Coupling of semantic and syntactic graphs generated via tweets to detect local events

Abstract

Local events are important for the local people, tourists and local authorities. However, information about them is limited, or even absent on official websites. We propose an approach to automatically generate a data graph on local events via social network and mobile positioning data. This graph is constructed by the research results of syntactic and semantic data collection and will be used for an intelligent local event recommendation application. The coupling of these two research results makes it possible to discover little-known events or places that are not in the Linked Open Data and at the same time to connect them to it. An online experiment was set up to evaluate the performance of the techniques used in the study on semantic data which concerns the automatic ontology generation. The results allow us to conclude that the performance of the Named Entity Recognition and the choice of threshold score in automatic ontology generation depend on the categories of detected words types.