Completeness of metadata is one of the most essential characteristics of their quality. An incomplete metadata record is a record of degraded quality. Existing approaches to measure metadata completeness limit their scope in counting the existence of values in fields, regardless of the metadata hierarchy as defined in international standards. Such a traditional approach overlooks several issues that need to be taken into account. This paper presents a fine-grained metrics system for measuring metadata completeness, based on field completeness. A metadata field is considered to be a container of multiple pieces of information. In this regard, the proposed system is capable of following the hierarchy of metadata as it is set by the metadata schema and admeasuring the effect of multiple values of multivalued fields. An application of the proposed metrics system, after being configured according to specific user requirements, to measure completeness of a real-world set of metadata is demonstrated. The results prove its ability to assess the sufficiency of metadata to describe a resource and provide targeted measures of completeness throughout the metadata hierarchy.

Climate change has become a major area of concern over the past few years and consequently many governments, international bodies, businesses, and institutions are taking measures to reduce their carbon footprint. However, to date very little research has taken place on information and sustainable development in general, and on the environmental impact of information services in particular. Based on the data collected from various research papers and reports, this review article shows that information systems and services for the higher education and research sector currently generate massive greenhouse gas (GHG) emissions, and it is argued that there is an urgent need for developing a green information service, or green IS in short, that should be based on minimum GHG emissions throughout its lifecycle, from content creation to distribution, access, use, and disposal. Based on an analysis of the current research on green information technology (IT), it is proposed that a green IS should be based on the model of cloud computing. Finally, a research agenda is proposed that will pave the way for building and managing green ISs to support education and research/scholarly activities.

Privacy concerns can greatly hinder consumers' intentions to interact with a website. The success of a website therefore depends on its ability to improve consumers' perceptions of privacy assurance. Seals and assurance statements are mechanisms often used to increase this assurance; however, the findings of the extant literature regarding the effectiveness of these tools are mixed. We propose a model based on the elaboration likelihood model (ELM) that explains conditions under which privacy assurance is more or less effective, clarifying the contradictory findings in previous literature. We test our model in a free-simulation online experiment, and the results of the analysis indicate that the inclusion of assurance statements and the combination, understanding, and assurance of seals influence privacy assurance. Privacy assurance is most effective when seals and statements are accompanied by the peripheral cues of website quality and brand image and when counter-argumentation—through transaction risk—is minimized. Importantly, we show ELM to be an appropriate theoretical lens to explain the equivocal results in the literature. Finally, we suggest theoretical and practical implications.

Multisource web news portals provide various advantages such as richness in news content and an opportunity to follow developments from different perspectives. However, in such environments, news variety and quantity can have an overwhelming effect. New-event detection and topic-tracking studies address this problem. They examine news streams and organize stories according to their events; however, several tracking stories of an event/topic may contain no new information (i.e., no novelty). We study the novelty detection (ND) problem on the tracking news of a particular topic. For this purpose, we build a Turkish ND test collection called BilNov-2005 and propose the usage of three ND methods: a cosine-similarity (CS)-based method, a language-model (LM)-based method, and a cover-coefficient (CC)-based method. For the LM-based ND method, we show that a simpler smoothing approach, Dirichlet smoothing, can have similar performance to a more complex smoothing approach, Shrinkage smoothing. We introduce a baseline that shows the performance of a system with random novelty decisions. In addition, a category-based threshold learning method is used for the first time in ND literature. The experimental results show that the LM-based ND method significantly outperforms the CS- and CC-based methods, and category-based threshold learning achieves promising results when compared to general threshold learning.

