« 出口(2004) | Main | 次読む論文 »

December 16, 2004

Mihavics(1996)

Kenneth W. Mihavics and Aris M. Ouksel, "Learning to Align
Organizational Design and Data", Computational and Mathematical
Organization Theory, Vol.1, No.2, pp.143-155, 1996.
Carleyのモデルをちょっと変えてみたよ，の論文．
タスクのビットにウェイティングをつけてみたり．
contigency mapを作っているのはいただけそうではある．
基本的には私のオリジナリティのための位置づけように使うと．
Carleyモデルを概説してくれている．批判も楽になります．

The major assumptions of Carley's model include: 1)
organizational behavior is history based (individual decision makers
learn from their own past experience to guide future decision making);
2) organizational decisions are derived from the decisions of the
boundedly rational individuals that make up the organization; 3) these
decisions do not require a consensus to be reached; 4) many problems
that need to be solved within organizations can be classified as
"quasi-repetitive integrated" decision making tasks (Carley
1992).(pp.145)

メインリザルト．

While the work is still ongoing, we have discovered several
interesting phenomena. 1) Our work indicates that Carley's model of
organizational learning is a good one in that it is extensible and
produces results that have intuitive appeal. 2) The general
relationships between organizational design and learning that Carley
discovered have been corroborated, in the case of uniform weights of
evidence. 3) However, in the case of non-uniform weights, the
distribution of weights of evidence affects the performance of Majority
Teams but not Hierarchies.
(snip)
4) Both Hierarchies and Expert Teams relatively stable final
performance levels, regardless of weight distributions. 5) Expert Teams
are significantly slower in learning than either Hierarchies or
Majority Teams. These findings are summarized in Figure 6, which
represents a contigency model for choosing organizational structures
given the assumptions and parameters identified in our study.
(pp.152)

今後の展開．
はい，このような内容は私達のモデルでは実装済みでございます．

Our future research directions include an analysis of
relevant cost factors that affect organizaitonal learning, the further
relaxation of underlying assumptions of the model (such as data
independence), and a continued push to extend the model so as to
capture additional aspects of organizational learning, especially in
terms of the effects of "shared mental models" (Kim 1993; Senge
1994).(pp.153)

Posted by ysk5 at December 16, 2004 10:31 PM