Download 8.7 Authentication Logs and more Lecture notes Architecture in PDF only on Docsity!
8.7 Authentication Logs
VIUS REPORTS 8.
Authentication Logs
From August, 2002 to the present, traffic on 3 images databases licensed by the
University Libraries was monitored with authentication logging. A web page was
interposed and users who had selected one of the databases were asked to provide a
user ID before connecting. These IDs were then searched against a library patron file to
identify status (faculty, student, staff, etc.) and departmental or campus affiliation. The
IDs were then encrypted to protect confidentiality and so that future authentications by
the same ID could be identified as repeat uses. (VIUS project staff IDs were removed
from the data.) The 3 databases monitored were:
• The AMICO Library™ – a cooperative collection of more than 65,000 digital
images of works of art from more than 30 museums, mostly in North America.
• The AP Multimedia Archive™ – a database of approximately 700,000 historical
and current photos collected by the Associated Press.
• The VIUS/CONTENTdm™ Database – a database of approximately 12,
images. This database came online August 29, 2002.
Authentication logging cannot provide enough context to answer the general question
“Are these databases frequently used?” However, we can obtain some useful
perspective by comparing traffic on these databases to the roughly 400 others offered
by the University Libraries (primarily indexes and full-text resources). Both the AP
Multimedia Archive™ and the AMICO Library™ were available at Penn State
throughout the calendar years 2002 and 2003 and statistics were available for that
period through September 2003. When the number of connections for those years for all
University Libraries databases we re ranked, these two pictorial databases stood well
above the median (near the top of the second quartile) but well below the average
(which was raised by half-a dozen databases with extremely high numbers). A
respectable showing that might be described as “moderate.”
Turning to the authentication logs, we examined the data collected from August 29,
2001, through September 16, 2003. During this time 2,814 people authenticated on the
picture databases 8,307 times (for an average of 2.95 authentications per user).
Looking at the subset of this time when all three databases were available (August 15,
2002 – September 16, 2003), 1865 people authenticated 5022 times for an average of
2.69 authentications per user. This lower average is expected due to the shorter
timeframe. We can then compare the authentication numbers for all three databases
over the same period of time: AMICO 403 users/ 1112 authentications for an average of
2.76 per user; CONTENTdm™ 930 users/ 2387 authentications for an average of 2.
per user; AP Multimedia Archive 657 users/ 1523 authentications for an average of 2.
per user.
8.7 Authentication Logs
Also of interest is looking at the authentications per user ID over time. The 2 year
averages are higher than the 1 year averages. However, they are not double but rather
just somewhat higher. For instance, the overall authentications per user ID average
went from 2.69 to 2.95 when doubling the amount of time. Additionally, by looking at the
changes in the numbers of unique IDs over that time and the raw authentication
numbers, it becomes clear that continued usage of the system over time was dominated
more by new users who use it a couple of times rather than retaining previous users
who will continue to use the system. In other words, there seems to a revolving door of
users who will use it two or three times and then be replaced by new users. This is true
of both databases but is more true of AP where increasing the time by 100% increased
the number of users by 111% versus AMICO where doubling the time increased users
by 73%. Since overall usage hasn't increased dramatically (as seen in the following
chart) it is possible to deduce that although many new users where added many must
have been lost -- hence the revolving door theory. Of course, this is based on just over
two years of data. It would be interesting to see this if this holds true over the longer run.
By looking at the total number of unique user IDs across all three databases compared
to the subtotals within each database, it can be determined that crossover usage is
fairly low. Out of 1865 users over the course of a year, an estimated 3.4% to 6.7% of
users used more than one image database within that time frame.
The following chart and table show significant fluctuations by month (apparently due to
assignments or seasonal variations in class offerings, etc). The two peaks associated
with CONTENTdm™ usage corresponds precisely with an extra credit assignment
initiated in cooperation with the VIUS project. The assignment occurred in a fall and
spring introductory Landscape Architecture class with enrollments of 700 students each
semester. The assignment intended to promote usage of the CONTENTdm™ database
in order to obtain informed opinions about the database for a survey taken in that class.
It was completed by approximately 1/
rd
of the students each semester. In addition, the
most extensive publicity for the VIUS database (a large format color postcard) was
distributed during the week of March 24, 2003. The fact that these stimulants show so
clearly on the chart illustrates how sensitive database use is to matters such as
assignments or publicity.
8.7 Authentication Logs
Considered by status, the authentication logs suggest that students constitute a larger
percentage of authentications on the CONTENTdm™ database and AP Multimedia
Archive™ than they do the AMICO Library™.
AMICO Authentications by Status (n=2113)
Faculty 25%
Graduate 9%
Other 1%
Resident 9% Scholar 3%
Staff 17%
Student 3%
UnivStud 1%
Undergrad 32%
CONTENTdm Authentications by Status (n=2400)
Faculty 8% Graduate 3%
Scholar 3%
Staff 13%
Undergrad 69%
Resident 4%
8.7 Authentication Logs
By comparing the demographics above to the generalized distribution of the Penn State
population, we can see that faculty usage of the AP Multimedia Archive and in particular
of the AMICO Library™, is higher than their mere size in the population would suggest.
AP Multimedia Archive Authentications by Status (n=3766)
Graduate 7%
Other 1% Resident 5%
Scholar 4%
Staff 12% Student 3%
Faculty 18%
UnivStud 2%
Undergrad 48%
8.7 Authentication Logs
AMICO Library™ Use with All Users by College at UP Campus
8/29/01-9/16/03 n=
AGRICULTURAL SCIENCES
ARTS AND ARCHITECTURE
BUSINESS
ADMINISTRATION
GRADUATE NON -DEGREE
HEALTH AND HUMAN
DEVELOPMENT
INFORMATION SCIENCES
AND TECHNOLOGY
SCIENCE
LIBERAL ARTS
UNDERGRADUATE NON -
DEGREE
EDUCATION
ENGINEERING
EARTH AND MINERAL
SCIENCES
DIVISION OF
UNDERGRADUATE
STUDIES
COMMUNICATIONS
8.7 Authentication Logs
CONTENTdm™ Use with All Users by College at UP Campus
8/29/01-9/16/03 n=
ARTS AND ARCHITECTURE
BUSINESS
ADMINISTRATION
COMMUNICATIONS
DIVISION OF
UNDERGRADUATE
STUDIES
EARTH AND MINERAL
SCIENCES
EDUCATION
ENGINEERING
GRADUATE NON -DEGREE
HEALTH AND HUMAN
DEVELOPMENT
INFORMATION SCIENCES
AND TECHNOLOGY
SCIENCE
LIBERAL ARTS
UNDERGRADUATE NON -
DEGREE
AGRICULTURAL SCIENCES
