Yinlin Chen

Superimposed image description and retrieval for fish species identification

Fish species identification is critical to the study of fish ecology and management of fisheries. Traditionally, dichotomous keys are used for fish identification. The keys consist of questions about the observed specimen. Answers to these questions lead to more questions till the reader identifies the specimen. However, such keys are incapable of adapting or changing to meet different fish identification approaches, and often do not focus upon distinguishing characteristics favored by many field ecologists and more user-friendly field guides. This makes learning to identify fish difficult for Ichthyology students. Students usually supplement the use of the key with other methods such as making personal notes, drawings, annotated fish images, and more recently, fish information websites, such as Fishbase. Although these approaches provide useful additional content, it is dispersed across heterogeneous sources and can be tedious to access. Also, most of the existing electronic tools have limited support to manage user created content, especially that related to parts of images such as markings on drawings and images and associated notes. We present SuperIDR, a superimposed image description and retrieval tool, developed to address some of these issues. It allows users to associate parts of images with text annotations. Later, they can retrieve images, parts of images, annotations, and image descriptions through text- and content-based image retrieval. We evaluated SuperIDR in an undergraduate Ichthyology class as an aid to fish species identification and found that the use of SuperIDR yielded a higher likelihood of success in species identification than using traditional methods, including the dichotomous key, fish web sites, notes, etc.

Ensemble PDP-8: eight principles for distributed portals

Ensemble, the National Science Digital Library (NSDL) Pathways project for Computing, builds upon a diverse group of prior NSDL, DL-I, and other projects. Ensemble has shaped its activities according to principles related to design, development, implementation, and operation of distributed portals. Here we articulate 8 key principles for distributed portals (PDPs). While our focus is on education and pedagogy, we expect that our experiences will generalize to other digital library application domains. These principles inform, facilitate, and enhance the Ensemble R&D and production activities. They allow us to provide a broad range of services, from personalization to coordination across communities. The eight PDPs can be briefly summarized as: (1) Articulation across communities using ontologies. (2) Browsing tailored to collections. (3) Integration across interfaces and virtual environments. (4) Metadata interoperability and integration. (5) Social graph construction using logging and metrics. (6) Superimposed information and annotation integrated across distributed systems. (7) Streamlined user access with IDs. (8) Web 2.0 multiple social network system interconnection.

Using ACM DL paper metadata as an auxiliary source for building educational collections

Some digital libraries harvest metadata records from multiple content providers to build their collections. However, the quality and quantity of such metadata records are limited by what is harvested. To ensure collection growth, and to expand the scope beyond just what can be harvested, additional content acquisition methods are needed. Accordingly, we discuss how the Ensemble project (a pathway effort in the NSDL) is broadening its collection with the help of machine learning. Since Ensemble aims to aid computing education, we make use of ACM Digital Library records as a resource to help with transfer learning. We have built classifiers that can identify if a potential additional resource is about computing education. We approached this as a cross-domain text classification problem and developed suitable methods for feature extraction and bootstrapping for classifier training. Our experiments on three datasets of computing education metadata records show our approach can enhance the quality and quantity of records being added to Ensemble.

Digital library 2.0 for educational resources

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.

Categorization of computing education resources with utilization of crowdsourcing

The Ensemble Portal harvests resources from multiple heterogeneous federated collections. Managing these dynamically increasing collections requires an automatic mechanism to categorize records in to corresponding topics. We propose an approach to use existing ACM DL metadata to build classifiers for harvested resources in the Ensemble project. We also present our experience with utilizing the Amazon Mechanical Turk platform to build ground truth training data sets from Ensemble collections.

Species identification: fish images with CBIR and annotations

Uma Murthy, Edward A. Fox, and Yinlin Chen Department of Computer Science, Virginia Tech Blacksburg, VA 24061, USA {umurthy, fox, ylchen}@vt.edu Eric Hallerman Department of Fisheries and Wildlife Scienes Virginia Tech Blacksburg, VA 24061, USA ehallerm@vt.edu Ricardo Torres, Evandro J. Ramos, and Tiago R. C. Falcão Institute of Computing University of Campinas Campinas, SP, Brail rtorres@ic.unicamp.br

Evaluating Cost of Cloud Execution in a Data Repository

In this paper, we utilize a set of controlled experiments to benchmark the cost associated with the cloud execution of typical repository functions such as ingestion, fixity checking, and heavy data processing. We focus on the repository service pattern where content is explicitly stored away from where it is processed. We measured the processing speed and unit cost of each scenario using a large sensor dataset and Amazon Web Services (AWS). The initial results reveal three distinct cost patterns: 1) spend more to buy up to proportionally faster services; 2) more money does not necessarily buy better performance; and 3) spend less, but faster. Further investigations into these performance and cost patterns will help repositories to form a more effective operation strategy.

On-Demand Big Data Analysis in Digital Repositories: A Lightweight Approach

We describe a use and reuse driven digital repository integrated with lightweight data analysis capabilities provided by the Docker framework. Using building sensor data collected from the Virginia Tech Goodwin Hall Living Laboratory, we perform evaluations using Amazon EC2 and Container Service with a Fedora 4 repository backed with storage in Amazon S3. The results confirm the viability and benefits of this approach.

SuperIDR: A Tool for Fish Identification and Information Retrieval

ABSTRACT Students, fisheries professionals, and the general public may value computer-facilitated assistance for fish identification and access to ecological and life history information. We developed SuperIDR, a software package supporting such applications, by utilizing the search and data retrieval capabilities of digital libraries, as well as key features of tablet PCs. We demonstrated SuperIDR utilizing a database with information on 207 freshwater fishes of Virginia. A user may annotate fish images and identify fishes by using a dichotomous key; searching for key words, similar images, subimages, or annotations on images; or combinations of these approaches. Students using the software demonstrated enhanced ability to correctly identify specimens. Their comments led to improvements, including the addition of new features. The PC-based system for identifying freshwater fishes of Virginia may be downloaded and modified. SuperIDR is a prototype for PC-based species identification applications—the syste...

Developing Institutional Research Data Repository: A Case Study

We introduce VTechData, a Sufia/Fedora based institutional repository specifically implemented to meet the needs of research data management at Virginia Tech. Despite the rapid maturity of Hydra and Fedora code bases, the gaps between the released packages and a launched production-level service are still many and far from trivial. In this practitioner paper we describe the strategy and efforts through which these gaps were filled and lessons learned in the process of creating our first Hydra/Sufia-based repository.