cases, one subject heading is actually a combination of several concepts, such as a topic bounded in geographical and temporal scopes. In these cases, the label of the concept actually carries several concepts which are not represented in structured form. Our work explores machine learning techniques to recognize the sub concepts represented in the labels of SKOS subject headings. This paper describes a language independent named entity recognition technique based on conditional random fields, a machine learning algorithm for sequence labelling. This technique was evaluated on a subset of the Library of Congress Subject Headings, where we measured the recognition of geographic concepts, topics, time periods and historical periods. Our technique achieved an overall F

1

score of 0.98.

The Finnish National Digital Library (NDL) project aims to improve online accessibility and usability of digital content held by libraries, museums and archives. The lack of standardized metadata and numerous different collection management systems without sufficient set of technical standards in the museum sector led us to create a set of instructions and a template mapping of the Lightweight Information Describing Objects (LIDO) XML schema. This national LIDO schema for museum sector described in our paper is unique both in coverage in museum object types as well as number of institutions using it. A common schema presents heterogeneous metadata uniformly, thus enabling easy retrieval, browsing and versatile linking between different object types as well as data fields. In the pilot phase we have mapped the three most commonly used Finnish collection management systems with three different metadata formats to the top level LIDO schema.

The FRBR model has the potential for new services and discovery techniques for cultural items such as books, movies and music. In this paper, we present an approach to interpret descriptions found in Web resources and identify the FRBR entities these pertain to. To verify the resulting set of FRBR entities, we have used the Linked Open Data and the verifications have been validated by a group of experts. The results of this work demonstrates applicability of FRBR in a new context and establishes a firm basis for further exploitation.

During the last decades many information providers, such as libraries, have been collecting, organizing and delivering information in both print and digital format, forming a hybrid information environment. However, exploration of a hybrid information environment does not result in a unified seeking experience, which exploits most effectively the available resources. This paper aims to identify the main factors that influence the adoption of wireless, mobile devices (e.g., smartphones) as a means of integrating the information seeking process in hybrid environments. Therefore it presents a prototype system and an evaluation study that provides an insight about the services design.

We report on focus group feedback regarding the services provided by existing education-related Digital Libraries (DL). Participants provided insight into how they seek educational resources online, and what they perceive to be the shortcomings of existing educational DLs. Along with useful content, social interactions were viewed as important supplements for educational DLs. Such interactions lead to both an online community and new forms of content such as reviews and ratings. Based on our analysis of the focus group feedback, we propose DL 2.0, the next generation of digital library, which integrates social knowledge with DL content.

Virtual Research Environments are internet-based working environments tailored to serve needs of diverse and evolving user communities. These environments are oriented to promote new ways of dealing with modern research tasks. Their realization requires user interfaces that are dynamically built to provide their clients with

organised views

on the data and services aggregated to meet specific community needs. This paper presents an approach to the problem of Virtual Research Environment user interfaces dynamic construction. This approach is characterized by user interfaces built through a component-oriented strategy and an heuristic for user interface constituents arrangement on the screen. The implementation and exploitation of the proposed approach in the context of the D4Science-II, EU funded project is discussed as well as future plans are presented.

Emory University Library teamed with a commercial firm to develop a prototype system for using Amazon’s EC2 to properly size web application server deployment environments. This approach has been successfully applied to both high-transaction commercial environments with hundreds of thousands of users and to lower transaction digital library environments with hundreds of users. Starting with the same EC2-based product, our goal was to assess whether a similar strategy is practical for an academic library as well as for commercial systems. We examined cloud configuration and deployment costs, test preparation and analysis, and overall feasibility of this approach. Typically, for digital libraries, the user levels are significantly lower, the deployment costs are lower, and the return on investment (ROI) is not as immediately obvious. We conclude that the effort is worth the investment only (a) when there are significant repercussions from under-sizing a newly deployed digital library and (b) sufficient engineering staff are on hand to develop and debug the deployment scenarios.

