logseq/pages/hls__icse2023-paper685_1666614263567_0.md

file:: icse2023-paper685_1666614263567_0.pdf file-path:: ../assets/icse2023-paper685_1666614263567_0.pdf

• evaluations, however, do not investigate how developers perceive the use of RRs and what are the benefits of using RRs ls-type:: annotation hl-page:: 1 hl-color:: purple id:: 635c0dc4-0d1f-4685-a72b-79d47e9f591e
• s. In this work, we report a controlled experiment carried out with 24 developers where they repaired design models in eight different tasks, with RRs being provided in half of the tasks. ls-type:: annotation hl-page:: 1 hl-color:: yellow id:: 635c0dda-2f1c-40e8-81c2-3a5a0e186183
• the provision of RRs leads to developers considering alternatives to fix the inconsistency that were not considered without RRs. ls-type:: annotation hl-page:: 1 hl-color:: green id:: 635c106a-3a7e-4366-9e8a-8ae555b187e3
• epairing models, however, is an error-prone and laborious task requiring a considerable amount of time and effort from developers ls-type:: annotation hl-page:: 1 hl-color:: green id:: 635c10b9-ae03-4476-b849-a8d3ac4bc645
• maintain quality and reduce the time and effort required for the repair, mainly on UML model ls-type:: annotation hl-page:: 1 hl-color:: green id:: 635c10c6-36a0-45fa-b353-f65287cea138
• ser studies are also relevant to analyze the implications of applying an RRs strategy from the perspective of its end-users, i.e., developer ls-type:: annotation hl-page:: 1 hl-color:: green id:: 635c10dd-be0c-4068-8a9b-c3cb50a94662
• In this paper, we report an experiment to investigate if RRs bring benefits to developers in terms of effectiveness and efficiency when repairing design models ls-type:: annotation hl-page:: 1 hl-color:: purple id:: 635c1672-e5d1-4c47-8a8c-ca1860206644
• The goal of our study is described by three research questions (RQ): RQ1. Do developers benefit from recommendations when repairing inconsistent design models? RQ2. How do developers perceive the use of repair recommendations when repairing inconsistent design models? RQ3. Do developers have preferred recommendations when repairing inconsistent design models? ls-type:: annotation hl-page:: 1 hl-color:: purple id:: 635c1695-1b9d-4d93-a931-9e5aa79d2e68
• o answer RQ1, we measure inconsistencies fixed (effectiveness) and the completion time required (efficiency). For RQ2, we asked developers to give us feedback regarding the difficulty and their confidence when repairing the models. To obtain the results of RQ3, we ask the developers to select and rank RRs and create new ones. ls-type:: annotation hl-page:: 1 hl-color:: purple id:: 635c16c1-bed9-4507-8ede-d0456ba202ee
• A RR identifies the operator, the model element, the model element property, and, optionally, a value to change the model element propert ls-type:: annotation hl-page:: 2 hl-color:: purple id:: 635c1872-ea83-4b71-b049-f8ce5c14c01f
• A. Reder and A. Egyed, “Computing repair trees for resolving inconsistencies in design models.” in ASE, M. Goedicke, T. Menzies, and M. Saeki, Eds. ACM, 2012, pp. 220–229. [Online]. Available: http://doi.acm.org/10.1145/2351676.2351707 ls-type:: annotation hl-page:: 11 hl-color:: green id:: 635c18ed-54b0-4bb3-ac37-1f7bb73abe30
• archezan, R. Kretschmer, W. K. Assunc ̧ ̃ao, A. Reder, and A. Egyed,“Generating repairs for inconsistent models,” Software and Systems Modeling, pp. 1–33, 2022. ls-type:: annotation hl-page:: 11 hl-color:: green id:: 635c18fb-44dc-4aaf-aa3d-b34f11ca5184
• R. Kretschmer, D. E. Khelladi, R. E. Lopez-Herrejon, and A. Egyed,“Consistent change propagation within models,” Software and Systems Modeling, pp. 1–17, 2020. ls-type:: annotation hl-page:: 11 hl-color:: green id:: 635c1903-1df7-42ff-bdde-0be3182f28d0
