Search

Researchers seeking relevant data across disciplines confront the challenge of navigating technical descriptions. How can curation support the serendipitous discovery of related research data? Everyday spaces like bookshelves are designed to support browsing and exploration by placing similar resources closer together. Space and time are foundational ordering relations for knowledge organization. I ask how this ordering, which is well-established in the geographic context, can be translated to locate and organize research data in abstract topic spaces. This dissertation develops methods for making the latent topics of research metadata explicit. These methods produce spatial configurations where related research topics are co-located in neighborhoods. This has the potential to support serendipitous discovery by offering researchers ways to discover related data. I test this notion in three studies that develop topic spaces for research data curation. The first part of this dissertation in Chapter 2 focuses on supporting research data discovery with a common terminology. I develop a crosscutting base vocabulary of geospatial topics to help users discover related government data in a ubiquitous open civic data platform. Semantic annotation expands search terms by mapping users’ vernacular onto the language of metadata. In the second part of this dissertation, I shift away from addressing terminological search to supporting spatial curation by developing topic spaces. In Chapter 3, I develop two kinds of topic spaces for curating research theses and dissertations: landscapes and networks. I use topic modeling to determine the latent semantic similarity of research metadata and then produce topic spaces from these using spatialization techniques. In Chapter 4, I spatialize an institute’s multidisciplinary body of research, producing topic maps at two distinct levels of detail. Emerging spatial patterns, like centrality and proximity, support high-level narratives about cross-disciplinary research activities that complement the quantitative metrics currently cited in reviews of institutional research. Together, these three studies demonstrate strategies for developing topic spaces in which diverse, yet related, multidisciplinary research data are curated. Future research will extend these methods by tracing the impact of specific curatorial actions contributing to research data discovery and reuse.

It is challenging for scholars to discover thematically related research in a multidisciplinary setting, such as that of a university library. In this work, we use spatialization techniques to convey the relatedness of research themes without requiring scholars to have specific knowledge of disciplinary search terminology. We approach this task conceptually by revisiting existing spatialization techniques and reframing them in terms of core concepts of spatial information, highlighting their different capacities. To apply our design, we spatialize masters and doctoral theses (two kinds of research objects available through a university library repository) using topic modeling to assign a relatively small number of research topics to the objects. We discuss and implement two distinct spaces for exploration: a field view of research topics and a network view of research objects. We find that each space enables distinct visual perceptions and questions about the relatedness of research themes. A field view enables questions about the distribution of research objects in the topic space, while a network view enables questions about connections between research objects or about their centrality. Our work contributes to spatialization theory a systematic choice of spaces informed by core concepts of spatial information. Its application to the design of library discovery tools offers two distinct and intuitive ways to gain insights into the thematic relatedness of research objects, regardless of the disciplinary terms used to describe them.

We describe a method and system design for improved data discovery in an integrated network of open geospatial data that supports collaborative policy development between governments and local constituents. Metadata about civic data (such as thematic categories, user-generated tags, geo-references, or attribute schemata) primarily rely on technical vocabularies that reflect scientific or organizational hierarchies. By contrast, public consumers of data often search for information using colloquial terminology that does not align with official metadata vocabularies. For example, citizens searching for data about bicycle collisions in an area are unlikely to use the search terms with which organizations like Departments of Transportation describe relevant data. Users may also search with broad terms, such as "traffic safety", and will then not discover data tagged with narrower official terms, such as "vehicular crash". This mismatch raises the question of how to bridge the users' ways of talking and searching with the language of technical metadata. In similar situations, it has been beneficial to augment official metadata with semantic annotations that expand the discoverability and relevance recommendations of data, supporting more inclusive access. Adopting this strategy, we develop a method for automated semantic annotation, which aggregates similar thematic and geographic information. A novelty of our approach is the development and application of a crosscutting base vocabulary that supports the description of geospatial themes. The resulting annotation method is integrated into a novel open access collaboration platform (Esri's ArcGIS Hub) that supports public dissemination of civic data and is in use by thousands of government agencies. Our semantic annotation method improves data discovery for users across organizational repositories and has the potential to facilitate the coordination of community and organizational work, improving the transparency and efficacy of government policies.

of conference oral presentation on Using specimen citations to show the impact of natural history collections: metrics, machine learning, and manual annotation by Assistant Professor Andrea Thomer ofUniversity of Michigan Ann Arbor United States for SPNHC2022 Conference, Edinburgh. June 5-10th 2022.

