logseq/pages/hls__COLA-D-23-00096_1699177983001_0.md

file:: COLA-D-23-00096_1699177983001_0.pdf file-path:: ../assets/COLA-D-23-00096_1699177983001_0.pdf

• AI-based Clustering of Similar Issues in GitHub’s Repositories Article Type: Full Length Article Keywords: Similar Issues; GitHub; Machine Learning; Maintenance; Clustering Corresponding Author: Hamzeh Eyal Salman Mutah University Karak, JORDAN Corresponding Author Secondary Information: Corresponding Author's Institution: Mutah University ls-type:: annotation hl-page:: 1 hl-color:: green id:: 654cf55c-0711-4113-8546-4963e635cb90
• I-based Clustering of Similar Issues in GitHub’s Repositories ls-type:: annotation hl-page:: 3 hl-color:: green id:: 654cf565-f328-4e8b-9126-0ca88fe5ce9a
• The attractive repositories on Github receive a large number of issues daily. ls-type:: annotation hl-page:: 3 hl-color:: green id:: 654cf582-ed5c-4879-849b-0e06165c581e
• Assigning similar issues individually to different developers for validating and fixing introduces inconsistencies when asynchronously independent developers fix them, in addition to slowing the fixing process. ls-type:: annotation hl-page:: 3 hl-color:: purple id:: 654cf595-c370-437b-9b2f-dabaf84d8c0e hl-stamp:: 1699542423639
• grouping similar issues into clusters and assigning each cluster to the same and appropriate developer/team speeds up the fixing process. ls-type:: annotation hl-page:: 3 hl-color:: purple id:: 654cf5c4-718b-4d37-aaea-b4658b5ed7ad
• machine learning algorithm-based approach has been proposed to support issue management on GitHub by grouping similar issues together ls-type:: annotation hl-page:: 3 hl-color:: purple id:: 654cf5d6-940f-4078-a6fe-65af2bca9afb
• adding new features, reporting bugs to be fixed, or asking a question about the capability of the software. ls-type:: annotation hl-page:: 3 hl-color:: green id:: 654cfab8-9162-4963-a9c3-30dd587e4443 hl-stamp:: 1699543739600
• These issues differ in their nature and quality (asking for support, for improvement of a functionality, bug reporting). Therefore, the issues management in such cases will be harder ls-type:: annotation hl-page:: 3 hl-color:: green id:: 654cfb34-605f-4a2b-9758-c4a94e7d410b
• (i) to speed up the process of fixing issues and reduce the turnaround time for fixing, ls-type:: annotation hl-page:: 4 hl-color:: green id:: 654cfb5d-d6e3-4bec-9e2b-bfa61a354b5f
• ii) when related issues are grouped together, this speeds up the assignment process by finding the appropriate reviewer(s) for a cluster instead of25 individual issues, especially, if the assignment process is done manuall ls-type:: annotation hl-page:: 4 hl-color:: green id:: 654cfb6a-d45c-4079-874d-2e5b2b0633bf
• M. Borg, L. Jonsson, E. Engstrom, B. Bartalos, A. Szab’o, Adopting automated bug assignment in practice: A longitudinal case study at ericsson, ArXiv abs/2209.08955. ls-type:: annotation hl-page:: 21 hl-color:: green id:: 654cfb75-1118-48da-92f2-0850c197b241
• Hamzeh Eyal Salman ls-type:: annotation hl-page:: 3 hl-color:: green id:: 654cfb86-d7a3-4fec-870a-b6a38d62a5ee
• o consolidate similar issues into a single, well-defined issue that covers all the variations and aspects of the problem ls-type:: annotation hl-page:: 4 hl-color:: green id:: 654d0919-93c4-4619-81ba-14a90b26610c
• The difference between similar and duplicate issues is that the former refers to a set of issues with some common characteristics or are closely related, but they are not30 necessarily exact duplicates. ls-type:: annotation hl-page:: 4 hl-color:: green id:: 654d0928-2741-4db7-9e95-5a6eda44bf53
