MaxiBook: User-Centered Hypertext on an ASCII ... - Semantic Scholar

5 downloads 193 Views 110KB Size Report
By reusing the search engine from the original SuperBook system and ... By paying close attention to positioning the cursor, we found that the ASCII terminal.
MaxiBook: User-Centered Hypertext on an ASCII Terminal Michael L. Littman Bellcore Cognitive Science Research

Abstract The SuperBookTM1 browser (Remde, Gomez, Landauer, Eberhardt, and Lochbaum, 1987) was developed to provide users with improved online access to structured textual materials. It employs hypertext-like features in a graphical windowing environment to allow users to carry out search and browsing tasks. The MaxiBook project was carried out in order to make the features of the SuperBook browser available to people who do not have access to graphical display terminals. The successful creation of a robust and usable program involved three phases: prototyping in a windowing environment, prototyping in an ASCII terminal environment, and nal development. The second and central phase is the topic of this paper.

1 Motivation The information retrieval needs of Bellcore's owners and ongoing research in Cognitive Science came together in a system called the SuperBookTM browser (Remde, et al, 1987). This system was created to allow casual users to move through electronic documents and nd information at least as easily as they had with corresponding paper documents. The system has shown itself to be highly successful in behavioral experiments. Studies have revealed that people were able to use it to nd information more quickly and accurately than they could using paper (Egan, Remde, Gomez, Landauer, Eberhardt, and Lochbaum 1989). The work has generated interest within the Bell operating companies but surprisingly few wanted to attempt to put it into operation. One of the reasons was that the majority of the potential users have access to only ASCII terminals. Equipping users with graphical workstations is expensive. In addition, an organization may worry about abandoning its current systems and moving to something less well known and tested. By creating an ASCII terminal version of the SuperBook browser, we hoped to satisfy two goals. We wanted to respond to the needs of our owners by providing the functionality of the SuperBook browser to as many people as possible as soon as possible. In addition, we wanted to make it possible for our users to accumulate experience with their ASCII 1

SuperBook is a trademark of Bellcore.

1

terminals that could eventually be transferred to the graphical workstation version of the browser.

2 Phases of Development The creation of MaxiBook had three main phases. The rst phase was the creation of the original SuperBook browser in a graphical environment. It was this version (written by Joel Remde in Franz Lisp and then in C) which was used to identify the functions of the interface most important for carrying out the intended tasks. Formal usability experiments were performed and the code was rewritten from scratch at least twice. The second phase involved recreating the fundamental features of the graphical version (bitmapped display and mouse) using only ASCII terminal functionality (24  80 ASCII characters and keyboard). This version (written by Michael Littman in GNU Emacs Lisp) is discussed in detail in the following section. The focus in this phase was on mapping as much of the original functionality as possible to an ASCII terminal. In addition, e ort was spent identifying and eliminating interface constraints speci c to ASCII terminals. The conversion from graphical to ASCII terminal was reasonably successful. The resulting interface retained a strong family resemblance to the original and seemed to be useful for the same class of tasks by the same types of people. Nonetheless, without the SuperBook system to serve as a model, many fatal aws could have been made at this stage. The behavioral studies of the original system helped indicate which design decisions were important and which were inconsequential. The nal stage of development consisted of rewriting the prototype from scratch in C and CURSES. This program was written by Wai Nam Tam and Debbie Patel of Bellcore's ST&S as a near-identical copy of the ASCII prototype. The focus here was on implementation and speed instead of exploration and exibility. As a result, this version maintains all of the functionality from the original ASCII terminal version but runs signi cantly faster. In addition, we believe it is more portable, as use of CURSES is more wide-spread than GNU Emacs. This nal phase of development was very important in building an ecient system. We were lucky to have the collaboration of extremely capable programmers.

3 ASCII Prototype Development As described above, the second phase of development was mapping the functionality of the graphical version to the capability of an ASCII terminal.

3.1 Functionality of the SuperBook system The SuperBook program is intended for users to browse through textual material such as technical manuals. Therefore, aside from graphical gures, it is not heavily dependent on bitmapped displays. This made the mapping to an ASCII terminal considerably easier for the SuperBook system than it would have been for, say, a CAD program. 2

On the input side, the SuperBook system makes only limited use of the mouse. It primarily supports two simple functions: selecting a subwindow and indicating a word or section name in the subwindow. Again, this made the job of conversion to an ASCII display relatively simple.

3.2 Mapping to an ASCII terminal Development of the ASCII prototype was under GNU Emacs editor/operating environment (Stallman, 1985). We found GNU Emacs to be a powerful interface prototyping tool for ASCII displays. Many important functions were provided as primitives and it was possible to test out changes to the interface while the system was running. In addition, the program is distributed by the Free Software Foundation free of charge, which simpli ed the legal issues in distributing the code to other researchers for testing. Simulating the multi-window display was relatively easy using GNU Emacs. We chose to represent the two most signi cant windows using a tiled, vertical split-screen approach (about 40 columns apiece). For ease of development, we moved the window titles to the bottom of each subwindow. We also retained a blank line at the bottom to act as a simple sort of pop-up dialogue box. It seemed clear that cursor keys or arrow keys could assume some of the functionality of the mouse (with the RETURN key acting as the single mouse button). However, instead of emulating the notion of clickable buttons, characters on the keyboard served the same function (s for search, n for next, etc.).

3.3 Further re nement: informal testing By reusing the search engine from the original SuperBook system and prototyping the interface with GNU Emacs Lisp, we were able to write a simple ASCII terminal front-end to the above speci cations rather quickly. Inviting naive but friendly users to try the system helped us quickly identify many significant usability problems and opportunities for improvement. This observation is consistent with the work of people who have been advocating the use of informal methods for usability engineering (Nielsen, 1989). The loudest message from the users was that the interface should hide details of the GNU Emacs environment. This surprised the designers but was relatively easy to do. Several vi commands were provided as alternatives for moving the cursor and a script was written to allow users to enter the system from outside GNU Emacs. Before this change, several users refused to even start the program on their terminals. With the clickable buttons absent from the interface, we found that users missed their ubiquity| by being visible, they remind users of the available commands. This de ciency was remedied by including a short list of keyboard commands at the bottom of the screen and also by adding an easily viewed help screen which lists the entire set of commands. A subtler point was that the position of the cursor was much more important in MaxiBook than it had been in the SuperBook system. Since cursor motion with the keyboard is more tedious than pointer motion with the mouse, users craved automatic and sensible 3

default positioning of the cursor. They seemed to need the cursor to have a context speci c \home" location in each of the subwindows. To make this possible, we made \jumping between windows" an atomic action (TAB). The \home" position, then, became either the current search word or the current section name, depending on which window the user was in. Furthermore, after a scroll, the cursor remained xed with respect to the text| not the window.

3.4 Bene ts of the ASCII system By paying close attention to positioning the cursor, we found that the ASCII terminal version actually displays some advantages over the graphical system. For one, since the cursor is automatically moved to positions of interest, users found that they could accomplish some basic functions with no cursor motion at all. This is impossible in the mouse-based system because the pointer moves only when the user moves it. We also noticed that people seem to use the ASCII terminal version more e ectively for reading long passages. This may be due to their use of the cursor as a place holder (much like speed readers use their ngers). Users often lose their place after scrolling in the graphical version.

4 Conclusion A three-phase model of development: graphical version, exploratory ASCII terminal version, and development ASCII terminal version, was successful in producing a hypertext system more ecient for searching and browsing than several other more popular systems (Rada and Murphy, 1990). Informal testing and user feedback at all levels of development contributed a great deal to this success.

References [1] Egan, D.E., Remde, J.R., Gomez, L.M., Landauer, T.K., Eberhardt, J. and Lochbaum, C.C. (1989). Formative design-evaluation of SuperBook. ACM Transactions on Information Systems, Vol. 7, No. 1, 30-57. [2] Nielsen, J., (1989). Usability Engineering at a Discount. Proc. Third International Conference on Human-Computer Interaction (Boston, MA). [3] Rada, R., Murphy, C. (1990). Searching versus browsing in Hypertext. University of Liverpool Department of Computer Science. [4] Remde, J.R., Gomez, L.M., & Landauer, T.K. (1987). SuperBook: an automatic tool for information exploration| hypertext? In Proceedings of Hypertext'87 (Chapel Hill, N.C.), pp. 175-188. [5] Stallman, R., (1985) GNU Emacs Manual. Free Software Foundation. 4

Suggest Documents