Field study of sixteen users to analyze daily user practices on document
classification and retrieval. Descriptions of problems found and proposal of
solutions.
Introduction
The organization of personal information resources can be divided in four
basic underlying tasks:
Information classification
Information categorization
Information filing
information retrieval.
This study aims at providing an up to date analysis of user practices.
Hypothesis
Users’ intuitive ideas of the organization of their personal information space do not match with the strict hierarchical data storage paradigm given by the file system.
Users consider it difficult to find a distinct piece of information within their own personal information space (in contradiction to Nardi et al. [1995]).
Users have clear ideas as to what is wrong with a system from their points of view, and of how to improve it. They are able to actually (verbally) express their ideas on these matters.
Research Questions
What are the habitual practices employed in the area of classification and retrieval, respectively?
Do the users employ the desktop (i.e. the plane of the screen) in some structured way?
If so, for what?
And, in what way?
What types of inconveniences and problems do users encounter during daily computer use?
What are the users’ strategies to cope with these problems?
From the users’ point of view: What is wrong with the user interface utilized today?
What are users’ suggestions for the system’s evolution?
Previous work
Method
Semi structured interview, videotaped, in field.
Interview guidelines:
Habitual sequence of activities at work.
Classification and maintenance
Use of desktop screen plane
Search habits and strategies
Results for Classification
Observed Classification Practices
Archiving: average age six months to eight years. Users consider archives
important
Maintenance: it was considered an important activity.
Use of hierarchical structures: new subfolders when 3 to 7 docs on the
same subject or when the user wants an overview. Classification and storage are
considered on-going processes.
Proper classification is perceived as difficult.
User interface issues:
Users used concepts and thematic proximity and document type to group the
content on the desktop.
Users used the screen for storage and kept actively used things, things to
archive and things to delete since they were not useful anymore. They also
used it to remember things to do.
Problems
User interface
Desktop metaphor was not understood: low level users didn't knew they
could use the desktop to put things on.
Installation programs asked the user to put icons on the screen and some
users didn't dare to touch them.
Irritation that file system and desktop looks differently but have the
same functionality.
Someone was bother by flying papers from one folder to the other.
Underlying system
Separation between files, emails and bookmarks.
No place for short-term notes and remarks
Link together resources that are the same or belong together (example
citation and cited articles, author and address book) and in general to have
more content and content based access to their information.
Difficult to classify and gain overview: they would like the system to
create overview and have the possibility to modify it.
Conclusions
Separate users and systems information.
Provide small but useful tools to manage info (such as annotation).
Integrate rather than separate information
Results for Retrieval
Observed Search Practices
Reluctance to use search tools
Manual search was most common. The common sequence was:
Direct access strategy
Semantic proximity
Exhaustive search
Users retrieved and ordered their data spatially even in the file system
if they had the opportunity to do so.
Searching the web vs. searching "My personal information space".
Why are web searches attractive?
internet search for unknown information
simple interface (Google), misspelled correction, search refine without
starting over
Problems
Today's built-in systems search tools are not easy to handle
Document metadata (such as name or path) did not help
No content search
Context is missing
Conclusions
Search tools are to complicated. Users are able to brush up on their general
knowledge of the general organization of their information space while searching
manually.
Discussion
Two different categories of problems for the users:
actual use of the system on their computers.
the system doesn't allow for the desired work or organizational functions
sought.
Skills are an important factor for working with the desktop systems.
Conclusion
Interviewer proposed solutions to today's problems:
Annotations as a new type of information
Store docs in publicly available formats
Content search
Simplify the UI of searching tools
Authors also proposed abstract metaphors, database underlying structure for
file systems