In the summer of 2008 all University of North Carolina libraries switched from a traditional library catalog interface supporting text-based searching (TextOnly) to a text and facet-based interface (TextFacet) to improve users' search experiences. This study seeks to understand the differences between these two interfaces and how they affect the search experience of the novice user. In this study, 40 participants were asked to search for resources using both interfaces. Their search times and accuracy were measured across three types of search tasks (known, partially known, and exploratory). After completing the searches, they were asked a series of questions about their experiences. The data were analyzed in order to identify strengths and weaknesses in both search interfaces. Thirty-six out of 40 participants preferred the TextFacet interface to the TextOnly interface. Using three dependent variables—time, accuracy, and rating—the two interfaces were compared and interactions were tested with the three task types. Search times for the TextFacet were shorter and participants preferred the TextFacet search interface over the TextOnly search interface. Performances across the three task types were different in terms of search time. The partially known and exploratory task types showed similar distributions for rating and accuracy. These distributions were distinctly different from the known task type. The results of this study may assist libraries in developing improved library catalog search interfaces that utilize facets as well as text searching.

Rapid increase in global competition demands increased protection of intellectual property rights and underlines the importance of patents as major intellectual property documents. Prior art patent search is the task of identifying related patents for a given patent file, and is an essential step in judging the validity of a patent application. This article proposes an automated query generation and postprocessing method for prior art patent search. The proposed approach first constructs structured queries by combining terms extracted from different fields of a query patent and then reranks the retrieved patents by utilizing the International Patent Classification (IPC) code similarities between the query patent and the retrieved patents along with the retrieval score. An extensive set of empirical results carried out on a large-scale, real-world dataset shows that utilizing 20 or 30 query terms extracted from all fields of an original query patent according to their log(tf)idf values helps form a representative search query out of the query patent and is found to be more effective than is using any number of query terms from any single field. It is shown that combining terms extracted from different fields of the query patent by giving higher importance to terms extracted from the abstract, claims, and description fields than to terms extracted from the title field is more effective than treating all extracted terms equally while forming the search query. Finally, utilizing the similarities between the IPC codes of the query patent and retrieved patents is shown to be beneficial to improve the effectiveness of the prior art search.

We characterize the research performance of a large number of institutions in a two-dimensional coordinate system based on the shapes of their h-cores so that their relative performance can be conveniently observed and compared. The 2D distribution of these institutions is then utilized (1) to categorize the institutions into a number of qualitative groups revealing the nature of their performance, and (2) to determine the position of a specific institution among the set of institutions. The method is compared with some major h-type indices and tested with empirical data using clinical medicine as an illustrative case. The method is extensible to the research performance evaluation at other aggregation levels such as researchers, journals, departments, and nations.

Since the end of World War II, the English language has become the lingua franca of science publications worldwide. Science publications written in other languages do not gain the same exposure to the international scientific community as does the material in English. In this sense, non-English articles constitute an “invisible science” for the rest of the scientific world. This study compares publications indexed in the academic-oriented Hebrew Index of Periodicals (IHP) database with those in the Science Citation Index Expanded (SCIE) in order to document the amount of scientific material published in Israel, where Hebrew is the native language. Except for abstracts, which are sometimes given in English, as well as Hebrew, and therefore provide some idea of a paper's content, most of this research remains hidden from the international scientific community. The SCIE and IHP databases for our examination cover the three grand disciplines: the exact and life sciences, the social sciences, and the humanities. Additionally, the study probes the coverage of medical publications in the two databases. The difference between old and emerging disciplines in the use of a language other than Hebrew is observed and non-English citation patterns for various disciplines are examined. The results confirm the dominance of English as the lingua franca of science and point to the large number of scientific studies in Hebrew that lack international exposure.

We conducted a fine-grained prosopography of six distinguished information scientists to explore commonalities and differences in their approaches to scholarly production at different stages of their careers. Specifically, we gathered data on authors' genre preferences, rates and modes of scholarly production, and coauthorship patterns. We also explored the role played by gender and place in determining mentoring and collaboration practices across time. Our biobibliometric profiles of the sextet reveal the different shapes a scholar's career can take. We consider the implications of our findings for new entrants into the academic marketplace.