Search engines allow users to retrieve documents with respect to a given query. These provide advanced search options, such as query operators (e.g.,

+term

,

term^10

). Previous work studied how query operators are employed by end-users. In this paper, we study the extent to which using query operators may lead to improved results, regardless of specific users. We hypothesize that the proper use of query operators improves search results. To validate this hypothesis, we present a methodology relying on standard IR test collections. We applied this methodology to TREC-7 and TREC-8 test collections with five IR models implemented in the

Terrier

search engine. Experiments show that queries enriched with operators give an improvement in effectiveness up to 35.1% over regular queries. This result suggests that end-users would benefit from using operators more often.

In all subfields of information retrieval, test datasets and ground truth data are important tools for testing and comparison of new search methods. This is also reflected by the image retrieval community where several benchmarking activities have been created in past years. However, the number of available test collections is still rather small and the existing ones are often limited in size or accessible only to the participants of benchmarking competitions. In this work, we present a new freely-available large-scale dataset for evaluation of content-based image retrieval systems. The dataset consists of 20 million high-quality images with five visual descriptors and rich and systematic textual annotations, a set of 100 test query objects and a semi-automatically collected ground truth data verified by users. Furthermore, we provide services that enable exploitation and collaborative expansion of the ground truth.

YouTube is the largest, most popular video digital library in existence, and is quite possibly the most popular digital library regardless of format type. Furthermore, music videos are one of the primary applications of YouTube. Based on our experiences of linking to music videos in YouTube, we observed that while any single URI had a short half-life, music videos were always available at another URI. For this study we collected 1291 music videos and found that very few had zero or one copies in YouTube at any given time, and some had several thousand copies at any given time. Furthermore, individual URIs had a half-life of anywhere from 9 to 18 months, depending on the publication date and remaining commercial potential.

We present on-going work on the linguistic and semantic processing of the labels of the Thompson’s Motif-Index of Folk-Literature, which has been proposed by Stith Thompson for the classification of narrative elements in folk-literature. We automatically extracted the labels of an on-line version of the Index, and wrote specialised grammars for providing for a multi-layer linguistic annotation of them. We are currently working on enriching the linguistically annotated labels with semantic classes and relations, allowing for a better access to the content of the Index. With this resource, we expect to be able to semi-automatically annotate digitised literary works at the sub-document level by means of automatically comparing the annotated Index with the results of text processing tools applied to those works, and so contribute to a better inter-textual interlinking and understanding of related works in the folk-literature, offering a new way of semantically accessing digital libraries.

The Walden’s Paths Project, as part of our philosophy of continual evaluation, actively seeks out user communities who may find our system to be of interest. In the past few years we noticed a recurring trend of user issues, needs, and sought-after features. In order to better support our users, we initiated a redesign of Walden’s Paths that not only solves these problems, but enables us to perform more rapid prototyping and experimentation of new features and interfaces. In order to accomplish these goals, we have created a web service that handles the storage, modification, and representation of our path data structures. This service is completely isolated from user interface layers, allowing many different interface designs to be implemented on top of the basic Walden’s Paths data structures. We also present several prototype interfaces - Marginalia, CoWPaths, Walden’s Drupal, PathCompiler v2, mWalden - that represent new areas in which we believe our ideas can be applied such as collaborative work, location-aware services, large educational databases, offline presentation, and mobile computing.

We conducted a study with an objective to learn more about the dynamic information needs, information-seeking behavior, information use and other scholarly activities of researchers. Our focus was on the collaborative and social usage and on the social reference managers. We compared the current practices and strategies of scholars and researchers from multidisciplinary research areas. Our findings provide valuable insights and augment the understanding of how the social web is having a significant effect on the current researchers’ activities and digital libraries.

Fingerprint management systems support millions of images and complicated but imperfect image identification algorithms. The forensic community requires a set of digital library services to support large image collections, execute identification algorithms, and analyze experiments that test identification algorithms in development. We present a model and prototype system capable of testing and analyzing fingerprinting algorithms in terms of identification performance based on matches of a known image to partial images, distortions of the images, and sub-regions of the images. These services are provided based on our framework for composing a set of services and a fingerprint image collection. The prototype will be useful in collaborations connecting several algorithm development efforts, and in composing an experimentation workflow. We also describe extensions of these services into other domains.

