Conference thebay::joyoflex

203.0. "Word-Network Analysis (jargon from another field)" by SUPER::MATTHEWS (Don't panic) Tue Jun 10 1986 23:27

Received: from DECWRL by DEC-RHEA with SMTP; Wed,  4 Jun 86 20:29-PST
Received: from MC.LCS.MIT.EDU by decwrl.DEC.COM (4.22.05/4.7.34)
    	id AA27227; Wed, 4 Jun 86 20:29:00 pdt
Received: from CCC.MIT.EDU by MC.LCS.MIT.EDU via Chaosnet; 4 JUN 86  23:15:15 EDT
Received: from MIT-MC.ARPA by MIT-CCC via Chaosnet;  4 Jun 86 23:02-EST
Received: from BLUE.RUTGERS.EDU by MC.LCS.MIT.EDU  4 Jun 86 23:01:53 EDT
Date: 4 Jun 1986 23:02:45-EDT
From: U45571%UICVM.BITNET@MIT-CCC
Resent-Message-Id: <[email protected]>
Message-Id: <[email protected]>
Orig-To:   Computers at MIT-MC
Resent-From: [email protected]
Resent-To: info-cobol@mc
Resent-Date: Wed 4 Jun 1986 22:44-EDT
Apparently-To: <callas%[email protected]>

[from the CRTNET Newsletter, issue 42 - if anyone has the actual paper, please
 drop me a line --Dave]

                       AUTOMATED WORD-NETWORK ANALYSIS:
       An Illustration with Electronic Mail Over a Monthly Time Series

                              James A. Danowski
                      University of Illinois at Chicago

Paper presented at annual meetings of the International Communication
Association, Chicago, 1986.

The author is grateful to John R. Andrews, Computer Center, University of
Illinois at Chicago, and Paul Edison-Swift, North Central Computer Institute
for their contributions to this research.

                                   ABSTRACT

     This research tested hypotheses about linkages between the interpersonal
networks and the message content networks occurring in an organization over a
one-year period.  Automated network analysis procedures were developed to
content analyze raw natural language electronic mail texts.  Word were treated
as network nodes and their cooccurrence in texts as definers of links among
words.

     The data were obtained from a state-wide extension organization whose
electronic mail traffic and message content was captured for one year.

     Support was found for the hypotheses that:  as node integration in the
interpersonal network increased, node integration in the word network
increased;  and, crisis destablizes content networks.  Crisis nearly
quadrupled the amplitude of oscillation over the subsequent six months in size
of the largest word group.

It was also observed that:

  o at the node-centric level interpersonal integration was
    2.5 times more elastic than content network integration;

  o a crisis decreased the number of word groups by 50% the
    following month, but the second month afterward the number
    of groups reached a yearly high.  Subsequently, however,
    number of groups dropped to crisis-level lows for several
    months;

  o as content node integration decreased, the number of groups
    declined, as one of the groups became very large;

  o size of the largest word group was positively correlated with
    the number of group members, indicating that volume of shared
    content increases;

  o there was an apparent one-month lagged relationship between
    the number of messages (and users) and the number of content
    groups, with changes in number of messages occurring prior
    to content group changes.  This suggests that the infusion
    of diverse information in a month, results in differentiation
    of word groups in the subsequent month.

     The methods appear useful for mapping the social concepts or culture of a
community.

    

    I'm underwhelmed.
       
    -bs