Design Points: Finding Latent Design Concepts in Developer Discussions

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 Design Points: Finding Latent Design Concepts in Developer Discussions Anonymous Author(s) ABSTRACT A software system’s design determines many of its properties, such as maintainability and performance. An understanding of design is needed to maintain system properties as changes to the system occur. Unfortunately, many systems do not have up-to-date design documentation and approaches that have been developed to recover design often focus on how a system works by extracting structural and behaviour information rather than information about what a system is intended to do to meet its desired design properties. In this paper, we explore whether latent design that is talked about in on-line developer discussions can be identified automatically so it can subsequently be provided to help developers perform change tasks on a system. Specifically, we perform a detailed analyses of one form of developer discussions, pull requests. We show that in these discussions, design often arises as design points, statements about aspects of design for which developers are called upon to make a decision. To build a foundation on which tools can be built to leverage latent design information for developers, we investigate and introduce a classifier that can identify design points in pull requests with high accuracy (average F1 and AUC scores of 0.62 and 0.87, respectively). We provide initial results that show humans identify similar design points in a discussion, whilst also agreeing with the automatic classifier in the majority of cases. We also show the promise of identified design points in aiding developer tasks by investigating a proof-of-concept recommender that can learn project-specific design considerations from previous pull requests.