Information retrieval in a knowledge rich domain poses challenges that are different from other domains. The domain of fine arts and cultural heritage is an exemplar of such a domain. The many facets of, and complex interrelations between, works of fine art are not easily addressed by conventional keyword-based approaches or even by structured cataloguing systems. Information retrieval challenges in this domain include: the conversion of existing legacy data into knowledge representations that emulate the semantics of the domain's relationships; and easy access to a robust knowledge representation for users unfamiliar with query languages. Our research addresses aspects of both challenges as they are connected and may benefit from being addressed in conjunction. Based on a study on user preferences in art image search and a review of existing structured resources for cataloguing art and heritage information, we have developed two prototypes:
Ontology Populator
and
Artfinder
. The first prototype,
Ontology Populator
, is used to automatically enrich data akin to legacy data kept by heritage institutions and transform it into a knowledge base. The second prototype is a graphical query builder,
Artfinder
, which interacts with the knowledge base. The
Artfinder
interface, is constructed dynamically from the structure of the underlying knowledge. A task-based evaluation of
Artfinder
was carried out with 10 expert and 10 layperson evaluators. Participants reviewed the interface favourably and the evaluation also revealed potential for improvement.
Artfinder
and its “query logic,” perhaps is a semantically richer mode of accessing knowledge repositories, allowing for logically more complex queries than are currently supported outside the realm of dedicated query languages. We believe that domain experts and perhaps informed laypersons will benefit from this retrieval approach.
Abstract. Tasks and difficulties inherent in the largely open problem of temporal information extraction from legal text are outlined. We demonstrate the efficacy of tools and concepts available "off-the-shelf" and suggest refinements for such applications. In particular, the frequent references between regulatory texts have to be addressed as a separate named entity recognition task that bears relevance to an analysis of the temporal ordering of legislation. A regular expression-based approach as a robust first step towards addressing this problem is tested.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.