Tip:
Highlight text to annotate it
X
Automatic evaluation of user interfaces, UI, allows studies with more users without increasing
costs.
So, testing with more users tends to be costly and the automatic evaluation makes this possible,
to involve more users without making this kind of study costly.
Then, remote evaluation allows us to get real data from users in their regular environments.
So, we try to (?) the first point I showed.
Then, UI evaluations in real situations point to the use of event logs.
So, we can now how the access is occurring when they are using the UI in their regular
So, due to the typical amount of data of interface events when users are using an UI we are guided
to use automatic evaluation tools.
Now I'll talk about event-based evaluation tools. So, what kind of events we can have?
Here's a first example of server logs.
So, when you access a page, here's a simple example of Institute of Computing homepage,
you'd got... you no, the server would have something like that.
So, we get the remote host, the time that it occurred, the page that was requested,
And at each new access, it is logged on the server.
So, analyzing this log you can have all the path the users performed.
Data available at client logs. What kind of information we have in client logs?
Another example: We have a pointer here over here the Institute of Computing logo and here
what kind information we can have?
Like 400 ms, mouseover event, we have some coordinates, the target ID of the element,
and the location.
So, if the user moves the mouse over the x-axis, we got more information. And then he clicks
one time, and then another time.
So, this is just to show how kind of information we have in server log and client log.
Here we got a more detailed information, because we now that user moved the pointer and clicked
two times.
So, after these topics, our objective in this work is identify solutions, limitations, and
gaps in this kind of tools -- event-based evaluation tools of websites.
We tried to find out what characteristics strengthen this kind of tools, make them more
robust, less costly, and easier to use.
Easier to use, from the side of the tool administrator, website administrator, to the user.
So, how users use this tool? How administrators use this tool?
These have to be easy for both.
And the solutions and limitations found. We're trying to help the developers so that they
can reuse consolidated ideas and avoid the found limitations.
Some tools we analyzed.
We defined some aspects and they focusing in comparing, comparing is not the right word,
but analyzing different tools that are event-based.
First we analyzed a tool called WET, WebRemUSINE, and WebQuilt.
NAUTICUS is here because it represents one of aspects that we didn't find in any event-based
tool that addresses this requirement.
So, we present this only to illustrate the aspect.
These are the 8 aspects.
First, configuration of the environment, that means what the administrator of the tool or
what the site administrator must do to put a tool to work.
So, I have to host a file, edit my Web pages and some things like that.
Two, if the tool uses just server-side log or client-side logs, and things like that.
The third aspect was how are the solutions to store or transmit logs to the server.
So, how are the solutions?
Four, what are the dependencies on the users' actions?
What the users have to do to this tool work.
The user must select something, click on something to the tool capture, and something g like
that.
Five, what are the impacts of using this tool will cause to the user.
So, this tool will be showed in a popup or something like that.
What this tool impacts in the interface?
Six, scalability, from just the number of tests and how longer these tests sessions
can be.
And seventh is the high level information.
How the tool obtain high level information from logged data?
And eighth, finally, is about the integration of accessibility and usability.
These are the aspects we identify in the 3 tools tool we analyzed that use event logs.
Here I will present just a few examples of these the eight aspects.
The combined work is under... they don't...
Just talk about some points just to present how the analysis was made.
First, an example of the first aspect.
A solution depends only on the native script language.
It's a characteristic the Web tool because depends only...
The tool has some specific characteristics to capture data and just depends on the script
language, what we defined as a good solution.
A limitation we found: requires changes in the source of Web pages.
We find a tool that doesn't need this kind of changes and in websites this can be hard.
Since it is a difficult requirement to achieve, in some point of the pages, this can be hard
in huge websites.
Would be hard to use a tool like this, for example.
An interesting point, capture events without changing evaluated Web pages or using specific
browsers.
Because we find a tool that... exists a tool that make this kind of capture though the
use of specific browser.
So the user has to install a specific browser so it changes all the environment, so, it's
artificial again.
Examples of the third aspect: store and transmit logs.
An interesting solution is to use a proxy logger to record the data between the user
and the server-side he accesses.
So, there's a...
The tool mediates this communication.
So, it's an interesting solution to store and transmit logs, since it doesn't depend
on the resources of the client.
A limitation: depend on the client device, for example, for storing logs.
What we define as interesting is to use cyclic transmission to avoid some of the limitations
or constraint.
Some examples of the fourth aspect: an interesting solution is to use the access of the user
first time and no further access is required.
This is an interesting solution, because in some tools the user has to select some option
so the capture can start.
So, we find this as an interesting solution.
The user starts using and then after it brings the real use to the point.
And limitations: depend on actions to perform primary functions like log data or transfer
data.
An interesting, in fact, a subject we defined as interesting is to make the acceptance of
the user to be valid for several sessions.
So, he accepts one time and until the user chooses to stop the capture he can still going.
So the tool can capture data more related to common environment and common tests.
And finally, the aspect we defined as really interesting is the combination of accessibility
and usability.
We don't find...