Exploratory analysis of textual data streams is addressed. Documents are classified into fine-grained similarity clusters. Similar clusters are aggregated into larger topics. Newly extracted topics are linked with similar topics of different time windows. Thematic analysis of topics and topic trends in time is supported.