Abstract. Sensemaking is the process of analysing complex situations in order to make informed decisions. Semantic Web technology can be effectively used to create new sensemaking systems that focus on concepts and knowledge instead of documents. We demonstrate how this is achieved using information extraction to acquire knowledge and create a semantic repository that can then be semantically searched. A domain ontology is used to support the creation of an analysis tree; the semantic visualisation enables knowledge discovery, a core aspect of sensemaking.Keywords: Semantic web, Knowledge acquisition, Semantic visualisation, Sensemaking.
BackgroundSensemaking is the process of analysing complex and uncertain situations in order to take informed decisions [1]. It requires knowledge workers to find relevant information and formulate hypotheses based on the knowledge contained, test the hypotheses and revise the effectiveness of the decisions adopted. While some systems support only document manipulation in sensemaking (e.g., [2]), the X-Media project [3] uses Semantic Web (SW) technology and ontology-based visualisation to enable abstraction from the low level of multiple instances and more easily compose a conceptual model of a situation. Knowledge acquisition and retrieval, and visualisation are basic components of semantic sensemaking.Knowledge acquisition (KA) transforms information into knowledge via classification and extraction. While information is heterogeneous and dispersed, coming from many sources, e.g., text documents, images, tables, etc., knowledge is homogeneous and instantiates concepts and relations captured in domain ontologies. In knowledge retrieval, users search knowledge repositories to find those missing elements that would improve their understanding of the current situation. Knowledge visualisation supports visual thinking by explicitly laying out the retrieved knowledge, information and personal reflections.We demonstrate the steps we follow during analysis using the XMediaBox, developed as part of X-Media, using a non-restricted dataset. The XMediaBox has also been evaluated with experts in a realistic, knowledge-intensive setting: aerospace engineers working on gas turbine engines [4].