The REDCap consortium: Building an international community of software platform partners.

Despite progress in perceptual tasks such as image classification, computers still perform poorly on cognitive tasks such as image description and question answering. Cognition is core to tasks that involve not just recognizing, but reasoning about our visual world. However, models used to tackle the rich content in images for cognitive tasks are still being trained using the same datasets designed for perceptual tasks. To achieve success at cognitive tasks, models need to understand the interactions and relationships between objects in an image. When asked "What vehicle is the person riding?", computers will need to identify the objects in an image as well as the relationships riding(man, carriage) and pulling(horse, carriage) to answer correctly that "the person is riding a horse-drawn carriage." In this paper, we present the Visual Genome dataset to enable the modeling of such relationships. We collect dense annotations of objects, attributes, and relationships within each image to learn these models. Specifically, our dataset contains over 108K images where each image has an average of $$35$$35 objects, $$26$$26 attributes, and $$21$$21 pairwise relationships between objects. We canonicalize the objects, attributes, relationships, and noun phrases in region descriptions and questions answer pairs to WordNet synsets. Together, these annotations represent the densest and largest dataset of image descriptions, objects, attributes, relationships, and question answer pairs.

/pdf/visual-genome-connecting-language-and-vision-using-15ih2ylk6r.pdf

Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations

We present YAGO, a light-weight and extensible ontology with high coverage and quality. YAGO builds on entities and relations and currently contains more than 1 million entities and 5 million facts. This includes the Is-A hierarchy as well as non-taxonomic relations between entities (such as HASONEPRIZE). The facts have been automatically extracted from Wikipedia and unified with WordNet, using a carefully designed combination of rule-based and heuristic methods described in this paper. The resulting knowledge base is a major step beyond WordNet: in quality by adding knowledge about individuals like persons, organizations, products, etc. with their semantic relationships - and in quantity by increasing the number of facts by more than an order of magnitude. Our empirical evaluation of fact correctness shows an accuracy of about 95%. YAGO is based on a logically clean model, which is decidable, extensible, and compatible with RDFS. Finally, we show how YAGO can be further extended by state-of-the-art information extraction techniques.

/pdf/yago-a-core-of-semantic-knowledge-13k3nmj6l2.pdf

Yago: a core of semantic knowledge

This paper gives an overview of the development of the field of Knowledge Engineering over the last 15 years. We discuss the paradigm shift from a transfer view to a modeling view and describe two approaches which considerably shaped research in Knowledge Engineering: Role-limiting Methods and Generic Tasks. To illustrate various concepts and methods which evolved in recent years we describe three modeling frameworks: CommonKADS, MIKE and PROTEGE-II. This description is supplemented by discussing some important methodological developments in more detail: specification languages for knowledge-based systems, problem-solving methods and ontologies. We conclude by outlining the relationship of Knowledge Engineering to Software Engineering, Information Integration and Knowledge Management.

/pdf/knowledge-engineering-principles-and-methods-6do1hbj75r.pdf

Knowledge engineering: principles and methods

The National Human Genome Research Institute (NHGRI) Catalog of Published Genome-Wide Association Studies (GWAS) Catalog provides a publicly available manually curated collection of published GWAS assaying at least 100000 singlenucleotide polymorphisms (SNPs) and all SNP-trait associations with P <110 5 . The Catalog includes 1751 curated publications of 11912 SNPs. In addition to the SNP-trait association data, the Catalog also publishes a quarterly diagram of all SNP-trait associations mapped to the SNPs’ chromosomal locations. The Catalog can be accessed via a tabular web interface, via a dynamic visualization on the human karyotype, as a downloadable tab-delimited file and as an OWL knowledge base. This article presents a number of recent improvements to the Catalog, including novel ways for users to interact with the Catalog and changes to the curation infrastructure.