Interactive query expansion can assist users during their query formulation process. We conducted a user study with over 4,000 unique visitors and four different design approaches for a search term suggestion service. As a basis for our evaluation we have implemented services which use three different vocabularies: (1) user search terms, (2) terms from a terminology service and (3) thesaurus terms. Additionally, we have created a new combined service which utilizes thesaurus term and terms from a domain-specific search term recommender. Our results show that the thesaurus-based method clearly is used more often compared to the other single-method implementations. We interpret this as a strong indicator that term suggestion mechanisms should be domain-specific to be close to the user terminology. Our novel combined approach which interconnects a thesaurus service with additional statistical relations outperformed all other implementations. All our observations show that domain-specific vocabulary can support the user in finding alternative concepts and formulating queries.

Defective metadata is a significant problem of digital libraries. So far, automatic error detectors have been in the focus of research interest. However, recent public projects have shown that patrons are willing to invest time to report errors if they are called to contribute. In this case-study, we analyze the community contribution to error detection for DBLP, a public bibliographic collection. Our study is based on e-mails sent to the project between January 2007 and November 2010. We manually and automatically identify error reports and analyze their contribution to corrections of the DBLP collection. We show that users frequently report certain types of defects while others are ignored. The detection of homonym-name inconsistencies in particular strongly depends on user input. We also discuss who sends the reports and which communities are particularly active in this matter.

The advent of Internet and digital media has added more complications to the already complex copyright laws. This paper will first summarize the history of copyright laws in the United States and Europe. It will then analyze and compare the digital copyright laws as they are applied in higher education in the United States and major countries in Europe.

When a user knows exactly what they are looking for most library systems are adequate for their needs. However, when the user’s information needs are ill-defined - traditional library systems prove inadequate. This is because traditional library systems are not designed to support sense making rather for information retrieval. Visual analytics is the science of analytical reasoning facilitated by interactive visualizations and visual analytics systems can support both sense making and information retrieval. In this paper, we present INVISQUE - an approach and experimental software for interactive visual search and query. INVISQUE uses an index card metaphor to display library content, organized in a way that visually integrates attributes such citations and date published, making it easy to pick out the most recent and most cited paper. It uses design techniques such as focus+context to reveal relationships between documents, while avoiding the “what-was-I-lookingfor?” problem.

In this paper, we describe a comparative user evaluation of two multilingual thesaurus-enhanced visual user interfaces, namely T-Saurus and Searchling, developed for digital libraries. The study used 25 academic users carrying out three search tasks on both user interfaces to the UNESCO digital portal, holding 400,000 documents. It applied usability and affordance strength questionnaires, interviews, thinkalouds, and direct observation to investigate users’ evaluation of the key components of both user interfaces, namely multilingual features and thesaurus and search functions. The empirical data gathered will be useful for designers of search interfaces that use thesaurus and multilingual features. Results of the study show that users were able to successfully carry out the search tasks using thesaurus-enhanced search interfaces. However, they preferred Searchling for its flexible language option, thesaurus browsing and visualization.

We describe a framework for Adaptive Multilingual Information Retrieval (AMIR) which allows multilingual resource discovery and delivery using on-the-fly machine translation of documents and queries. Result documents are presented to the user in a contextualised manner. Challenges and affordances of both adaptive and multilingual IR, with a particular focus on digital libraries, are detailed. The framework components are motivated by a series of results from experiments on query logs and documents from The European Library. We conclude that factoring adaptivity and multilinguality aspects into the search process can enhance the user’s experience with online digital libraries.

We report on three years of data collected in the field from students in graduate and undergraduate seminars at two universities. The students annotated texts for discussion in classes where hypertext and computer interfaces were core topics. The results of our analysis show how annotation style changes with a combination of experience and study of material related to annotation. Our major conclusions are that there are essentially six purposes for scholarly user-readers to annotate; and support for textual glosses is a necessary part of any successful annotation technology for such use. Our study suggests tools that will be appreciated by e-text users.

Transaction log analyses are common practice to understand user behavior in both online databases and library catalogues. While there has been significant work done in each of these domains, there is little work comparing user queries between library catalogues and online resources. In this paper we report on an exploratory comparison between searches performed via the same interface in three different search systems: a library catalogue, an online research database, and Google Scholar.