This study enhances main path analysis by proposing several variants to the original approach. Main path analysis is a bibliometric method capable of tracing the most significant paths in a citation network and is commonly used to trace the development trajectory of a research field. We highlight several limitations of the original main path analysis and suggest new, complementary approaches to overcome these limitations. In contrast to the original local main path, the new approaches generate the global main path, the backward local main path, multiple main paths, and key-route main paths. Each of them is obtained via a perspective different from the original approach. By simultaneously conducting the new, complementary approaches, one uncovers the key development of the target discipline from a broader view. To demonstrate the value of these new approaches, we simultaneously apply them to a set of academic articles related to the Hirsch index. The results show that the integrated approach discovers several paths that are not captured by the original approach. Among these new approaches, the key-route approach is especially useful and hints at a divergence–convergence–divergence structure in the development of the Hirsch index.

This paper presents the results of a large-scale, qualitative study conducted in the homes of children aged 7, 9, and 11 investigating internet searching processes on Google. Seven search roles, representing distinct behavior patterns displayed by children when interacting with the Google search engine, are described, including Developing Searchers, Domain-specific Searchers, Power Searchers, Nonmotivated Searchers, Distracted Searchers, Rule-bound Searchers, and Visual Searchers. Other trends are described and selected to present a view of the whole child searcher. These roles and trends are used to make recommendations to designers, researchers, educators, and parents about the directions to take when considering how to best aid children to become search literate.

Multivariate linear regression models suggest a trade-off in allocations of national research and development (R&D). Government funding and spending in the higher education sector encourage publications as a long-term research benefit. Conversely, other components such as industrial funding and spending in the business sector encourage patenting. Our results help explain why the United States trails the European Union in publications: The focus in the United States is on industrial funding—some 70% of its total R&D investment. Likewise, our results also help explain why the European Union trails the United States in patenting, since its focus on government funding is less effective than industrial funding in predicting triadic patenting. Government funding contributes negatively to patenting in a multiple regression, and this relationship is significant in the case of triadic patenting. We provide new forecasts about the relationships of the United States, the European Union, and China for publishing; these results suggest much later dates for changes than previous forecasts because Chinese growth has been slowing down since 2003. Models for individual countries might be more successful than regression models whose parameters are averaged over a set of countries because nations can be expected to differ historically in terms of the institutional arrangements and funding schemes.

Bioinformatics journals publish research findings of intellectual synergies among subfields such as biology, mathematics, and computer science. The objective of this study is to characterize the citation patterns in bioinformatics journals and their correspondent knowledge subfields. Our study analyzed bibliometric data (impact factor, cited-half-life, and references-per-article) of bioinformatics journals and their related subfields collected from the Journal Citation Reports (JCR). The findings showed that bioinformatics journals' citations are field-dependent, with scattered patterns in article life span and citing propensity. Bioinformatics journals originally derived from biology-related subfields have shorter article life spans, more citing on average, and higher impact factors. Those journals, derived from mathematics and statistics, demonstrate converse citation patterns. Journal impact factors were normalized, taking into account the impacts of article life spans and citing propensity. A comparison of these normalized factors to JCR journal impact factors showed rearrangements in the ranking orders of a number of individual journals, but a high overall correlation with JCR impact factors.

The user-centered approach to information retrieval emphasizes the importance of a user model in determining what information will be most useful to a particular user, given their context. Mediated search provides an opportunity to elaborate on this idea, as an intermediary's elicitations reveal what aspects of the user model they think are worth inquiring about. However, empirical evidence is divided over whether intermediaries actually work to develop a broadly conceived user model. Our research revisits the issue in a web research services setting, whose characteristics are expected to result in more thorough user modeling on the part of intermediaries. Our empirical study confirms that intermediaries engage in rich user modeling. While intermediaries behave differently across settings, our interpretation is that the underlying user model characteristics that intermediaries inquire about in our setting are applicable to other settings as well.

