US Election 2016 Web Monitor Thumbnail

Relation Tracker – Opinion Leaders and Semantic Associations

To identify opinion leaders and reveal key factors influencing online discussion about a topic, the webLyzard platform accurately detects known entities – i.e. persons, organizations, and locations that are influencing news and social media coverage. To develop a deeper understanding of this process, the relation tracker helps analysts to not only understand how these entities impact topics of interest, but also reveals the interconnections among the entities themselves.

How did a public appearance of the CEO impact a company’s perceived relation to its main competitor, for example, and what are journalists associating with the competitor’s latest announcement?

The interactive Relation Tracker helps to answer such questions. It grants analysts an unprecedented look into the complex and influential relations around entities and topics of interest.


Visualization Components

  • Data Columns. The selected data entries (up to ten) are listed in two vertical columns. The font size indicates the number of documents that mention the data entry.
  • Relations. Connecting curves indicate a relation between a pair of data entries. The thickness of a curve represents how many times these entries co-occur.

Graph Configuration

  • Level of Detail. A slider element beneath the title of the central column determines the level of detail – i.e., it controls the threshold for showing relations among data entries.
  • Column Menus. Each column provides a settings menu via its title. The available options apply to the given column only:
    • Data Types. The selection includes entities (people, organizations, locations), predefined topics, sources, and other metadata attributes such as sentiment and language. Selecting the same in both columns visualizes relations within the same data type; e.g., entity-to-entity relations.
    • Sorting. Each column can be sorted by frequency, average sentiment, name, or number of relations.
    • Color Coding. There are three options: (1) no color coding; (2) by sentiment – from red to green depending on average sentiment, in line with other dashboard elements; and (3) by topic – using the color legend from the topic sidebar.

Interactive Features

  • Data Selection. Mouse-over checkboxes allow selecting individual data entries for specific comparisons. The selected entries stay highlighted while the opacity of the remaining elements is reduced. The selection can be cleared using the “Clear Selection” action from the respective column menu.
  • Highlighting and Tooltips. When hovering over a data entry, its outgoing relations and the connected data entries stay highlighted while the opacity of the remaining graph elements is reduced. After hovering over a relation, the curve and connected data entries remain highlighted. Additional context and search options are provided via adaptive tooltips.

Last major update with release 2016-12 (Leopard Gecko).