How are definitions of content and the design of digital documents being determined in practice

? In this paper the authors present the relationship between document encoder and document as the central unit of analysis in a framework for making sense of documentary practice at community, organisational and implementation levels. The paper presents the integrated findings from a global survey of document encoders participating in the Text Encoding Initiative, providing important insights into the characteristics of an emergent documentary practice. By focusing on documentation as a field of practice the paper reveals a rich and generative practice at play and provides valuable lessons for other complex metadata and markup initiatives.

In this paper, we present an approach to automatically link FRBR works identified in metadata to the corresponding entity in Linked Open Data resources. The main contribution is a basis for semantic enrichment and verification of works identified in existing metadata. Through experiments, we demonstrate that FRBR works can be identified in the LOD cloud, which provides a solid ground for further work.

In many heritage institutes, objects are routinely described using terms from predefined vocabularies. When object collections need to be merged or linked, the question arises how those vocabularies relate. In practice it often unclear for data providers how well alignment tools will perform on their specific vocabularies. This creates a bottleneck to align vocabularies, as data providers want to have tight control over the quality of their data. We will discuss the key limitations of current tools in more detail and propose an alternative approach. We will show how this approach has been used in two alignment use cases, and demonstrate how it is currently supported by our Amalgame alignment platform.

Laboratories have long been seen as reasonable proxies for user experience research. Yet, this assumption may have become unreliable. The trend toward multiple activities in the users’ natural environment, where people simultaneously use a digital library, join a chat or read an incoming Facebook post, changes users’ behavior. The effects of these disruptions generate a gap that is generally not taken into account in user-experience research. This paper presents a psychological experiment that measured how differently people behave in a laboratory and in a natural environment setting. The existence and impact of distraction is measured in a standard laboratory setting and in a remote setting that explicitly allows users to work in their own natural environment. The data indicates that there are significant differences between results from the laboratory and natural environment setting. Distractions like email or chat influence the users’ performance and their ratings.

In this paper we present a framework for improving the ranking learning process, taking into account the implicit search behaviors of users. Our approach is query-centric. That is, it examines the search behaviors induced by queries and groups together queries with similar such behaviors, forming

search behavior clusters

. Then, it trains multiple ranking functions, each one corresponding to one of these clusters. The trained models are finally combined to re-rank the results of each new query, taking into account the similarity of the query with each cluster. The main idea is that similar search behaviors can be detected and exploited for result re-ranking by analysing results into feature vectors, and clustering them. The experimental evaluation shows that our method improves the ranking quality of a state of the art ranking model.

Evaluation is a central digital library practice. It provides important data for managing digital libraries and informing strategic decision-making. Digital library evaluation and management are organizational as well as technical practices. What evaluation models can account for these organizational factors, in practice as well as in theory? To address these questions, this paper integrates two models, one from the organizational literature (Porter’s value chain), and one from the evaluation literature (evaluation logic models), into a generic, flexible and extensible evaluation model that supports the goal-oriented evaluation and management of digital libraries in specific sociotechnical contexts. A case study is provided.

While the concept of digital library (DL) is well perceived and applied in developed countries, it is still a big challenge to the developing nations. There are great disparities, known as digital divide between developed countries and developing countries in terms of electronic resource funding, availability, and accessibility. DL, together with the information retrieval (IR) system, is believed to be an effective way to mend the gap of digital divide. This paper will employ a real case to discuss the significance of developing a national level of digital library for pre-university schools of Albania, the challenges of designing such information system both economically and technologically, and considerations of designing the digital library.

The Digital Assets Repository (DAR) is a system developed at the Bibliotheca Alexandrina to manage the full lifecycle of a digital asset: its creation and ingestion, its metadata management, storage and archival in addition to the necessary mechanisms for publishing and dissemination. In its third release, the system architecture has been revamped into a modular design including components that are best of the breed, in addition to defining a flexible content model for digital objects based on current standards and a focus on integrating DAR with different sources and applications. The goal of this paper is to demonstrate the building blocks of DAR as an example of a modern repository, in addition to discussing the challenges that face an institution in consolidating its assets and DAR’s answer to these challenges.