YouTube is one of the world's most popular websites and hosts numerous amateur and professional videos. Comments on these videos might be researched to give insights into audience reactions to important issues or particular videos. Yet, little is known about YouTube discussions in general: how frequent they are, who typically participates, and the role of sentiment. This article fills this gap through an analysis of large samples of text comments on YouTube videos. The results identify patterns and give some benchmarks against which future YouTube research into individual videos can be compared. For instance, the typical YouTube comment was mildly positive, was posted by a 29-year-old male, and contained 58 characters. About 23% of comments in the complete comment sets were replies to previous comments. There was no typical density of discussion on YouTube videos in the sense of the proportion of replies to other comments: videos with both few and many replies were common. The YouTube audience engaged with each other disproportionately when making negative comments, however; positive comments elicited few replies. The biggest trigger of discussion seemed to be religion, whereas the videos attracting the least discussion were predominantly from the Music, Comedy, and How to & Style categories. This suggests different audience uses for YouTube, from passive entertainment to active debating.

Few studies have been done concerning document components and their effects on information use. This research empirically tested a taxonomy of functional units in a prototype journal system. This taxonomy was developed by identifying functions of the smallest information units within four journal article components (i.e., introduction, methods, results, discussion), and their associations with information tasks of using journal articles. Experimental results show that functional units can be utilized in supporting navigation, close reading, comprehension, and information use of journal articles to various extents. The results provide evidence that an individual functional unit has varying relevance to information use tasks, and has varying relevance to other functional units in the same or another component for a particular task. This research suggests that the information within a journal article can be organized and presented by functions to enhance effectiveness and efficiency in reading process and reading outcome.

This study replicates a previous study based on work in psychology, which demonstrates that students who score as below proficient in information literacy (IL) skills have a miscalibrated self-view of their ability. Simply stated, these students tend to believe that they have above-average IL skills, when, in fact, an objective test of their ability indicates that they are below-proficient in terms of their actual skills. This investigation was part of an Institute of Museum and Library Services-funded project and includes demographic data about participants, their scores on an objective test of their information literacy skills, and self-estimates of their ability. Findings support previous research that indicates many students come to college without proficient IL skills, that students with below-proficient IL skills have inflated views of their ability, and that this miscalibration can also be expressed by students who test as proficient. Implications for research and practice are discussed.

A critical part of a scientific activity is to discern how a new idea is related to what we know and what may become possible. As the number of new scientific publications arrives at a rate that rapidly outpaces our capacity of reading, analyzing, and synthesizing scientific knowledge, we need to augment ourselves with information that can effectively guide us through the rapidly growing intellectual space. In this article, we address a fundamental issue concerning what kinds of information may serve as early signs of potentially valuable ideas. In particular, we are interested in information that is routinely available and derivable upon the publication of a scientific paper without assuming the availability of additional information such as its usage and citations. We propose a theoretical and computational model that predicts the potential of a scientific publication in terms of the degree to which it alters the intellectual structure of the state of the art. The structural variation approach focuses on the novel boundary-spanning connections introduced by a new article to the intellectual space. We validate the role of boundary-spanning in predicting future citations using three metrics of structural variation—namely, modularity change rate, cluster linkage, and Centrality Divergence—along with more commonly studied predictors of citations such as the number of coauthors, the number of cited references, and the number of pages. Main effects of these factors are estimated for five cases using zero-inflated negative binomial regression models of citation counts. Key findings indicate that (a) structural variations measured by cluster linkage are a better predictor of citation counts than are the more commonly studied variables such as the number of references cited, (b) the number of coauthors and the number of references are both good predictors of global citation counts to a lesser extent, and (c) the Centrality Divergence metric is potentially valuable for detecting boundary-spanning activities at interdisciplinary levels. The structural variation approach offers a new way to monitor and discern the potential of newly published papers in context. The boundary-spanning mechanism offers a conceptually simplified and unifying explanation of the roles played by commonly studied extrinsic properties of a publication in the study of citation behavior.

