Word Tree

The Word Tree module is a visual tool to show the different contexts in which brands or products appear. Its graph-based display facilitates the rapid exploration of search results and conveys a better understanding of how language is being used surrounding a certain topic. The module processes the list of concordances of the sentence view and presents them in a more structured manner to emphasize the usage context of a term. It complements the tag cloud, which gives a good overview of the most frequent terms, but does not reflect their usage context within specific sentences.

Visual Representation. The Word Tree of the webLyzard dashboard [1] is based on the popular keyword-in-context technique [3] and adopts a symmetrical approach [2]. The root of the tree is the search term. The left part of the tree displays all sentence parts that occur before the search term (prefix tree), the right part those that follow the search term (suffix tree). These branches to the left and to the right help users to spot repetition in contextual phrases that precede or follow the search term. Visual cues include different font sizes to indicate the frequency of phrases, and connecting lines to highlight typical sentence structures.

Example. The tree-like structure is built by (1) searching for a term – e.g., science, (2) grouping identical phrases containing the term into nodes – e.g., political science, and (3) creating additional sub-nodes once the sentences start to differ – e.g., political science at …, political science professor …, etc. This grouping together of equal phrases into a connected tree structure sheds light on word usage within the selected source(s) in a given time interval.

Interactive Features

  • Hovering over a node highlights all connected sentences – only a single (complete) sentence in the case of leaf nodes, or all sentences containing the phrase from the root to the hovered branch in the case of intermediate nodes.
  • Single clicking on the root node (= a term matching the search query) displays alternative root terms, which can be used to create variations of the tree based on the same set of search results. This is useful when searching for multiple terms – e.g., clicking on a topic or using the logical “or” operator of the advanced search.
  • Single clicking on intermediate nodes reprocesses the shown data to create a new tree (the phrase from the previous root to the clicked word becomes the new root). This drill down operation can be used to limit the amount of information shown, and to explore sub-branches of the tree containing specific phrases.
  • Double clicking on a term triggers a new full-text search.
  • ‘Plus’ and ‘Minus’ buttons allow users to extend or trim the tree – i.e. add or remove hierarchical layers of branches. This hides most of the individual sentences and focuses the display on the primary tree structure.

References

  1. Scharl, A., Kamolov, R., Fischl, D., Rafelsberger, W. and Jones, A. (2014). Visualizing Contextual Information in Aggregated Web Content Repositories. 9th Latin American Web Congress (LA-WEB 2014). Ouro Preto, Brazil: Forthcoming.
  2. Fischl, D. and Scharl, A. (2014). “Metadata Enriched Visualization of Keywords in Context”, 6th ACM SIGCHI Symposium on Engineering Interactive Computing Systems. Rome, Italy: Association for Computing Machinery. 193-196.
  3. Wattenberg, M. and Viégas, F.B. (2008). “The Word Tree, an Interactive Visual Concordance”, IEEE Transactions on Visualization and Computer Graphics, 14(6): 1221-1228.