X hits on this document

72 views

0 shares

0 downloads

0 comments

11 / 28

Brunswikian Architecture

Robert Gifford Victoria BC Canada E-mail: rgifford@uvic.ca

Reviewing all the newsletters online, I see that our group has only contributed to it 3 times, the last time in 1999, but we have never stopped conducting lens model studies over the last 20 years.

In terms of a lens model paradigm, we appear to be fixed at the place where Brunswik (1945) left off (see the 1956 book, pages 26-29, Experiment D), but this simple paradigm has been employed in contexts ranging from personnel hiring to judged intelligence and personality, burglars' and police assessments of potential burglary targets, and human physical attractiveness.

Our last three published Brunswikian articles have concentrated on how different groups of key players judge building aesthetics. Studies in Environment and Behavior (2001) and Journal of Architectural and Planning Research (2002) examine how architects and laypersons use different objective building cues (fenestration, cladding, height, shape, etc.) to arrive at different aesthetic conclusions. It has long been known that architects judge buildings differently from the rest of us, but these studies begin to elucidate exactly how and why that happens, beginning with objective cues, and including the observer's intermediate-level assessments such as the building's judged complexity, coherence, friendliness, etc. A similar

study

(Educational

Research,

2001)

compared

student and professor judgements of university classrooms.

The architecture studies seemed to catch the media eye, and were summarized in the national Canadian university newspaper University Affairs, the UK popular science journal NewScientist, and a few other places.

Are all judgments equal?

Claudia Gonzalez Vallejo Ohio University E-mail: gonzalez@ohiou.edu

My Brunswikian research continues to look at the extent to which global judgments differ in

systematic piece by

ways from judgments performed in a

piece

fashion

(disaggregated-where

attribute levels are judged for their quality).

In earlier

work we preference

found orders

that the relationship of objects evaluated via

between the global

method and choice preference orders than that of choice and disaggregated

was smaller orders. We

are

further

pursuing

this.

In

addition,

we

are

looking

at whether irrelevant cues influence global judgments to a greater extent than do disaggregated ones in a job selection task.

Using Judgment Analysis to Predict the Targets of Crime

Louise Gunderson University of Virginia E-mail: lfg4a@virginia.edu

I am using Judgment Analysis to create a model of target selection by criminals. Generally , (robbery, when a criminal commits a crime

burglary) the details of the crime are recorded in

police collect

reports.

This allows the investigator

information

about

the

crime

such

to as

location, time, date, and type the features of the target can

of offence. Many of be determined from

this

information, the

demographics of

the location

can

be found from

the US Census

block group

information, the weather can be found date, and so on. I am using the lens

from the model to

construct data mining methodologies the preferences of the criminals.

to

discover

One of the preliminary results was a methodology to predict the type of object stolen

from homes, from stolen together in

a set the

of objects period of

that were not

study.

First,

association were not

rules were used to stolen together.

find the objects that For the city of

Richmond, Virginia in 1996, bicycles, goods, firearms, livestock, and tools

consumable were never

stolen these

together. objects

All was

of the crimes in which one stolen were selected.

of A

classification tree was built using the 1996-1997, and the features derived

data from from the

police

reports.

This

classification

tree

was

used

to

predict

the

stolen

objects

in

the

crimes

occurring

in

1998. This method correctly predicted object 44.2 % of the time, as compared the time by random draw.

the stolen to 28% of

Further work is continuing on the discovery of target preferences by a clustering methodology based on the lens model. This methodology has been demonstrated to outperform other clustering methods on synthetic data sets. At this point, it is being modified to take into account the distribution of features in the natural environment.

This work has been presented at various engineering conferences, such as PerMIS 2002 and IEEE SMC, 2002. The work presented at last

year's Brunswik Society meeting published as a chapter in Advances in

has been Computers,

Vol 56, M. V. Zelkowitz. Ed.

Newsletter 2002 page 11 of 28

Document info
Document views72
Page views72
Page last viewedFri Dec 09 08:43:49 UTC 2016
Pages28
Paragraphs1455
Words18198

Comments