Domestic citation to papers from the same country and the greater citation impact of documents involving international collaboration are two phenomena that have been extensively studied and contrasted. Here, however, we show that it is not so much a national bias, but that papers have a greater impact on their immediate environments, an impact that is diluted as that environment grows. For this reason, the greatest biases are observed in countries with a limited production. Papers that involve international collaboration have a greater impact in general, on the one hand, because they have multiple “immediate environments,” and on the other because of their greater quality or prestige. In short, one can say that science knows no frontiers. Certainly there is a greater impact on the authors' immediate environment, but this does not necessarily have to coincide with their national environments, which fade in importance as the collaborative environment expands.

Online communication is an indispensable tool for communication and management. The network structure of communication is considered to affect team and individual performances, but it has not been not empirically tested. In this article, we collected a set of 1-month e-mail logs of a company and conducted an e-mail network analysis. We calculated the network centralities of 72 managerial candidates, and investigated the relationship between positions in the network and leadership performance with partial least squares structural equation modeling. Betweenness and in-degree network centralities of those middle managers are correlated with their leadership performance; on the other hand, for this management group, out-degree has no correlation, and PageRank is a negative indicator of leadership. Leaders with high performance are trusted in their communities as a hub of the information channel of the communication network.

In Web 2.0 environments, people commonly share their knowledge and personal experiences with others, but little is known about their background characteristics and motivations. Thus, the current study examines some of the characteristics and motivations common among answerers, who produce health-related answers to questions asked by anonymous others in a social Q&A site, Yahoo! Answers. An online survey questionnaire was distributed to top and recent answerers to investigate their demographics, areas of health expertise, and other characteristics related to answering behaviors online. Also, 10 motivation factors are proposed and tested in the survey: enjoyment, efficacy, learning, personal gain, altruism, community interest, social engagement, empathy, reputation, and reciprocity. Findings show that altruism is the most influential motivation, while personal gain is the least. Enjoyment and efficacy are more influential than other social motivations, such as reputation or reciprocity, although there are some variations across different groups of answerers. Motivational factors among top answerers or health experts are further analyzed. The findings of this study have practical implications for promoting health answerers to share knowledge and experiences in social contexts. Furthermore, the study design of the current study can be used to examine motivations of answerers in other topic areas as well as other social contexts.

The authors review five paradigms from the discipline of community sociology (functionalism, evolution, conflict, interactionism, and exchange) to assess their potential utility for understanding everyday life information behavior and technology use. Their analysis considers the ways in which each paradigm defines the concepts of community, information, and technology. It also explores the insights offered by each paradigm regarding relationships between community and both information and technology. Accordingly, the authors highlight the ways in which existing information behavior and informatics scholarship draws from similar conceptual roots. Key insights drawn from this research, as well as remaining gaps and research questions, are examined. Additionally, they consider the limitations of each approach. The authors conclude by arguing for the value of a vigorous research program regarding information behavior and technology use in communities, particularly that which takes the community as the central unit of analysis. They consider key questions that could drive such a research program, as well as potentially fruitful conceptual and methodological approaches for this endeavor.

Literature-based discovery (LBD) refers to a particular type of text mining that seeks to identify nontrivial assertions that are implicit, and not explicitly stated, and that are detected by juxtaposing (generally a large body of) documents. In this review, I will provide a brief overview of LBD, both past and present, and will propose some new directions for the next decade. The prevalent ABC model is not “wrong”; however, it is only one of several different types of models that can contribute to the development of the next generation of LBD tools. Perhaps the most urgent need is to develop a series of objective literature-based interestingness measures, which can customize the output of LBD systems for different types of scientific investigations.

