Greenberg, S. (1993). The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-based Tools. New York: Cambridge University Press.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... the physical world, to amplify capabilities (e.g. written lists as memory extensions) or to translate problems into more manageable form [10, 17] Software environments have also taken lessons from environments for physical tools for example, a blacksmith s shop [12] a carpenter s workbench [8], or a chef s kitchen [1] by incorporating some of the same design properties: effective use of space [13] conceptual organization that reflects functional relationships, and exploitation of conventions [19] It is easy to see how spatial and functional organization of tools in a physical ....

S. Greenberg. The computer user as toolsmith : the use, reuse, and organization of computer-based tools. Cambridge University Press, Cambridge, UK, 1993.

....future behavior. A second approach consists of learning frequent sequences of actions performed by the user, which is explained in section 4. 2 Discovering Rules in Shell Usage Analyzing the behavior of shell users, either by hand or by applying machine learning techniques, has been done before. [4] analyses di erent types of users by applying statistical techniques. 2] use 5 di erent machine learning techniques to build classi ers for predicting the next command, given the history of previous commands. As mentioned in [2] a machine learning technique that uses a richer data and ....

Saul Greenberg. The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-Based Tools. Cambridge Series on HumanComputer Interaction. Cambridge University Press, 1993. QA76.9H85G73.

....the system; their only control is to turn the adaptive menus on off and to reset the data collected in the user model. Two examples in the research literature that incorporate intelligence are Greenberg s work on Workbench that makes frequently used commands easily accessible for reuse [4] and Linton, Joy, and Schaefer s recommender system that alerts users to functionality currently being used by co workers doing similar tasks [9] No user testing has been reported in the literature for any of the interfaces given above except for Training Wheels. DESIGN OVERVIEW ....

.... (rather than a predetermined static small interface) Not only do users typically use very few features [9, 12] but the overlap between the command vocabulary of different users is minimal, even for users in the same group who perform similar tasks and who have similar computer expertise [4]. Users also generally customize very little, likely because customization facilities are often powerful and complex in their own right and therefore require time both for learning and for doing the customization. Mackay found the latter to be true in the case of UNIX customization [10] ....

Greenberg, S. (1993). The computer user as toolsmith: The use, reuse, and organization of computer-based tools. Cambridge: Cambridge University Press.

....had larger numbers of unique commands. Approximately one out of five commands were the same 32 as the previous command executed (that is, the user repeated the last command 20 of the time) The second (Greenberg) dataset is a larger collection of user activities with the csh UNIX shell [Gre88, Gre93] It contains more than twice as many users and approximately twice as many commands overall. This dataset was acquired after completing the tests on the first, but we will present the results of both datasets when appropriate. 3.3.1.3 Earlier Results In previous work [DH97a, DH97b] we ....

....cross user training to leverage the experience of others. Unfortunately, preliminary experiments indicate that, at least for this dataset, starting with the average of all other users command prediction tables does not improve predictive accuracy. This result matches with those of Greenberg [Gre93] and Lee [Lee92] who found that individual users were not as well served by systems tuned for best average performance over a group of users. The goal of our work has been to discover the performance possible without domain knowledge. This can then be used as a benchmark for comparison against ....

[Article contains additional citation context not shown here]

Saul Greenberg. The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-based Tools. Cambridge University Press, New York, NY, 1993.

....to have a relatively higher research profile [2] Software development is memory intensive, so external memory use is, as one might expect, pervasive. Individual, personal external memory use is common. External memory can be found in something as simple as command line histories (e.g. Greenberg [275]) or browser histories (e.g. Tauscher et al. 633] In these, the memory acts as a type of external working memory [95] The study by Bowdidge et al. 64] already mentioned in the section on strategic artifact use, also provides good examples of the use of external working memories in the form ....

....interfaces provide a history of commands or responses. An example of such a history is the scroll of paper that used to cascade down behind the old teletype machines. The modern equivalent of this is the terminal emulator buffer and its scrollbar. History also includes shell command histories [275]. Breadcrumbing, dog earing, tick marks. A special form of visitation history; people often mark their trail or important places they ve been to indicate relevance [454] They also regularly record The terms snapshot , trace , and path are not further defined here as they are meant to be ....

[Article contains additional citation context not shown here]

Greenberg, S. The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-based Tools. Cambridge University Press, 1993.

....pages (those that are considered likely to be cacheable) can be reached by examining the links of the current page and its two immediate predecessors in the current user s request stream. For comparison, the per client average recurrence rate (that is, the rate at which requests are repeated [35]) for clients making more than 500 prefetchable requests was 49.2 , and for all clients was 22.5 (since some clients made very few requests) This means that if each client utilized an infinite cache that that passively recorded cacheable responses, the per client average hit rate on prefetchable ....

S. Greenberg. The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-based Tools. Cambridge University Press, New York, NY, 1993.

