Arts & Culture

18 pages
8 views

A Keystroke Level Analysis of Manual Map Digitizing12

of 18
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Share
Description
Abstract The acquisition of digital spatial data is a key economical factor in GIS projects. Transforming analog graphic data by manual digitizing is slow and therefore extremely expensive. The work reported here investigates the possibility of
Transcript
  A Keystroke Level Analysis of Manual   Map Digitizing 12 Peter HaunoldWerner Kuhn Department of GeoinformationTechnical University ViennaGusshausstrasse 27-29A-1040 Vienna (Austria)Fax: (+43 1) 504 3535Email: haunold@geoinfo.tuwien.ac.atkuhn@geoinfo.tuwien.ac.at Abstract The acquisition of digital spatial data is a key economical factor in GISprojects. Transforming analog graphic data by manual digitizing is slow andtherefore extremely expensive. The work reported here investigates thepossibility of applying the Keystroke-Level Model to the modeling andoptimization of manual digitizing tasks. This model predicts the time it takesan experienced user to perform routine tasks on a given system. It has beenapplied successfully for text editing tasks. Here, its suitability for manualdigitizing is being tested and additional unit tasks are determined.The suitability of the Keystroke-Level Model for manual digitizing tasksis analyzed in the context of a major national map digitizing effort. The agencyresponsible for topographic and cadastral mapping in Austria (Bundesamt fürEich- und Vermessungswesen, BEV) is currently developing a digital cadastralmap (Digitale Katastral-Mappe, DKM). Scanned cadastral maps are beingmanually digitized on screen in order to transform analog graphic data intodigital form. The article describes the principles of the model, the design of anexperiment, and encouraging first results. 1. Introduction Collecting GIS data is a difficult and time consuming process. Usually, existinganalog map data have to be transformed into digital form by complex conversionprocesses. In many application areas manual digitizing is the primary method of acquiring map data for a GIS [Marble, Lauzon, & McGranaghan, 1984]. Automatic   1 The contribution from the "Bundesamt für Eich- und Vermessungswesen"  and the supportfrom Intergraph Corporation for user interface research from Intergraph Corporation isgratefully acknowledged. 2   Haunold, P., & Kuhn, W. 1993. "A Keystroke Level Analysis of Manual MapDigitizing". In Spatial Information Theory: Theoretical Basis for GIS  . (Frank,A.U., & Campari, I., eds.), Lecture Notes in Computer Science, 1 vols., Vol.716 , Heidelberg-Berlin, Springer Verlag, pp: 406-420.  digitizing methods, e.g., scanning and vectorizing, are being developed, but so farthey can only be used in prototype implementations or special cases [Crosilla, &Piccinini, 1991]. The resulting data are often of limited value because of data qualityproblems [Connealy, 1992].The slow and expensive nature of manual digitizing processes necessitates theinvestigation of the user interface characteristics of these operations [Kuhn, 1990a;Kuhn, 1990b; Frank, in Press]. Research on human-computer interaction hasproduced many methods and models for analyzing and developing user interfaces.The hypothesis underlying the work reported here is that one of these models, theKeystroke-Level Model [Card, Moran, & Newell, 1980a; Card, Moran, & Newell,1980b] is applicable to manual digitizing tasks. If true, this would offer greatpotential for economizing on these very expensive operations [Haunold, & Kuhn,1993].The Keystroke-Level Model provides a simple method to predict the time forstandard functions performed by expert users on computer systems. The timerequired for these functions is modeled by counting keystrokes and other low-leveloperations which have known average performance times. The predictiondetermines the time spent on performing each task. Individual tasks can then beoptimized by comparing different execution strategies. The model has been usedsuccessfully for text editing tasks [Card, Moran, & Newell, 1983].A study investigating the Keystroke-Level Model as a tool to optimize manualdigitizing is run by the authors in cooperation with the Austrian agency responsiblefor topographic and cadastral mapping (Bundesamt für Eich- undVermessungswesen, BEV). The suitability of the model is being tested in anexperiment. This experiment compares predicted times from the Keystroke-LevelModel with actual performance times. In the overall scheme of planning andoptimizing the map data acquisition process, this project analyzes only elementaryactions and their performance times. These actions are at the level of pressing a key,pointing to a target or homing the hand between cursor and keyboard. Some of themonly take fractions of a second, but they are repeated very often.The paper describes the Keystroke-Level Model (section two), an experimentfor comparing predicted and observed performance times (section three) and earlyresults on the suitability of the model (section four). 