Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).
Paper Session 10: Information Organization [SDGs 4, 9, 11, 12]
4:00pm - 5:30pm
Session Chair: Ann Graf, Simmons University, United States of America
4:00pm - 4:15pm ID: 107 / PS-10: 1 Long Papers Topics: Library and Information Science Keywords: Diffusion of innovations, libraries, linked data, innovation adoption
Diffusion and Adoption of Linked Data Among Libraries
University of South Florida, USA
Through content analyses of literature about the Linked Data adoption of individual libraries, this study found that the diffusion of Linked Data among libraries is a decentralized process with high-degree of reinvention and a continuous process that includes multiple stages and might last for many years. The diffusion of Linked Data among libraries involves the diffusion of numerous related innovations, follows three paths (inter-library, intra-library, and inter-librarian), and is facilitated by four types of institutions (professional associations, vendors, external funders, and leading libraries) and three factors (avoiding re-inventing the wheel, standardization, and commercialization).
4:15pm - 4:30pm ID: 302 / PS-10: 2 Long Papers Topics: Human Computer Interaction (HCI) Keywords: Social tagging, user interfaces, text entry, folksonomies, classification
The Effects of Suggested Tags and Autocomplete Features on Social Tagging Behaviors
University of Washington, USA
Many websites employ social tagging to allow users to label and classify information. These tagging user interfaces use a variety of features to support efficient and consistent tag creation, including suggested tags and autocomplete for tags. This study uses a custom-built tagging interface in a controlled experiment to determine how these features affect social tagging behavior. The study finds that suggested tags do not have a significant effect on the number of tags, number of unique tags, number of typos, or time elapsed per tagged provided. However, autocomplete significantly increases the number and consistency of tags provided, significantly decreases the rate of typos, and significantly decreases the elapsed time per tag provided. These findings for the autocomplete feature align with the priorities and constraints of social tagging folksonomies that support retrieval and site navigation and suggest that autocomplete is an important aid for text entry in social tagging user interfaces.
4:30pm - 4:40pm ID: 240 / PS-10: 3 Short Papers Topics: Data Science; Analytics; and Visualization Keywords: Content analysis, Annotation quality, Text classification.
Using Text Classification to Improve Annotation Quality by Improving Annotator Consistency
Emi Ishita1, Satoshi Fukuda1, Yoichi Tomiura1, Douglas W. Oard2
1Kyushu University, Japan; 2University of Maryland, College Park, USA
This paper presents results of experiments in which annotators were asked to selectively reexamine their decisions when those decisions seemed inconsistent. The annotation task was binary topic classification. To operationalize the concept of annotation consistency, a text classifier was trained on all manual annotations made during a complete first pass and then used to automatically recode every document. Annotators were then asked to perform a second manual pass, limiting their attention to cases in which their first annotation disagreed with the text classifier. On average across three annotators, each working independently, 11% of first pass annotations were reconsidered, 46% of reconsidered annotations were changed in the second pass, and 71% of changed annotations agreed with decisions made independently by an experienced fourth annotator. The net result was that for an 11% average increase in annotation cost it was possible to increase overall chance corrected agreement with the annotation decisions of an experienced annotator (as measured by kappa) from 0.69 to 0.73.
4:40pm - 4:50pm ID: 269 / PS-10: 4 Short Papers Topics: Archives; Data Curation; and Preservation Keywords: language archive metadata, free-text metadata, comparative analysis, metadata quality
Descriptive Richness of Free-Text Metadata: A Comparative Analysis of Three-Language Archives
Mary Burke, Oksana L. Zavalina
University of North Texas, USA
As archiving became a priority in documentary linguistics only in the 1990’s, existing research indicates that language archives are not yet up to date on best practices in information organization. As a result, metadata in language archives varies substantially, depending on the depositor, self-upload procedures, and metadata creation guidelines. Focusing on free-text metadata, known to provide rich information, this study analyzes item-level metadata in three language archives: the Endangered Language Archive, Pacific Regional Archive for Digital Sources in Endangered Cultures, and the Archive of the Indigenous Languages of Latin America. The study identified the categories of information included in Description metadata fields and the relative distribution of these categories. The most commonly occurring categories of information observed in this study can serve as a basis for the development of best practice guidelines for item-level metadata in language archives.