Information Technology Reference
In-Depth Information
from one another's' messages. Thus, for example, A and B are connected in
the social network and the arc between A and B is labeled with a theme - e.g.,
“sports” - if and only if A and B have had at least one interchange like the fol-
lowing: A posts a message about baseball, B replies with a post about football, B
posts a message about swimming, and A cites or responds to B's message with
one about skiing. Since baseball, football, swimming, and skiing are all sports,
the link between A and B might be labeled with the more abstract term “sports”
(computed by the Conversation Map system using the WordNet thesaurus, ver-
sion 1.6). So, the themes listed in the menus are only there if there has been one
or more reciprocated responses in which the theme (or a semantically similar)
term was mentioned in each of the exchanged messages.
Figure 6, showing the reciprocated discussion themes in the messages of
archive 2, is a surprisingly short list. Usually the menu of themes lists many
items. Clicking on the items to highlight the parts of the social network that
they label shows even more clearly how fragmented the discussion of archive
2 is. All of the themes listed connect only one pair of posters. In short, only
a small handful of the interchanges concerning the “Hell Money” episode are
focused around a specific theme of discussion.
Figure 5, showing the reciprocated discussion themes in the messages of
archive 2, again shows that the social interchange visible in the message archives
is more cohesive in the first archive than it is in the second archive. This can be
interpreted from the longer list of reciprocated themes for archive 1.
Semantic networks
The semantic networks shown in Figures 7 and 8 show that the conversations
after both episodes are concerned with the main characters (Scully and Mul-
der). Moreover, it is interesting to see the computed similarities between the
main characters and the more generic terms of “you,” “me,” “someone,” “any-
one” etc. These calculations provide a way of seeing how the audience members
talk about themselves in ways comparable to the way they talk about the main
characters. This calculation might be compared to analyses of character “iden-
tification” discussed in the literatures of film theory and other media studies.
Conclusions
A computational sociolinguistic analysis of stories has been proposed and im-
plemented in the
Conversation Map
system. The significance of a story is seen as
