Temporal Knowledge Extraction for Dataset Discovery

Linked data datasets are usually created with different data and metadata quality. This makes the exploration of these datasets a quite difficult task for the users. In this paper, we focus on improving discoverability of datasets based on their temporal characteristics. For this purpose, we identify the typology of temporal knowledge that can be observed inside data. We reuse existing temporal information extraction techniques available, and employ them to create temporal search indices. We present a particular use-case of dataset discovery based on more detailed and completed temporal descriptions for each dataset in the Czech LOD cloud based on the analyzing of the unstructured content in the literals as well as the structured properties, taking into consideration varying data and metadata quality.