.... in the physical world, to amplify capabilities (e.g. written lists as memory extensions) or to translate problems into more manageable form [9, 15] Software environments have also taken lessons from environments for physical tools for example, a blacksmith s shop [11] a carpenter s workbench [7], or a chef s kitchen [2] by incorporating some of the same design properties: effective use of space [12] conceptual organization that reflects functional relationships, and exploitation of conventions [16] It is easy to see how spatial and functional organization of tools in a physical ....

....as systems for enduser programming. TOOL USE IN THE PHYSICAL WORLD Researchers in a variety of fields, including social psychology [18] anthropology [11] animal cognition [10, 19] experimental psychology [22] artificial intelligence [2] design disciplines [13, 14, 16] and of course HCI [7, 9], have examined the concept of tool use. Descriptions are not always consistent across fields, but a few strong themes recur. We can begin with Beck s definition, widely accepted in non human primate cognition research [3] Thus tool use is the external employment of an unattached environmental ....

Saul Greenberg. The computer user as toolsmith : the use, reuse, and organization of computer-based tools. Cambridge University Press, Cambridge, UK, 1993.

....fully utilise the power of the shell, either because they do not know all the tools the shell can provide or they do not want to go through the e ort of using the tools. Since the shell is such a complex and powerful user interface many people have investigated how users use this shell. Greenberg [6] collected logs from 168 di erent users, classi ed in four categories: computer scientists, experienced programmers, novice programmers and users performing no programming tasks (such as people from the administration et cetera) He used the standard statistical techniques to analyse e.g. which ....

....75 of the commands correctly, and about 50 when also the extra parameters of the command (switches, lenames et cetera) are predicted. The accuracy is even higher when multiple predictions are shown. Although this looks promising, there are some drawbacks: the user often uses short commands [6]. For those commands, it is as little e ort to type the command as to select or verify a prediction, so nothing is gained. long commands are dicult to predict correctly. Verifying and correcting the prediction can take as much time as typing the command (or even more) a special shell is ....

Saul Greenberg. The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-Based Tools. Cambridge Series on Human-Computer Interaction. Cambridge University Press, 1993. QA76.9H85G73.

....to new positions rather than abruptly shifting location. Patterns of use of the Web Studies of Web use report patterns of behaviour that support the idea that users are often interested in assembling and revisiting a limited number of pages what Saul Greenberg calls a recurrent system [19]. According to research in U.Calgary s Grouplab [20] 1. About 30 of all logged navigation activity involve use of the Back button 2. About 60 (s.d. 9 ) of all page visits are to pages a person has been to before 3. A list of 6 or so URLs just visited contain the majority of pages a person ....

Greenberg, S. The Computer User As Toolsmith : The Use, Reuse and Organization of Computer-Based Tools (Cambridge Series on Human-Computer Interaction, No 6) 1993

....of an application, it must be focused by observational and experimental studies of Web use. Studies of Web use report patterns of behaviour that support the idea that users are often interested in assembling and revisiting a limited number of pages what Saul Greenberg calls a recurrent system [19]. According to research in U.Calgary s Grouplab [20] About 30 of all logged navigation activity involve use of the Back button . About 60 (s.d. 9 ) of all page visits are to pages a person has been to before . A list of 6 or so URLs just visited contain the majority of pages a person will ....

Greenberg, S. The Computer User As Toolsmith: The Use, Reuse and Organization of Computer-Based Tools (Cambridge Series on Human-Computer Interaction, No 6) 1993

....out redundant patterns Fig. 3. Finding frequent behaviour The next section shows an example of the use of this system. 4 Automated Macro Construction Unix shells are still often used to interact with a Unix system. There have been previous e orts to analyse how users use this shell interface. [4] uses statistical techniques to analyse shell usage. Most user modeling approaches use machine learning techniques to learn a classi er that, given the history of previous typed commands, predicts what the next command will be. 2] compare di erent approaches, 5] presents an approach which can ....

....of these actions and rewrite these results so that we can present to the user a macro de nition for the sequence of actions. It is up to the user to decide which of these macro descriptions to use. We chose this approach because previous work shows that the user often uses short commands [4]. For those commands, it is as little e ort to type the command as to select or verify a prediction. long commands are dicult to predict correctly. Verifying and correcting the prediction can take as much time as typing the command (or even more) a special shell is necessary to integrate ....

[Article contains additional citation context not shown here]

Saul Greenberg. The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-Based Tools. Cambridge Series on Human-Computer Interaction. Cambridge University Press, 1993. QA76.9H85G73.

....to present to the user. 6.1.3 Pruning duplicates from a recency list The sequential ordering by recency method described in Section 6.1.2 contains every URL the user has visited, including revisits to the same URL. These duplicates occupy valuable space on a history list of a limited length. Greenberg (1993a) applied two 105 URL Visit No. URL Navigation Action a) Sequential ordering by recency 16 acsl.cs.uicu.edu kaplan applets.html Open URL 15 acsl.cs.uicu.edu kaplan worlds environment.html Open URL 14 acsl.cs.uicu.edu kaplan worlds.html Open Hotlist 13 www.acm.org sigchi chi96 forms ProcFormat.html Open URL 12 ....