• . Liao et al. [7] studied the impact of labeled or tagged issues on their management using six popular repositories ls-type:: annotation hl-page:: 4 hl-color:: green id:: 654d0956-2e2f-4b33-90ba-0d2785d223e8
• GitGub ls-type:: annotation hl-page:: 4 hl-color:: red id:: 654d098f-d572-4f9d-af6e-d8c141179f66 hl-stamp:: 1699547537664
• ull-request (PR) ls-type:: annotation hl-page:: 4 hl-color:: green id:: 654d09d4-103e-48cd-bdcc-af0c0437e17d hl-stamp:: 1699547606864
• ssue reporting ls-type:: annotation hl-page:: 4 hl-color:: green id:: 654d09da-6d06-4e05-9f35-4707b04e02eb
• Figure 1: ls-type:: annotation hl-page:: 5 hl-color:: yellow id:: 654d0a15-8279-4622-85d1-a8d2abb01f6d hl-stamp:: 1699547670418
• he ls-type:: annotation hl-page:: 5 hl-color:: red id:: 654d0a34-dfd6-453b-9354-0d8407dfb7d1 hl-stamp:: 1699547701974
• ssues1,2,3, 4 ls-type:: annotation hl-page:: 5 hl-color:: red id:: 654d0a7d-0b3b-4802-8d5b-dcc99ac20fe1
• Figure 2: ls-type:: annotation hl-page:: 6 hl-color:: red id:: 654d0aa6-3ed1-4ec6-85a0-548508d471ae
• cookie handling ls-type:: annotation hl-page:: 5 hl-color:: green id:: 654d0b36-f779-484b-920d-0576fc0ab546 hl-stamp:: 1699547960656
• there is no research work in the literature addressing the problem of clustering similar issues into groups in GitHub. ls-type:: annotation hl-page:: 7 hl-color:: yellow id:: 654d0b5e-8aa3-4eb4-ac26-83e9fc224d13 hl-stamp:: 1699548001102
• natural language processing (NLP ls-type:: annotation hl-page:: 7 hl-color:: green id:: 654d0bc1-6a33-4f0d-bbfc-a360a082d835 hl-stamp:: 1699548099501
• information retrieval (IR) ls-type:: annotation hl-page:: 7 hl-color:: green id:: 654d0bc7-ac78-4001-932d-c868c79e9f9c
• machine learning (ML) ls-type:: annotation hl-page:: 7 hl-color:: green id:: 654d0bcb-ee33-4375-b1f3-f80a968cfe7a
• The idea behind using IR and NLP is to find textual overlapping between bug report queries and already existing report queries in the repository ls-type:: annotation hl-page:: 7 hl-color:: green id:: 654d0c2a-31c0-4c27-8c7f-193094faea16
• Jacard, cosine, and dice measur ls-type:: annotation hl-page:: 7 hl-color:: green id:: 654d0ce7-8016-4942-b414-3bd21ea81e0d
• ug reports i ls-type:: annotation hl-page:: 7 hl-color:: green id:: 654d0cf0-42bc-4786-b1fc-31032e6e8a58
• This ranked list allows triggers to compare the incoming bug report with this list to discover the duplication ls-type:: annotation hl-page:: 7 hl-color:: green id:: 654d0d14-426e-4710-bb84-37f6f5dfb4ca
• uplicate bug report detection with a combination of information retrieval and topic modeling, i ls-type:: annotation hl-page:: 22 hl-color:: green id:: 654d0ef4-c76f-4b08-916d-0ebb44b6a959
• In this study, title and body features are only considered during the clustering process while Label is ignored as it is repetitive and missed in many issues. ls-type:: annotation hl-page:: 8 hl-color:: green id:: 654d0f20-3e2b-4b6d-bddd-015d4a3ad297
• Then,175 T ls-type:: annotation hl-page:: 8 hl-color:: red id:: 654d0f3f-322b-4db1-9b86-509a6d740969 hl-stamp:: 1699548993168
• As mentioned above, these important features are the title and body of each issue. ls-type:: annotation hl-page:: 8 hl-color:: green id:: 654d0f66-4f5e-4b82-a369-653543dd3c36 hl-stamp:: 1699549038638
• ground-truth dataset. ls-type:: annotation hl-page:: 12 hl-color:: green id:: 654d0ffb-b347-4fbe-82ed-ff7c00d2e88e
• refuted ls-type:: annotation hl-page:: 12 hl-color:: yellow id:: 654d100f-1b43-44ec-ab81-2e0005a65b48 hl-stamp:: 1699549201228