An efficient and robust medical-image indexing procedure should be user-oriented. It is essential to index the images at the right level of description and ensure that the indexed levels match the user's interest level. This study examines 240 medical-image descriptions produced by three different groups of medical-image users (novices, intermediates, and experts) in the area of radiography. This article reports several important findings: First, the effect of domain knowledge has a significant relationship with the use of semantic image attributes in image-users' descriptions. We found that experts employ more high-level image attributes which require high-reasoning or diagnostic knowledge to search for a medical image (Abstract Objects and Scenes) than do novices; novices are more likely to describe some basic objects which do not require much radiological knowledge to search for an image they need (Generic Objects) than are experts. Second, all image users in this study prefer to use image attributes of the semantic levels to represent the image that they desired to find, especially using those specific-level and scene-related attributes. Third, image attributes generated by medical-image users can be mapped to all levels of the pyramid model that was developed to structure visual information. Therefore, the pyramid model could be considered a robust instrument for indexing medical imagery.

This study proposes an approach for visualizing knowledge structures that creates a “research-focused parallelship network,” “keyword co-occurrence network,” and a knowledge map to visualize Sci-Tech policy research structure. A total of 1,125 Sci-Tech policy-related papers (873 journal papers [78%], 205 conference papers [18%], and 47 review papers [4%]) have been retrieved from the Web of Science database for quantitative analysis and mapping. Different network and contour maps based on these 1,125 papers can be constructed by choosing different information as the main actor, such as the paper title, the institute, the country, or the author keywords, to reflect Sci-Tech policy research structures in micro-, meso-, and macro-levels, respectively. The quantitative way of exploring Sci-Tech policy research papers is investigated to unveil important or emerging Sci-Tech policy implications as well as to demonstrate the dynamics and visualization of the evolution of Sci-Tech policy research.

Social media is frequently used as a platform for the exchange of information and opinions as well as propaganda dissemination. But online content can be misused for the distribution of illicit information, such as violent postings in web forums. Illicit content is highly distributed in social media, while non-illicit content is unspecific and topically diverse. It is costly and time consuming to label a large amount of illicit content (positive examples) and non-illicit content (negative examples) to train classification systems. Nevertheless, it is relatively easy to obtain large volumes of unlabeled content in social media. In this article, an artificial immune system-based technique is presented to address the difficulties in the illicit content identification in social media. Inspired by the positive selection principle in the immune system, we designed a novel labeling heuristic based on partially supervised learning to extract high-quality positive and negative examples from unlabeled datasets. The empirical evaluation results from two large hate group web forums suggest that our proposed approach generally outperforms the benchmark techniques and exhibits more stable performance.

As online communities grow and the volume of user-generated content increases, the need for community management also rises. Community management has three main purposes: to create a positive experience for existing participants, to promote appropriate, socionormative behaviors, and to encourage potential participants to make contributions. Research indicates that the quality of content a potential participant sees on a site is highly influential; off-topic, negative comments with malicious intent are a particularly strong boundary to participation or set the tone for encouraging similar contributions. A problem for community managers, therefore, is the detection and elimination of such undesirable content. As a community grows, this undertaking becomes more daunting. Can an automated system aid community managers in this task? In this paper, we address this question through a machine learning approach to automatic detection of inappropriate negative user contributions. Our training corpus is a set of comments from a news commenting site that we tasked Amazon Mechanical Turk workers with labeling. Each comment is labeled for the presence of profanity, insults, and the object of the insults. Support vector machines trained on these data are combined with relevance and valence analysis systems in a multistep approach to the detection of inappropriate negative user contributions. The system shows great potential for semiautomated community management.