• Furthermore, the documentation provided by them was used to implement the approach using their algorithm logic and repair generator functions. Details about how RRs are generated are described in the following. ls-type:: annotation hl-page:: 2 hl-color:: green id:: 635c1947-b18e-42c3-abc2-f255adaa38eb
• Consistency Rules (CR ls-type:: annotation hl-page:: 2 hl-color:: purple id:: 635c1963-b3f4-4a10-9dc9-bddfc819eca5
• epair Recommendation ls-type:: annotation hl-page:: 2 hl-color:: purple id:: 635c1968-3888-4d4b-984f-d7ea09b99e63
• setCookingMode from CookingMod ls-type:: annotation hl-page:: 2 hl-color:: green id:: 635c1995-a04b-47fa-af08-ea7f1cdcead5
• SlowCookingMode to CookingMode, and the operation setCookingMode from SlowCookingMode. ls-type:: annotation hl-page:: 2 hl-color:: green id:: 635c19a2-fa93-4ae0-a076-3462ea5fa1a7
• These three model elements are the cause of the inconsistency and should be modified to repair the model ls-type:: annotation hl-page:: 2 hl-color:: green id:: 635c19ca-cac7-40a2-b28e-fb7c22b83427
• are abstract since they suggest modifying the signature without saying how ls-type:: annotation hl-page:: 2 hl-color:: green id:: 635c1d73-f97e-49b9-bc27-2df45e6d672f
• In these cases, a developer is required to define a value for the RRs ls-type:: annotation hl-page:: 2 hl-color:: purple id:: 635c1d7e-2f3e-4269-9275-84a320fa61e6
• randomly ordered ls-type:: annotation hl-page:: 3 hl-color:: purple id:: 635c1d90-eafe-422a-b941-e67a50b8b7bf
• To the best of our knowledge, however, the use of repair recommendations has never been evaluated from the developers’ perspective ls-type:: annotation hl-page:: 3 hl-color:: yellow id:: 635c1e65-6cd9-4980-8840-c8f2228f6196
• we designed our experiment with the goal to collect and analyze these implications ls-type:: annotation hl-page:: 3 hl-color:: yellow id:: 635c1e75-dc61-4a8e-b14d-b7623255328a
• effectiveness and efficiency ls-type:: annotation hl-page:: 3 hl-color:: purple id:: 635c1f70-a208-45c5-b020-09169345fa7b
• Fallessi et al. [32], for instance, proposes the R3 scheme (Real, Relevant, and Recen ls-type:: annotation hl-page:: 3 hl-color:: purple id:: 635c1fa5-c090-4d7e-9833-fa5323736fe9
• mpirical software engineering experts on the use of students and professionals in experimen ls-type:: annotation hl-page:: 11 hl-color:: green id:: 635c1fb1-3117-49c0-91b5-1cca2bafbef1
• This scheme is important as the experience of participants may impact the results of the experiment. ls-type:: annotation hl-page:: 3 hl-color:: purple id:: 635c1fdd-7307-4ffb-923c-0d760f93839c
• The main goal of our experiment is to analyze if developers benefit from the provision of RRs and how developers perceive the difference between using RRs compared to not using them. We also aim at investigating if developers have preferred RRs when fixing inconsistent models. ls-type:: annotation hl-page:: 3 hl-color:: yellow id:: 635c1fff-cda0-492c-9dd6-24c4a0857d9f hl-stamp:: 1666981891291
• The dependent variables are: T: Time spent on the tasks; I: number of inconsistencies per task; IF: number of inconsistencies fixed per task. ls-type:: annotation hl-page:: 3 hl-color:: purple id:: 635c20f4-3fae-4e26-bfad-6e9dcbc01120
• RC: repairs created by participants in WR tasks ls-type:: annotation hl-page:: 3 hl-color:: purple id:: 635c2116-d3d4-4f01-8542-aeb8a05b5816
• 60 ls-type:: annotation hl-page:: 3 hl-color:: yellow id:: 635c213f-585f-465f-b19b-bb2f60a862c0
• For the EFT metric, the null hypothesis H0 states that RR and WR have the same effectiveness. ls-type:: annotation hl-page:: 4 hl-color:: green id:: 635c21a4-7325-4c00-8f83-2fa820261048
