Touching hands
5.5Example of an (unobtrusive) Agent Application: The Remembrance Agent
Navigation:

Back to the previous page Previous page

To the Table of Contents of this chapter Chapter ToC

To the Table of Contents of this paper Paper ToC

Most of the current information retrieval applications concentrate on query-based searches. However, they do not help when someone does not remember enough to even know to ask a question, or what question to ask;

"With more powerful desktop computers, most of a computer's CPU time is spent waiting for the user to hit the next keystroke, read the next page, or load the next packet off the network. There is no reason it can't be using those otherwise wasted CPU cycles constructively by performing continuous searches for information that might be of use in its user's current situation. For example, while an engineer reads email about a project an agent might remind her of project schedules, status reports, and other resources related to the project in question. When she stops reading email and starts editing a file, it should automatically change its recommendations accordingly."

The Remembrance Agent ([RHOD96]) performs this continuous, associative form of recall by continuously displaying relevant information which might be relevant to an individual user in this person’s current context. However, it does not just display this information, as this would distract the person too much and would not be of much help to him or her at all! The philosophy behind the Remembrance Agent (RA) therefore is that is should never distract from the user’s primary task, but only augment it. It accomplishes this by suggesting information sources (deemed possibly relevant to the user’s current situation) in the form of one-line summaries at the bottom of the screen or current active window. Here, they can be easily monitored (at an eye’s glance) without distracting from the primary activity the user is engaged in. The full text of a suggestion can be brought up with a single keystroke. Effectively, the agent thus becomes an augmentation of the a person’s memory. It is important to note that - unlike most information retrieval systems - the RA runs continuously without user intervention.


How is this agent constructed?
The agent consists of two parts: a front end and a back end. The front end continuously watches what the user types and reads, and sends this information to the back end. The back end then finds old email messages, personal documents, and on-line documents which are somehow relevant to the user’s context. This information is then displayed by the front end in a way which does not distract from the user’s primary task; in case of, say, a text editor, this would lead to the information being displayed in one-line suggestions at the bottom of the editing window, along with a numeric rating indicating the relevance of the displayed document(s). With a simple key combination, the user can bring up the full text of a suggested document.
The power of this construction is that the information displayed is just enough for the user to get an idea of the full document being suggested, but not too verbose to take up too much screen estate and/or too obtrusive to distract the user from its primary activity. Also, the frequency with which the front end provides new suggestions, as well as the number of suggestions, is kept low enough as not to be distracting.
Currently, to find similar and relevant documents, the SMART information retrieval program is used as a back end. While SMART is not the most sophisticated system, it was chosen as it has the advantage that it requires no human pre-processing of the documents being indexed (and searched).

When looking for relevant and useful documents, the agent does not have very much information (e.g. context) it can use to perform its searches. As a result, many of the suggestions of the RA or false positives, and not all that useful. However, as there are almost no ‘costs’ to the user to see a suggestion and ignore it if deemed not useful at the time, this is not as big a problem as one might at first think. The fact that no colour cues or highlights are used when displaying suggestions, and the fact that suggestions are displayed at regular intervals, contribute to the low ‘costs’ of checking them (they do not distract too much).
Just as it is important that suggestions are unobtrusive, it is also important for the success of the RA that it is trivial to both access the entire suggested document and to return to the primary activity once it has been viewed.

While using the system, it became clear that suggestions are much more useful when the document being suggested only contains one ‘nugget’ of information, and when this nugget is clearly displayed on its one-line suggestion:

"This ‘less is more’ approach solves several problems. First, it allows the user to tell what a suggestion contains from its description without having to peruse the entire document. Second, a document with only one primary point is more likely to be a good hit. Documents which address several issues will rarely match the user's situation exactly, but will often partially match. Finally, if a suggested document is read, the shorter it is the quicker the user can get on with their primary work. Email and short notes files seem to be a very good length for RA suggestions."

Using personal email and documents turned out to be a good idea for other reasons as well. The most important reason being that these are pre-personalised to each individual user and that these files will automatically change as the user’s interest changes. It turned out that this compensated very well for other RA shortcomings (as mentioned earlier on).

In future applications of the RA, it is planned to provide the back-end with an algorithm which learns based on which suggestions the user asked to have displayed. In this way, the RA can display more documents that are actually useful, and cull out those which seem to be useless suggestions.

On to the next page Next page
To the Hermans' Home Page Home Page
Chapter 5 - Tomorrow's Internet: an Ubiquitous and Agent-serviced Online Market Place...? "Desperately Seeking: Helping Hands and Human Touch" -
by Björn Hermans