Transcript Poster - University of Maryland

BasketLens:
Searching for baskets of words in text collections
Darya Filippova, Catherine Plaisant, Ben Shneiderman
Searching for words
Basket library
Overview
BasketLens is an interface that allows users browse a
document collection and search for preset baskets of
words, revealing the distribution of those words in the
collection.
Users can create their own baskets,
or use predefined baskets, e.g.
baskets imported from Inquirer
(www.wjh.harvard.edu/~inquirer/).
The list of words are saved in a
simple text format so they can be
easily edited and shared with other
users.
Users can type lists of words and see which document
contain those terms. The table view has one column per
term. Each column contains the number of matches of
this particular word in each document. Cells are color
coded: the more matches the document had, the brighter
the color of the cell.
Revealing individual words of the baskets
Searching for baskets of words
The document view
show multiple
documents at once,
sorted by the number
or matches.
When searching for baskets of words, the table view
has one column for each basket. Here comparing the
use of terms for pain and pleasure (which in this
example rarely overlap). Users can chose a different
color for each basket.
www.cs.umd.edu/hcil/textvis/basketlens
Users can ungroup the
baskets to reveal where
individual words are used.
Options allow users to see
more compact views when
the number of words
becomes large.
University of Maryland, Human-Computer Interaction Lab www.cs.umd.edu/hcil