Fig. 1. The architecture of Storylines.

Fig. 2. The distribution of dimension contribution to the latent concept space. Dimension contribution is measured by S², where S is the singular value of the dimension.

Fig. 3. An overview of a semantic network of terms based on the top 100 most significant latent semantic dimensions in the VAST data set. The list of concept dimensions is filtered through a fisheye view for easy selection. Each term is represented by a red node. The node size reflects the global contribution of the term to the whole latent concept space.

Fig. 4. Identifying dimensions in which the term “mad cow disease” has a strong presence from a network of terms derived from the top 100 most representative latent concept dimensions. In this example, the term has the strongest presence in the 4th dimension.

Fig. 5. A semantic network of terms revealing the nature of the 4th dimension identified by LSI in the VAST data set.

Fig. 6. Story generation. A story is generated when a document is selected from the Timeline view. The red square of the document will turn in to a blue square. The generated story will appear in the Storyline window at the lower left corner of the interface. Each storyline is represented as a tree structure, containing subject line of document, time and named entities extracted from the corresponding documents. This feature allows analysts to study people who are involved in a particular thread of events as well as the locations and organizations related to these events.

Fig. 7. Story generation procedures.

Fig. 8. Social Network Analysis procedures.

Fig. 9. Named Entity Inter-relationship Explorer. After selecting a cluster in one type of named entity associative network, for instance people, related location entities, organization entities, subjects and time of events are highlighted accordingly. The color of nodes and labels is changed from red to blue.