In plagiarism detection (PD) systems, two important problems should be considered: the problem of retrieving candidate documents that are globally similar to a document q under investigation, and the problem of side-by-side comparison of q and its candidates to pinpoint plagiarized fragments in detail. In this article, the authors investigate the usage of structural information of scientific publications in both problems, and the consideration of citation evidence in the second problem. Three statistical measures namely Inverse Generic Class Frequency, Spread, and Depth are introduced to assign a degree of importance (i.e., weight) to structural components in scientific articles. A term-weighting scheme is adjusted to incorporate component-weight factors, which is used to improve the retrieval of potential sources of plagiarism. A plagiarism screening process is applied based on a measure of resemblance, in which component-weight factors are exploited to ignore less or nonsignificant plagiarism cases. Using the notion of citation evidence, parts with proper citation evidence are excluded, and remaining cases are suspected and used to calculate the similarity index. The authors compare their approach to two flat-based baselines, TF-IDF weighting with a Cosine coefficient, and shingling with a Jaccard coefficient. In both baselines, they use different comparison units with overlapping measures for plagiarism screening. They conducted extensive experiments using a dataset of 15,412 documents divided into 8,657 source publications and 6,755 suspicious queries, which included 18,147 plagiarism cases inserted automatically. Component-weight factors are assessed using precision, recall, and F-measure averaged over a 10-fold cross-validation and compared using the ANOVA statistical test. Results from structural-based candidate retrieval and plagiarism detection are evaluated statistically against the flat baselines using paired-t tests on 10-fold cross-validation runs, which demonstrate the efficacy achieved by the proposed framework. An empirical study on the system's response shows that structural information, unlike existing plagiarism detectors, helps to flag significant plagiarism cases, improve the similarity index, and provide human-like plagiarism screening results.

Although Web 2.0 contains many tools with different functionalities, they all share a common social nature. One tool in particular, social bookmarking systems (SBSs), allows users to store and share links to different types of resources, i.e., websites, videos, images. To identify and classify these resources so that they can be retrieved and shared, fragments of text are used. These fragments of text, usually words, are called tags. A tag that is found on the inside of a resource text is referred to as an obvious or explicit tag. There are also nonobvious or implicit tags, which don't appear in the resource text. The purpose of this article is to describe the present situation of the SBSs tool and then to also determine the principal features of and how to use explicit tags. It will be taken into special consideration which HTML tags with explicit tags are used more frequently.

International collaboration is being heralded as the hallmark of contemporary scientific production. Yet little quantitative evidence has portrayed the landscape and trends of such collaboration. To this end, 14,000,000 documents indexed in Thomson Reuters's Web of Science (WoS) were studied to provide a state-of-the-art description of scientific collaborations across the world. The results indicate that the number of authors in the largest research teams have not significantly grown during the past decade; however, the number of smaller research teams has seen significant increases in growth. In terms of composition, the largest teams have become more diverse than the latter teams and tend more toward interinstitutional and international collaboration. Investigating the size of teams showed large variation between fields. Mapping scientific cooperation at the country level reveals that Western countries situated at the core of the map are extensively cooperating with each other. High-impact institutions are significantly more collaborative than others. This work should inform policy makers, administrators, and those interested in the progression of scientific collaboration.

In an effort to understand how academic scientists seek information relevant to their research in today's environment of ubiquitous electronic access, a correlation framework is built and regression analysis is applied to the survey results from 2,063 academic researchers in natural science, engineering, and medical science at five research universities in the United States. Previous work has reported descriptive statistics about these scientists' information-seeking behavior. This study extends that work to examine relationships between scientists' information-seeking behaviors and their personal and environmental factors. Several regression models, including the Poisson model, the logit model, and the ordered logit model, are built to interpret the correlation among scientists' behaviors. In addition, exploratory factor analysis is used for data reduction. Overall, many factors were found to affect the specific information-seeking behaviors of scientists, including demographic, psychological, role-related, and environmental factors. Of the factors having an effect, academic position was the most important determinant of information behavior.

The goal in blog search is to rank blogs according to their recurrent relevance to the topic of the query. State-of-the-art approaches view it as an expert search or resource selection problem. We investigate the effect of content-based similarity between posts on the performance of the retrieval system. We test two different approaches for smoothing (regularizing) relevance scores of posts based on their dependencies. In the first approach, we smooth term distributions describing posts by performing a random walk over a document-term graph in which similar posts are highly connected. In the second, we directly smooth scores for posts using a regularization framework that aims to minimize the discrepancy between scores for similar documents. We then extend these approaches to consider the time interval between the posts in smoothing the scores. The idea is that if two posts are temporally close, then they are good sources for smoothing each other's relevance scores. We compare these methods with the state-of-the-art approaches in blog search that employ Language Modeling-based resource selection algorithms and fusion-based methods for aggregating post relevance scores. We show performance gains over the baseline techniques which do not take advantage of the relation between posts for smoothing relevance estimates.