• owever, such a dataset does not exist yet. ls-type:: annotation hl-page:: 12 hl-color:: yellow id:: 654d1019-abd0-47e7-9f81-34a8f1a8b786
• duplicate clusters of issues, as 100% duplicate issues are similar issues, too ls-type:: annotation hl-page:: 12 hl-color:: red id:: 654d1026-9678-4b52-b875-57d0c7e594a0 hl-stamp:: 1699549224448
• ese clusters are diverse to ensure free-bias evaluation ls-type:: annotation hl-page:: 13 hl-color:: yellow id:: 65537124-859a-4460-991f-5652b10380e5
• RQ1: To what extent are the identified issue clusters correct? ls-type:: annotation hl-page:: 13 hl-color:: green id:: 655373bd-50f7-4cb1-8767-9cbfa678ca5e
• to what extent the member issues of a cluster are related to each other (similar). ls-type:: annotation hl-page:: 13 hl-color:: green id:: 65537486-2805-4b02-896d-869221fe3882
• RQ2: To what extent are the identified issue clusters complete? ls-type:: annotation hl-page:: 14 hl-color:: green id:: 65537540-a5a6-4caa-ab48-c3c7664eb7fa
• AgglomerativeClustering ls-type:: annotation hl-page:: 15 hl-color:: green id:: 6553757f-f7a0-4aa0-90d7-0a0159cc3f28
• As a summary, the AHC algorithm can extract relevant similar issue clusters from GitHub repositories depending on only two issue features: title and description. ls-type:: annotation hl-page:: 20 hl-color:: green id:: 655375f7-0c9e-4efa-b858-da6f1bcf804f
• to return a number of relevant issue clusters equal to the number of ground-truth clusters ls-type:: annotation hl-page:: 20 hl-color:: green id:: 65537602-737f-487c-96dc-5148243c5ac2
• credability ls-type:: annotation hl-page:: 20 hl-color:: red id:: 65537619-ffb9-4fd6-aea0-a46bd19bb3b6
• Internal threat ls-type:: annotation hl-page:: 20 hl-color:: green id:: 6553764c-c160-4efc-a31b-dcbb5b54cc4b
• External threat ls-type:: annotation hl-page:: 20 hl-color:: green id:: 65537656-c324-42e1-bd1f-dcf886a36063
• n this paper, a machine learning-based approach is proposed to support the issue tracking system on GitHub via grouping similar submitted issues together ls-type:: annotation hl-page:: 21 hl-color:: green id:: 65537688-628f-476d-b11f-582180e09e81
• clustering process in this research work is guided by two important features extracted from the issue’s report. These features are the title and body of each issue ls-type:: annotation hl-page:: 8 hl-color:: yellow id:: 6554f0d6-ae81-4f8c-bfab-a7041da86d96 hl-stamp:: 1700065570293
• Figure 4: An example of dendrogram tree with cutting line.9 ls-type:: annotation hl-page:: 11 hl-color:: yellow id:: 6554f1eb-38bf-447e-85dc-b41d6594246e
• without a prior assumption about the number of clusters ls-type:: annotation hl-page:: 12 hl-color:: green id:: 6554f25c-2197-46bf-a790-e29a265917d1
• In this research work, this cutting point is set after many tries to find255 reasonable similar issue clusters based on title and description matching ls-type:: annotation hl-page:: 12 hl-color:: green id:: 6554f27b-e237-409f-ae81-1966cf5cac44
• A cluster from the dataset that shares a maximum number of issues with the extracted cluster is the target and is called a ground-truth cluster ls-type:: annotation hl-page:: 14 hl-color:: green id:: 655528b8-2fd3-4cb7-a9eb-f989e4abdc91
• The higher Recall value you have, the higher completeness cluster is extracted, and vice-versa. ls-type:: annotation hl-page:: 14 hl-color:: blue id:: 655528ec-ada3-400c-80b1-503e711d3a1a
• For each extracted cluster (EXC), ls-type:: annotation hl-page:: 14 hl-color:: green id:: 65552908-9bcb-4bc8-ab76-323e776f4928