• effectiveness (EFT) ls-type:: annotation hl-page:: 3 hl-color:: green id:: 635c21bc-c925-42c9-820b-0252d0cf5f98
• efficiency (EFY) ls-type:: annotation hl-page:: 3 hl-color:: green id:: 635c21c0-5cd7-462f-b477-bee8e77ae9f5
• We analyze these results to understand if when RRs are provided, participants perceive tasks to be less or more difficult while having less or more confidence in performing them. ls-type:: annotation hl-page:: 4 hl-color:: yellow id:: 635c21f0-a899-4d57-ba56-50412ee9ff65
• RQ3. Do developers have preferred recommendations when repairing inconsistent design models? ls-type:: annotation hl-page:: 4 hl-color:: yellow id:: 635c55a9-61e4-4da2-8312-6f645abb8be4
• one ideal RR for each inconsistenc ls-type:: annotation hl-page:: 4 hl-color:: green id:: 635c55e6-af53-4e24-ad65-fe504e39c393
• R3 scheme (Real, Relevant, and Recent ls-type:: annotation hl-page:: 4 hl-color:: green id:: 635c564f-5205-49bf-86b1-2d2ffd16bcd8 hl-stamp:: 1666995796072
• The experimental package used was designed following a strategy closely related to the strategy for mitigating threats to validity (see Section III-C) ls-type:: annotation hl-page:: 4 hl-color:: yellow id:: 635c578f-3138-4f2f-bae3-4f96bfa3925b
• , ls-type:: annotation hl-page:: 5 hl-color:: red id:: 635c5930-942d-47cd-81bd-2d6d1a55068b
• When designing the tasks, we had to consider the diagrams to be used and their inconsistencie ls-type:: annotation hl-page:: 5 hl-color:: green id:: 635c5977-ab20-49fd-9953-7863e2a51328
• Rs were generated for all tasks ls-type:: annotation hl-page:: 5 hl-color:: yellow id:: 635c5fb6-5e69-464b-acb8-1e4727d80780
• hree experimenters were in the room observing if participants were following the guidelines explained. ls-type:: annotation hl-page:: 6 hl-color:: green id:: 635c60be-5aea-4388-9c52-90b140fcba56
• he order and the type of the tasks changed for each document ls-type:: annotation hl-page:: 6 hl-color:: green id:: 635c60e8-b74a-40fc-8a06-363801125fa1
• internal, external, and conclusion ls-type:: annotation hl-page:: 6 hl-color:: green id:: 635c6100-2174-4480-b9ec-bd76c7be7c1d
• Construct Validity ls-type:: annotation hl-page:: 6 hl-color:: green id:: 635c6147-a614-451c-8370-cee335e307d5
• the models from the dataset were already evaluated and used in other studies ls-type:: annotation hl-page:: 6 hl-color:: yellow id:: 635c6265-d1a3-49ad-a14e-cc3a9b56490f
• task difficulty and completion time ls-type:: annotation hl-page:: 6 hl-color:: green id:: 635c62b1-e7ea-4d50-8fee-8772afcdd2a6
• articipants from the pilot studies also reported that most tasks were easy, so we re-balanced these tasks, increasing their difficulty ls-type:: annotation hl-page:: 6 hl-color:: green id:: 635c62de-c5e6-43ea-bfbc-e5e3da4fbc6a
• were part of the generated repairs from the same RR task ls-type:: annotation hl-page:: 6 hl-color:: green id:: 635c631a-3832-4ad0-89db-8463462bbd5d
• we argue that even inexperienced developers should be able to use RRs ls-type:: annotation hl-page:: 6 hl-color:: purple id:: 635c636e-caf6-4e55-a574-d1d37fe673e7
• RQ1. Do developers benefit from recommendations when repairing inconsistent design models? ls-type:: annotation hl-page:: 7 hl-color:: yellow id:: 635c7160-423e-43bd-8b64-9c0e7652798c
• feel that using RRs is better than not using them. ls-type:: annotation hl-page:: 9 hl-color:: yellow id:: 635c7497-8669-46ad-8fed-d78614d0893c
• This improvement is more evident when the repairs required are complex. ls-type:: annotation hl-page:: 10 hl-color:: yellow id:: 635c75f8-1804-44b9-b6aa-7093a3d523ae