/pdf/the-nhgri-gwas-catalog-a-curated-resource-of-snp-trait-3zt2r8ota7.pdf

The NHGRI GWAS Catalog, a curated resource of SNP-trait associations

There are many automatic approaches for generating matches between ontologies and schemas. However, these techniques are far from perfect and when the use case requires an accurate matching, humans must be involved in the process. Yet, involving the users in creating matchings presents its own problems. Users have trouble understanding the relationships between large ontologies and schemas and their concepts, remembering what they have looked at and executed, understanding output from the automatic algorithm, remembering why they performed an operation, reversing their decisions, and gathering evidence to support their decisions. Recently, researchers have been investigating these issues and developing tools to help users overcome these difficulties. In this chapter, we present some of the latest work related to human-guided ontology matching. Specifically, we discuss the cognitive difficulties users face with creating ontology matchings, the latest visual tools for assisting users with matching tasks, Web 2.0 approaches, common themes, challenges, and the next steps.

/pdf/interactive-techniques-to-support-ontology-matching-4qu7dpl7gv.pdf

Interactive Techniques to Support Ontology Matching

As knowledge engineering moves to the Semantic Web, ontologies become dynamic products of collaborative development rather than artifacts produced in a closed environment of a single research group. However, the projects differ--sometimes significantly--in the way the community members can contribute, the different roles they play, the mechanisms they use to carry out discussions and to achieve consensus. We are currently developing a flexible mechanism to support a wide range of collaborative workflows in the Protege environment. In this paper, we analyze workflows for several active projects, and describe the properties of these workflows. We discuss an ontology that we developed to represent different aspects of workflows for collaborative ontology development. This ontology is a key component of the customizable workflow support in Protege. We evaluate the coverage and flexibility of this ontology by using it to represent formally two different collaborative workflows described in the literature, Diligent and BiomedGT. This evaluation demonstrates that our workflow ontology is sufficiently flexible to represent these very different workflows.

A Generic Ontology for Collaborative Ontology-Development Workflows

The great success of Web 2.0 is mainly fuelled by an infrastructure that allows web users to create, share, tag, and connect content and knowledge easily. The tools for developing structured knowledge in this manner have started to appear as well. However, there are few, if any, user studies that are aimed at understanding what users expect from such tools, what works and what doesn–t. We organized the Collaborative Knowledge Construction (CKC) Challenge to assess the state of the art for the tools that support collaborative processes for creation of various forms of structured knowledge. The goal of the Challenge was to get users to try out different tools and to learn what users expect from such tools–features that users need, features that they like or dislike. The Challenge task was to construct structured knowledge for a portal that would provide information about research. The Challenge design contained several incentives for users to participate. Forty-nine users registered for the Challenge; thirty- three of them participated actively by using the tools. We collected extensive feedback from the users where they discussed their thoughts on all the tools that they tried. In this paper, we present the results of the Challenge, discuss the features that users expect from tools for collaborative knowledge constructions, the features on which Challenge participants disagreed, and the lessons that we learned.

Technical Report: The CKC Challenge: Exploring Tools for Collaborative Knowledge Construction

Using the wisdom of the crowds to find critical errors in biomedical ontologies: a study of SNOMED CT.

A new generation of tools supports the integration of Web 2.0 and semantic Web approaches. Some of these tools-such as Semantic MediaWiki (http://meta.wikimedia. org/wiki/SemanticMediaWiki), BOWiki, (http://onto.eva. mpg.de/bowiki), and Platypus Wiki (http://platypuswiki. sourceforge.net)-provide wiki extensions for creating semantic links between pages. Other tools let users organize tags in some semantic structure, and fully fledged ontology editors such as pOWL support the distributed and collaborative development of ontologies. Commercial tools such as Freebase (http://freebase.com) are also entering the field. Most of these tools are in early development-the collaborative-knowledge-construction field is in its infancy. Few, if any, user studies outline what users expect from such tools and what does or doesn't work. So, we organized the collaborative knowledge construction challenge. The CKC challenge let users try different tools and provide feedback to help us assess the state of the art.

/pdf/the-ckc-challenge-exploring-tools-for-collaborative-jivndbhbdu.pdf

Natalya F. Noy

Papers

Interactive Techniques to Support Ontology Matching

A Generic Ontology for Collaborative Ontology-Development Workflows

Technical Report: The CKC Challenge: Exploring Tools for Collaborative Knowledge Construction

Using the wisdom of the crowds to find critical errors in biomedical ontologies: a study of SNOMED CT.

The CKC Challenge: Exploring Tools for Collaborative Knowledge Construction