Video Recordings

We focus on the problem of named entity disambiguation on a very detailed level. To each entity is assigned a concrete identifier of a corresponding Wikipedia article describing the entity. We will compare various context representations for proper disambiguation: bag of words representation, sentence structure representation and entity co-occurrence representation. We will show how we preprocess freely available dumps from Wikipedia in order to form indexes representing the context of named entities. Briefly we will cover also the topic of Czech DBpedia and the way it can be used to enrich textual content with structured data.