News, multimedia and cultural heritage archives are increasingly offering opportunities to create connections between their collections. We consider the task of linking archives: connecting an item in one archive to one or more items in other, often complementary archives. We focus on a specific instance of the task: linking items with a rich textual representation in a news archive to items with sparse annotations in a multimedia archive, where items should be linked if they describe the same or a related event. We find that the difference in textual richness of annotations presents a challenge and investigate two approaches: (i) to enrich sparsely annotated items with textually rich content; and (ii) to reduce rich news archive items using term selection. We demonstrate the positive impact of both approaches on linking to same events and linking to related events.

In the information retrieval context, resource collections are frequently classified using thesauri. However, the limited semantics provided by thesauri restricts the collection search and browsing capabilities. This work focuses on improving these capabilities by transforming a set of resources indexed according to a thesaurus into a semantically tagged collection. The core mechanism for building this collection is based on the conversion of the domain specific thesaurus (indexing the collection of resources) into a domain ontology connected to an upper level ontology. The feasibility of this work has been tested in the urban domain by transforming the resources accessible through the European Urban Knowledge Network into a Linked Data repository.

Europeana is a long-term project funded by the European Commission with the goal of making Europe’s cultural and scientific heritage accessible to the public. Since 2008, about 1500 institutions have contributed to Europeana, enabling people to explore the digital resources of Europe’s museums, libraries and archives. The huge amount of collected multi-lingual multi-media data is made available today through the Europeana portal, a search engine allowing users to explore such content through textual queries. One of the most important techniques for enhancing users search experience in large information spaces, is the exploitation of the knowledge contained in query logs. In this paper we present a characterization of the Europeana query log, showing statistics on common behavioral patterns of the Europeana users. Our analysis highlights some significative differences between the Europeana query log and the historical data collected by general purpose Web Search Engine logs. In particular, we find out that both query and search session distributions show different behaviors. Finally, we use this information for designing a query recommendation technique having the goal of enhancing the functionality of the Europeana portal.

Europeana has put in a stretch many known procedures in digital libraries, imposing requirements difficult to be implemented in many small institutions, often without dedicated systems support personnel. Although there are freely available open source software platforms that provide most of the commonly needed functionality such as OAI-PMH support, the migration from legacy software may not be easy, possible or desired. Furthermore, advanced requirements like selective harvesting according to complex criteria are not widely supported. To accommodate these needs and help institutions contribute their content to Europeana, we developed a series of tools. For the majority of small content providers that are running DSpace, we developed a DSpace plug-in, to convert and augment the Dublin Core metadata according to Europeana ESE requirements. For sites with different software, incompatible with OAI-PMH, we developed wrappers enabling repeatable generation and harvesting of ESE-compatible metadata via OAI-PMH. In both cases, the system is able to select and harvest only the desired metadata records, according to a variety of configuration criteria of arbitrary complexity. We applied our tools to providers with sophisticated needs, and present the benefits they achieved.

Web archiving has been gaining interest and recognized importance for modern societies around the world. However, for web archivists it is frequently difficult to demonstrate this fact, for instance, to funders. This study provides an updated and global overview of web archiving. The obtained results showed that the number of web archiving initiatives significantly grew after 2003 and they are concentrated on developed countries. We statistically analyzed metrics, such as, the volume of archived data, archive file formats or number of people engaged. Web archives all together must process more data than any web search engine. Considering the complexity and large amounts of data involved in web archiving, the results showed that the assigned resources are scarce. A Wikipedia page was created to complement the presented work and be collaboratively kept up-to-date by the community.

We point out, in this paper, the issue of improving the coherence of web archives under limited resources (

e.g.

bandwidth, storage space, etc.). Coherence measures how much a collection of archived pages versions reflects the real state (or the snapshot) of a set of related web pages at different points in time. An ideal approach to preserve the coherence of archives is to prevent pages content from changing during the crawl of a complete collection. However, this is practically infeasible because web sites are autonomous and dynamic. We propose two solutions:

a priori

and

a posteriori

. As

a priori

solution, our idea is to crawl sites during the

off-peak

hours (

i.e.

the periods of time where very little changes is expected on the pages) based on patterns. A pattern models the behavior of the importance of pages changes during a period of time. As an

a posteriori

solution, based on the same patterns, we introduce a novel navigation approach that enables users to browse the most coherent page versions at a given query time.