2. The Keystroke-Level Model The Keystroke-Level Model provides a simple method to predict the time an expertuser needs to perform a given task. The central idea behind the model is that thetime for an expert to do a task on an interactive system is determined by the time ittakes to do the keystrokes. Therefore, one can write down the execution methodused for the task, count the number of keystrokes required, and multiply by the timea single keystroke takes in order to get the total time. Obviously other elementaryoperations must be added to the model. These operations are pointing, homing,mental preparing and a system response time [Card, Moran, & Newell, 1983].  A requirement for applying the Keystroke-Level Model is the dissection of functions into error-free repetitive tasks at the keystroke level. Only error-free tasksare predicted, because the model cannot predict where and how often errors occur.Previous experience shows that expert users rarely make errors in performing well-known tasks, therefore errors do not use much time, although this restricts theusability of the model. A second requirement is the subdivision of tasks intostandardized sequences of performance steps, called unit tasks.  2.1. The Time Prediction Problem The major goal of the Keystroke-Level Model is to predict performance times forsequences of unit tasks. The prediction problem is as follows: Given : A task; the command language of a system; the motor skillparameters of the user; the response time parameters of the system;the methods used for the task. Predict: The time an expert user will take to execute the task, provided heuses the method without error.Given a large task, a user will break it into a sequence of small, quasi-independentunit tasks. The time to do a unit task consists of two parts, the acquisition of the task and the execution of the task acquired. The total time is the sum of the time for thesetwo parts: T  task   = T  acquire + T  execute (2.1)The acquisition for a unit task is a very complex process and cannot be modeledwith the Keystroke-Level Model [Card, Moran, & Newell, 1983]. The model canonly be used for tasks performed by experienced users who do not spend timethinking about the steps to perform, i.e. acquisition-time is assumed to be zero. 2.2. Operators 3 of the Keystroke-Level Model The model uses simple keystroke level operators. The execution of one task can bedescribed with four physical-motor operators, one mental operator and a systemresponse operator. These operators are listed in Figure 1. The total execution time isthe sum of the times for each of the operators: T  execute = T  K  + T  P + T   H  + T   D + T   M  + T   R (2.2)The operators K , P , H and D are assumed to remain constant for each occurrence. The Operator K. The most frequently used operator is K , which represents akeystroke or a button pushed on a mouse or on a cursor. It refers strictly to keys andnot to characters. Therefore, pressing the "shift" or "control" key counts as aseparate K [Card, Moran, & Newell, 1980b]. Time varies with the typing skill of theuser; the range of typical values is shown in Figure 1. The total time for allkeystrokes T K used to do one task is: T  K  =  n K   t K  (2.3)Where  n K  is the number of keystrokes and  t K  is the time per keystroke.   3 "Operator" is being used here in a mathematical sense, following Card et al . The personoperating a system is called "user".  ______________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ Fig. 1. The Operators of the Keystroke-Level Model, from [Card, Moran, & Newell, 1980b] The Operator P. The operator P represents pointing to a target on a display with apointing device. Pointing time  t P varies as a function of the distance to the target,  d  ,and the size of the target,  s , according to Fitts’s Law [Card, English, & Burr, 1978]:  t P = 0.8 + 0.1 log 2 (  d   /   s + 0.5) sec (2.4)The time ranges from 0.8 to 1.5 sec, with 1.1 as an average time. This operator doesnot   include the button press that often follows pointing. The Operator H. Using different physical devices, the user will move his handsbetween them. This hand-movement is represented by the operator H ("homing"). Aconstant time  t  H  of 0.4 sec for movements between any two devices is assumed[Card, English, & Burr, 1978; Card, Moran, & Newell, 1980a]. The Operator D. The operator D represents the manual drawing of a set of straightlines using a device. D takes two parameters, the number of lines,  n  D  , and their totallength l   D . An average value is given with:OperatorDescription and RemarksTime (sec) K Keystroke or button press. Best typist (135 wpm)0.08Good typist (90 wpm)0.12Average skilled typist (55 wpm)0.20Average non-secretary typist (40 wpm)0.28Typing random letters0.50Typing complex codes0.75Worst typist (unfamiliar with keyboard)1.20 P Pointing to a target on a display with a mouse.1.10 H Homing the hand(s) on the keyboard or other device.0.40 D(  n  D  ,l   D ) Drawing (manually)  n  D   straight-line segmentshaving a total length of  l   D cm..9  n  D   + .16 l   De M Mentally preparing for executing physical actions.1.35 R(  t ) Response of   t   sec by the system . t
Related Documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x