Greenberg, S. (1993a). The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-based Tools. Cambridge series on human-computer interaction. Cambridge University Press.

....characters to match when applied to all of the data. Note that command completion when given 0 initial characters is just command prediction. Similar overall performance of IPAM can be seen in Figures 8 and 9 which shows command prediction accuracy and completion over a larger csh dataset (Greenberg 1988; 1993), containing more than twice as many users, and approximately twice as many commands overall. Again, IPAM outperforms the simpler Bayes and MRC algorithms, even when using the setting for alpha determined by the parameter study over the first dataset. Discussion While not shown, the results ....

....lets us examine cross user training to leverage the experience of others. Unfortunately, preliminary experiments indicate that, at least for this dataset, starting with the average of all other users command prediction tables does not improve predictive accuracy. This result matches with those of Greenberg (1993) and Lee (1992) who found that individual users were not as well served by systems tuned for best average performance over a group of users. The goal of our work has been to discover the performance possible without domain knowledge. This can then be used as a benchmark for comparison against ....

[Article contains additional citation context not shown here]

Greenberg, S. 1993. The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-based Tools.

....navigation burdens. Pages can be returned to with little effort, and they can show users where they have been. However, today s design of history mechanisms tend toward ad hoc approaches that do not appear to take advantage of previous research into history support within user interfaces e.g. Greenberg (1993) and Lee (1992) In particular, their designs are not based upon actual studies of how people revisit Web pages, and their actual use has been examined only superficially. Our goal is to place the design of history mechanisms within browsers on a more empirical footing. We had three sub goals. ....

....Web browsers contain a history mechanism, we do not know how often people revisit their pages. In other domains, research has quantified this repetition of user actions e.g. telephone numbers dialed (57 ) information retrieved in a technical manual (50 ) and Unix command lines entered (75 ) (Greenberg, 1993). We analyzed our own data and the Catledge and Pitkow (1995) data to derive a recurrence rate R: the probability that any URL visited is a repeat of a previous visit. An overall recurrence rate of 58 (s = 9 ) shows that users do revisit Web pages, and qualifies Web browsing as a recurrent ....

[Article contains additional citation context not shown here]

Greenberg, S. (1993). The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-based Tools. Cambridge series on human-computer interaction. Cambridge University Press.

....navigation burdens. Pages can be returned to with little effort, and they can show users where they have been. However, today s design of history mechanisms tend toward ad hoc approaches that do not appear to take advantage of previous research into history support within user interfaces e.g. Greenberg (1993) and Lee (1992) In particular, their designs are not based upon actual studies of how people revisit Web pages, and their actual use has been examined only superficially. Our goals are to provide an empirical foundation that can be applied to the design of better history mechanisms within ....

....This is important, as it gives us a bound for how useful a history system can be. In other domains, research has quantified this repetition of user actions e.g. telephone numbers dialed (57 ) how information is retrieval in a technical manual (50 ) and how Unix command lines are entered (75 ) (Greenberg, 1993). We analyzed our own data and the Catledge and Pitkow (1995) data to derive the recurrence rate R: the probability that any URL visited is a repeat of a previous visit. Each user s data was analyzed independently, and the statistics below represent either averages across all users, or ....

[Article contains additional citation context not shown here]

Greenberg, S. (1993). The computer user as toolsmith: The use, reuse, and organization of computer-based tools.

....within web browsers are only useful if users actually repeat their activities. Yet we do not know how often people revisit their pages. This is important, as it gives us a bound for how useful a history mechanism can be. In other domains, research has quantified this repetition of user actions [5] e.g. telephone numbers dialed (57 ) how information is retrieved in a technical manual (50 ) and how Unix command lines are entered (75 ) We analyzed our own data and the Catledge and Pitkow [2] data to derive the recurrence rate R: the probability that any URL visited is a repeat of a ....

....and 61 (s = 9 ) for 55 subjects from the Catledge and Pitkow study. These numbers clearly show that users revisit web pages heavily, although it also means that 40 of all page navigations are to new pages. These recurrence rates qualify web browsing activity as a recurrent system. Greenberg [5] coined this term to characterize systems where users predominately repeat activities they had invoked before, while still selecting new actions from the many that are possible. In post study interviews, people gave us their major reasons for visiting new pages and revisiting old ones. They ....

[Article contains additional citation context not shown here]

Greenberg, S. (1993). The computer user as toolsmith: The use, reuse, and organization of computer-based tools. Cambridge University Press.

No context found.

Greenberg, S. (1993). The Computer User as Toolsmith: The Use, Reuse, and Organization of Computer-based Tools. New York: Cambridge University Press.

No context found.

Greenberg, S. (1993). The computer user as toolsmith: The use, reuse, and organization of computer-based tools. Cambridge University Press, Cambridge.

No context found.

Saul Greenberg. The computer user as toolsmith : the use, reuse, and organization of computer-based tools. Cambridge University Press, Cambridge, UK, 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC