wcre - oct 2007 - 1 temporal model explorer - farah and lethbridge temporal exploration of software...
TRANSCRIPT
WCRE - Oct 2007 - 1Temporal Model Explorer - Farah and Lethbridge
Temporal Exploration of Software Models
HannaFarah
IBM Ottawa
TimothyLethbridgeUniversity of Ottawa
WCRE - Oct 2007 - 2Temporal Model Explorer - Farah and Lethbridge
Temporal Model Exploring (TME)Key Ideas
Smoothly slide a document back and forth as it appeared in time
Mark spots of interest (snapshots) to jump back and forth to
Filter the edit events– E.g. filter out position changes in a diagram so items
stay in their final position
Annotate history easily
WCRE - Oct 2007 - 3Temporal Model Explorer - Farah and Lethbridge
WCRE - Oct 2007 - 4Temporal Model Explorer - Farah and Lethbridge
Hypotheses
Participants using TME– Can solve modeling problems
• more quickly• more accurately
– Prefer it to not using TME
Participants using final positioning filtering in TME– Can solve modeling problems
• more quickly• more accurately
– Prefer it to not using final positioning
WCRE - Oct 2007 - 5Temporal Model Explorer - Farah and Lethbridge
Study
Participants– Twelve IBM employees
• Managers• Software developers• Coop students
Experiment steps– Participants were first given a consent form– Given a short training session in the TME feature – Given three class diagrams and asked questions
(more details of experiment blocking) (more details of questions asked)
WCRE - Oct 2007 - 6Temporal Model Explorer - Farah and Lethbridge
Measured performance results
No statistically significant difference in speed or accuracy
– When using TME or not– When using original position
– (more details of the measured results)
WCRE - Oct 2007 - 7Temporal Model Explorer - Farah and Lethbridge
Questionnaire results 1
I found that the TME feature helped me understand class diagrams more quickly
– 4.2 / 5
Using the page-up and page-down keys, I found that a useful set of steps in the development of the model (snapshots) were presented
– 4.1 / 5
I would use the TME feature if it was available to me in my work environment and I was asked to understand a class diagram.
– 3.9 / 5
WCRE - Oct 2007 - 8Temporal Model Explorer - Farah and Lethbridge
Questionnaire results 2
TME feature resulted in me taking a longer time to answer the questions
– 2.4 / 5
TME feature was awkward to use– 1.7 / 5
WCRE - Oct 2007 - 9Temporal Model Explorer - Farah and Lethbridge
Future work 1
Similar study with larger models that have multiple diagrams
Giving the TME tool to a group of users– use it in their daily work– surveying them to get more feedback
Additional filters– i.e. filter on name changes
WCRE - Oct 2007 - 10Temporal Model Explorer - Farah and Lethbridge
Future work 2
Automatically detecting snapshots based on:– Context– Time
Deploying the TME feature in word processors and other tools
WCRE - Oct 2007 - 11Temporal Model Explorer - Farah and Lethbridge
Summary
TME: Temporal exploration using a slider– Final positioning filtering– Adding, removing, or editing at ANY time of
• snapshots (controlled granularity)• temporal annotations
– Simple, would not add to ‘feature overload’
Study– Users gave positive feedback
A patent has been filed on key aspects of this work
WCRE - Oct 2007 - 12Temporal Model Explorer - Farah and Lethbridge
Questions?
WCRE - Oct 2007 - 13Temporal Model Explorer - Farah and Lethbridge
Supplemental slides
WCRE - Oct 2007 - 14Temporal Model Explorer - Farah and Lethbridge
Experiment blocking (back)
Blocking by– Problem– Problem sequence– Person creating original model– Model– Participant ability– Treatment pattern
• Order of No TME TME with final position TME with original position
WCRE - Oct 2007 - 15Temporal Model Explorer - Farah and Lethbridge
Experiment blocking grids (back)
Table 1: Allocation of participants to models and treatments
T1: Only final
diagram
T2: TMEP original
pos
T3: TMEP
final pos Investments-
A A1, E3 G1, K2 C2, I2 Investments-
B B1, F3 H1, J2 D2, L2 Elections-A C1, H3 F2, I3 A3, J3 Elections-B D1, G3 E2, L1 B3, K1
Airline
I1, K3, J1, L3
A2, C3, B2, D3
E1, G2, F1, H2
Table 2: Allocation of participants, controlling for experience
Participant Experience Treatment
pattern Problem sequence
A Higher 1 t23 IAE B Lower 1 t23 IAE C Higher 1 t32 EIA D Lower 1 t32 EIA E Higher t32 1 AEI F Lower t32 1 AEI G Higher t23 1 IAE H Lower t23 1 IAE I 1 t32 AIE J 1 t23 AIE K t23 1 EIA L t32 1 EIA
WCRE - Oct 2007 - 16Temporal Model Explorer - Farah and Lethbridge
Sample question asked (back)
Imagine the airline regularly flies from Ottawa to Toronto, but wants to create an extra charter flight tomorrow from Ottawa to Toronto. To enable this, the system will have to create instances of one or more classes. Which classes will the system have to make instances of? No need to worry about booking any passengers or assigning crew.
Figure 2: The Airline System class diagram used in the evaluation of TME
WCRE - Oct 2007 - 17Temporal Model Explorer - Farah and Lethbridge
More details of the measured data (back)
Table 3: Measures of speed and accuracy. All values are normalized, where the average for all participants over all models =1. A value greater than 1 means faster or more accurate.
Hypothesis and measure
Mean Max Min Std. Dev
95% Conf
Experts’ Mean
H1a: Time taken by participants in treatment T1 (without TME)
0.99 1.45 0.57 0.24 0.14 Same
Time taken by participants in T1 and T2 (TME)
1.00 1.21 0.77 0.12 0.07 Same
Conclusion: No significant difference H1b: Accuracy of participants in treatment T1 (without TME)
1.00 1.21 0.74 0.15 0.08 1.05
Accuracy of participants in T1 and T2 (TME)
1.00 1.13 0.90 0.07 0.04 0.98
Conclusion: No significant difference in general, but With experts showing a slight and not-statistically- significant loss of accuracy with TME
H2a: Time taken by participants in T2 (original positions)
1.04 1.51 0.58 0.26 0.14 Same
Time taken by participants in T3 (final position filter)
0.97 1.21 0.76 0.14 0.08 Same
Conclusion: Minor but not-statistically-significant Improvement in speed
H2b: Accuracy of participants in treatment T2 (original pos.)
1.01 1.26 0.80 0.14 0.08 1.00
Accuracy of participants in treatment T3 (final position)
0.98 1.08 0.79 0.09 0.05 0.98
Conclusion: No significant difference