Assessing the reach and impact of natural history collections requires reviewing scientific literature for citations to specimens, which can be labor-intensive. This project curates and analyzes a bibliography of literature citing specimens from the University of Michigan Museum of Zoology. We demonstrate a machine learning approach we are developing to detect specimen citations in scientific literature with the goal of enabling comprehensive studies of specimen reuse patterns across multiple natural history collections. This poster was presented at the 2022 July ESIP Meeting in Pittsburgh, PA.

The Inter-university Consortium for Political and Social Research (ICPSR) is developing a computational approach to detect informal data use and construct reliable data impact metrics. Formal data citations that use unique identifiers are readily discoverable; however, informal references made to data are challenging to infer and detect as they are described in many ways and tend to occur in article footnotes, tables, figures, or elsewhere where they are not indexed for search. Identifying data citations is an essential step toward characterizing the impact of research data (i.e., who reuses research data and for what purposes). We use features of text including the presence of indicator terms, sections of articles, and frequency of acronyms, to predict the portions of articles that are likely to indicate data use. We then use a natural language processing (NLP) pipeline to extract candidate data references. In production, our model will support the review of publications to ingest into the ICPSR Bibliography of Data-related Literature as part of a broader effort to measure the impact of research data.

Academic libraries have always supported research across disciplines by integrating access to diverse contents and resources. They now have the opportunity to reinvent their role in facilitating interdisciplinary work by offering researchers new ways of sharing, curating, discovering, and linking research data. Spatial data and metadata support this process because location often integrates disciplinary perspectives, enabling researchers to make their own research data more discoverable, to discover data of other researchers, and to integrate data from multiple sources. The Center for Spatial Studies at the University of California, Santa Barbara (UCSB) and the UCSB Library are undertaking joint research to better enable the discovery of research data and publications. The research addresses the question of how to spatially enable data discovery in a setting that allows for mapping and analysis in a GIS while connecting the data to publications about them. It suggests a framework for an integrated data discovery mechanism and shows how publications may be linked to associated data sets exposed either directly or through metadata on Esri’s Open Data platform. The results demonstrate a simple form of linking data to publications through spatially referenced metadata and persistent identifiers. This linking adds value to research products and increases their discoverability across disciplinary boundaries. Current data publishing practices in academia result in datasets that are not easily discovered, hard to integrate across domains, and typically not linked to publications about them. For example, discovering that two datasets, such as archaeological observations and specimen data collections, share a spatial extent in Mesoamerica, is not currently supported, nor is it easy to get from those data sets to relevant publications or other documents. In our previous work, we had developed a basic linked metadata model relating spatially referenced datasets to documents. The research reported here applies the model to a collection of spatially referenced researcher datasets, capturing metadata and encoding them as linked open data. We use existing RDF vocabularies to triplify the metadata, to make them spatially explicit, and to link them thematically. Our latest research has produced a simple and extensible method for exposing metadata of research objects as a library service and for spatially integrating collections across repositories.

The Inter-university Consortium for Political and Social Research (ICPSR) is developing a machine learning approach using natural language processing (NLP) to assist in the detection of informal data references. Formal data citations that reference unique identifiers are readily discoverable; however, informal references indicating research data reuse are challenging to infer and detect. We contribute a model that uses a combination of cues, such as the presence of indicator terms and syntactical patterns, to assign a likelihood score to dataset mentions and extract candidate data citations from academic text. In production, the model will support the evaluation of candidate documents for ingest into the ICPSR Bibliography of Data-related Literature. This work supports a larger effort to measure the impact of research data.

This project, “Measuring and Improving the Efficacy of Curation Activities in Data Archives” (MICA), is a three-year National Digital Infrastructures and Initiatives project led by investigators at the University of Michigan School of Information in partnership with the Inter-university Consortium for Political and Social Research. We are studying how curatorial actions impact the use of digital collections. Specifically, we are assessing stakeholder needs, priorities, and values for data reuse and creating curatorial metrics for measuring the impact of curation activities. This poster will present our ongoing efforts to leverage machine learning to track data citations and curation activities. This poster was presented at the 2021 Earth Science Information Partners (ESIP) Winter Meeting held virtually in January 2021.

This presentation, originally given during a webinar on February 19th, 2021, provides an overview of plenary and breakout sessions from the 2021 Earth Science Information Partners (ESIP) Winter Meeting held online (January 26th-29th, 2021).

Researchers seeking relevant data across disciplines confront the challenge of navigating technical descriptions. How can curation support the serendipitous discovery of related research data? Everyday spaces like bookshelves are designed to support browsing and exploration by placing similar resources closer together. Space and time are foundational ordering relations for knowledge organization. I ask how this ordering, which is well-established in the geographic context, can be translated to locate and organize research data in abstract topic spaces. This dissertation develops methods for making the latent topics of research metadata explicit. These methods produce spatial configurations where related research topics are co-located in neighborhoods. This has the potential to support serendipitous discovery by offering researchers ways to discover related data. I test this notion in three studies that develop topic spaces for research data curation. The first part of this dissertation in Chapter 2 focuses on supporting research data discovery with a common terminology. I develop a crosscutting base vocabulary of geospatial topics to help users discover related government data in a ubiquitous open civic data platform. Semantic annotation expands search terms by mapping users’ vernacular onto the language of metadata. In the second part of this dissertation, I shift away from addressing terminological search to supporting spatial curation by developing topic spaces. In Chapter 3, I develop two kinds of topic spaces for curating research theses and dissertations: landscapes and networks. I use topic modeling to determine the latent semantic similarity of research metadata and then produce topic spaces from these using spatialization techniques. In Chapter 4, I spatialize an institute’s multidisciplinary body of research, producing topic maps at two distinct levels of detail. Emerging spatial patterns, like centrality and proximity, support high-level narratives about cross-disciplinary research activities that complement the quantitative metrics currently cited in reviews of institutional research. Together, these three studies demonstrate strategies for developing topic spaces in which diverse, yet related, multidisciplinary research data are curated. Future research will extend these methods by tracing the impact of specific curatorial actions contributing to research data discovery and reuse.

This study surveys the state of search on open geospatial data portals. We seek to understand 1) what users are able to control when searching for geospatial data, 2) how these portals process and interpret a user’s query, and 3) if and how user query reformulations alter search results. We find that most users initiate a search using a text input and several pre-created facets (such as a filter for tags or format). Some portals supply a map-view of data or topic explorers. To process and interpret queries, most portals use a vertical full-text search engine like Apache Solr to query data from a content-management system like CKAN. When processing queries, most portals initially filter results and then rank the remaining results using a common keyword frequency relevance metric (e.g., TF-IDF). Some portals use query expansion. We identify and discuss several recurring usability constraints across portals. For example, users are typically only given text lists to interact with search results. Furthermore, ranking is rarely extended beyond syntactic comparison of keyword similarity. We discuss several avenues for improving search for geospatial data including alternative interfaces and query processing pipelines.

Access to public data in the United States and elsewhere has steadily increased as governments have launched geospatially-enabled web portals like Socrata, CKAN, and Esri Hub. However, data discovery in these portals remains a challenge for the average user. Differences between users' colloquial search terms and authoritative metadata impede data discovery. For example, a motivated user with expertise can leverage valuable public data about transportation, real estate values, and crime, yet it remains difficult for the average user to discover and leverage data. To close this gap, community dashboards that use public data are being developed to track initiatives for public consumption; however, dashboards still require users to discover and interpret data. Alternatively, local governments are now developing data discovery systems that use voice assistants like Amazon Alexa and Google Home as conversational interfaces to public data portals. We explore these emerging technologies, examining the application areas they are designed to address and the degree to which they currently leverage existing open public geospatial data. In the context of ongoing technological advances, we envision using core concepts of spatial information to organize the geospatial themes of data exposed through voice assistant applications. This will allow us to curate them for improved discovery, ultimately supporting more meaningful user questions and their translation into spatial computations.

This 15 minute talk was given by Sara Lafia during ESIP's 2018 Summer Meeting in Tucson, AZ for the 2018 Raskin Scholarship. This talk gives an overview of Sara's research at the University of California, Santa Barbara.