Yejun Wu
School of Library and Information Science,
Louisiana State University, USA
(wuyj@lsu.edu)
Background. Scientific documents often contain knowledge about what one entity did to another entity under what conditions (such as time, place, and method), which is related to another statement of what one entity did to another entity under what conditions. Such knowledge can be represented as relations between entities and events. Here what one entity did to another entity under what condition is defined as an event, which expresses the relationship between two entities under a condition.
Objective. The objective of this paper is to design a model of entity and event relationship that can be used to represent knowledge identified from scientific documents and to facilitate knowledge discovery and organization.
Method. The paper first presents a brief literature review on causal relationships, then evaluates four existing knowledge organization models and five event ontologies for their commonalities and differences in representing entity relationships and event relationships. The paper then proposes a combined entity and event relationship model based on the strengths of the existing event ontologies. Five main kinds of entity and event relationships are identified from an oil spill document set.
Results. The three domain event ontologies, CIDOC CRM, Event Ontology and NewsML-G2, are only useful in serving specific purposes. The two generic event ontologies, DOLCE+DnS and Event Model F, must be enriched to be useful for representing knowledge for discovery. An entity and event model is proposed based on the strengths of these event models for representing knowledge in scientific documents.
Download Article
Cite: Wu, Y. (2020). Modeling entity and event relations in scientific documents for supporting knowledge discovery and organization. LIBRES, 29(2), 77-90. https://doi.org/10.32655/LIBRES.2019.2.1