This article presents a search-intent-based method to generate pornographic blacklists for collaborative cyberporn filtering. A novel porn-detection framework that can find newly appearing pornographic web pages by mining search query logs is proposed. First, suspected queries are identified along with their clicked URLs by an automatically constructed lexicon. Then, a candidate URL is determined if the number of clicks satisfies majority voting rules. Finally, a candidate whose URL contains at least one categorical keyword will be included in a blacklist. Several experiments are conducted on an MSN search porn dataset to demonstrate the effectiveness of our method. The resulting blacklist generated by our search-intent-based method achieves high precision (0.701) while maintaining a favorably low false-positive rate (0.086). The experiments of a real-life filtering simulation reveal that our proposed method with its accumulative update strategy can achieve 44.15% of a macro-averaging blocking rate, when the update frequency is set to 1 day. In addition, the overblocking rates are less than 9% with time change due to the strong advantages of our search-intent-based method. This user-behavior-oriented method can be easily applied to search engines for incorporating only implicit collective intelligence from query logs without other efforts. In practice, it is complementary to intelligent content analysis for keeping up with the changing trails of objectionable websites from users' perspectives.

Understanding search behavior is important and leads to more effective interfaces that support searchers throughout the search process. In this article, through an observational user study, we investigate the search behavior of 15 visually impaired and 15 sighted searchers while they complete complex search tasks online. We study complex search tasks because they are challenging, cognitively intensive and affect performance of searchers. We compare the behavior of the two groups of searchers at four stages of the information-seeking process namely, Query Formulation, Search Results Exploration, Query Reformulation, and Search Results Management. For each stage, we identify research questions to investigate the impact of speech-based screen readers on the information-seeking behavior of visually impaired users. Significant differences were observed during query formulation and in the use of query-level support features such as query suggestions and spelling suggestions. In addition, screen-reader users submitted a lower number of queries and displayed comparatively limited exploratory behavior during search results exploration. We investigate how a lack of visual cues affected visually impaired searchers' approach towards query reformulation and observed different strategies to manage and use information encountered during the search process. We discuss the implications that our findings have for the design of search interfaces and propose a set of design guidelines to consider when designing interfaces that are usable and accessible with screen readers. This work also enhances our understanding of search behavior when using an auditory interface and could be useful when designing audio-based information retrieval systems.

Web accessibility (WA) is an innovation in Web design; it can be considered as part of the corporate social responsibility (CSR) strategy of the firms. As adoption of innovations and CSR commitment are linked with firm size and national culture/legislation, we hypothesize that size and national culture/legislation, may have an effect on WA level. The authors studied an international sample made up of companies included in EUROSTOXX600 (The STOXX Europe 600 Index). The main results suggest that both size and culture have a significant effect on WA. Large firms as well as Anglo-Saxon companies are more prone to have higher WA levels. A deeper analysis, which was done through the estimation of quantile regression equations, showed that the influence of size is significant for companies trying to excel or for those trying to avoid the worst WA. However, the effect of size is significant only in the lowest part of the conditional distribution.

The h-index is a popular bibliometric indicator for assessing individual scientists. We criticize the h-index from a theoretical point of view. We argue that for the purpose of measuring the overall scientific impact of a scientist (or some other unit of analysis), the h-index behaves in a counterintuitive way. In certain cases, the mechanism used by the h-index to aggregate publication and citation statistics into a single number leads to inconsistencies in the way in which scientists are ranked. Our conclusion is that the h-index cannot be considered an appropriate indicator of a scientist's overall scientific impact. Based on recent theoretical insights, we discuss what kind of indicators can be used as an alternative to the h-index. We pay special attention to the highly cited publications indicator. This indicator has a lot in common with the h-index, but unlike the h-index it does not produce inconsistent rankings.