logseq/pages/icse2025-paper149.md

type:: [[REVIEWS]]
tags:: 
year:: 2024
venue:: [[ICSE]] 
full-title:: Mining and Assessing Issue Resolution Processes In Open Source Software Repositories
date-start:: [[16-04-2024]] - 13:33
date-submitted:: [[03-06-2024]] 
external-links::
status:: [[DONE]] 
deadline-submission:: 
file:: [[@icse2025-paper149.pdf]] 
parent::
todoist:: https://app.todoist.com/app/task/149-mining-and-assessing-issue-resolution-processes-in-open-source-software-repo-7858696680

	- ### [[Highlights]]
	  collapsed:: true
		-
		- # Annotazioni
			- (3/6/2024, 14:17:58)
			- - “Resolving issues of various kinds is one of thecentral activities of open source software (OSS) developers.” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=2A7ZTKBS)) #5fb236
			- - “ac” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=4LFEB94K)) #ff6666
			- - “organization of issues,” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=YZBGY45C)) #5fb236
			- - “operationalized by the uncertainties involved in their processes” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=R8W4AC4N)) #ffd400
			- - “predictor of the issues’ processing efficiency in terms of their overall lifetime and transition time between consecutive steps” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=E3Y3LY5M)) #5fb236
			- - “dis-organization, inside the issues’ processes.” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=S9JYGM7Q)) #ffd400
			- *TO BE UNDERSTOOD BETTER! What does it mean?*
			- - “entropy of the DFGs’ transitions” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=FMJF26KC)) #a28ae5
			- - “rocess models” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=98PSQWMZ)) #ffd400
			- *How to cluster processes?*
			- - “higher level of process uncertainty is significantly associated with an increased issue lifetime.” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=P3PFV968)) #5fb236
			- - “The results suggest a potential way to improve issue resolution efficiency is to better organize (i.e. be more certain about) their processes, starting locally with each step, and eventually the overall processes” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=ZEC584U5)) #ffd400
			- *What does it mean?*
			- - “processing issues introduces a heavy workload to project maintainers, which calls for better understanding and management of issues and their workflows” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=FDKZ72AC)) #a28ae5
			- - “There are also evidences showing that having an organized, coordinated task resolution process is positively associated with OSS projects’ survivability and team performance” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=MVBSENRE)) #5fb236
			- - “open collaborations around issues are often carried out in a highly ad hoc manner without a predefined process, making it difficult to gain insights about the internal processes of complex issues, or making assessments about how well the issues are organized.” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=KHS47QGU)) #a28ae5
			- - “entropy” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=QP5XYHVJ)) #5fb236
			- - “a process with higher entropy is more complicated and involve more uncertainties, which indicates a lower level of organization in our context” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=7VWP2W4U)) #a28ae5
			- - “direct follow graphs (DFGs)” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=UYN5TQ7S)) #5fb236
			- - “we adopt DFGs to provide a unified representation of issues during process mining as introduced next.” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=PW3C3QXE)) #5fb236
			- - “entropy-based measure, we aim at quantifying the overall organization of issue resolution processes in an OSS project.” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=F3P6SRMF)) #e56eee
			- - “‘representative’, we aim at discovering concise models with each model representing a group of issues with similar processes.” ([“icse2025-paper149.pdf”, p. 1](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=1&annotation=KBK5Q95X)) #a28ae5
			- - “to group issues with similar processes to find the representative models” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=EQPN3HPL)) #ffd400
			- *I'm curious to see how processes are clustered!*
			- - “augment” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=XB8B4G53)) #a28ae5
			- - “effectiveness” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=NA88IES5)) #5fb236
			- - “usefulness” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=ECFK9RFZ)) #5fb236
			- - “representative yet simple processes from issues” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=GF3F76HG)) #ffd400
			- *What kind of processes and representative with respect to what?*
			- - “correlation” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=A2CBCWX2)) #5fb236
			- - “process entropy” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=P28LIEMY)) #5fb236
			- - “issue lifetime” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=JE73B2BX)) #5fb236
			- - “issue lifetime,” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=C5EGID4Q)) #a28ae5
			- - “extract” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=XTCPX2IJ)) #ff6666
			- *extracted?*
			- - “its following up transition time” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=WTYMF6GF)) #ffd400
			- *Similar to the previous research question, isn't it?*
			- - “follow transition time” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=BD96YTEZ)) #a28ae5
			- - “entropy-based metric to quantify the organization (uncertainty) of issue resolution processes in OSS projects.” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=2LIJ2LSN)) #a28ae5
			- - “mine representative yet simple models for issue resolution processes from event logs” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=XAECXEDW)) #a28ae5
			- - “level” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=8T6QAHZX)) #ff6666
			- *level of?*
			- - “Markov models are shown to be useful in generating probabilistic representations of the event sequences” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=D3MDSWI9)) #5fb236
			- - “DFGs as simple yet informative representations for both issue event sequences, and corresponding mined processes.” ([“icse2025-paper149.pdf”, p. 2](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=2&annotation=PK4RMA67)) #a28ae5
			- - “A recent survey” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=26GMQS3K)) #ffd400
			- *It's not a so recent work!*
			- - “In this work, we augment the KMeans++ algorithm for trace clustering when mining processes from the complex issue event logs.” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=7NV38TSU)) #a28ae5
			- - “there is a large body of literature that performs process mining in software repositories” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=6JNAE6ZZ)) #a28ae5
			- - “DFG-based issue resolution processes from event logs” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=GXF73FP3)) #5fb236
			- - “correlation between entropy and OSS issue resolution process, demonstrating the usefulness of the proposed process mining and entropy-based metric in understanding real-world collective behaviour in OSS repositories.” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=JG7Q5S3A)) #5fb236
			- - “There is also study aiming at issue characterization and classification for related issue recommendation” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=CCLHJWUX)) #5fb236
			- - “With the mined processes for resolving issues in OSS repositories, in this work, we propose to measure the level of organization of the issue processes with an information entropy based metric, and study its correlation with issues’ lifetime.” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=E55TKVZM)) #e56eee
			- - “This section presents the proposed approach to mine issue resolution processes.” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=X6AL24US)) #5fb236
			- - “fetch issue timeline” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=U3DZ59L6)) #5fb236
			- - “records users’ activities in issues” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=TKJKB43N)) #5fb236
			- - “key contents, including the actors’ roles, event types, action time and actor name from the raw data, and encode the issues into event sequences.” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=Q4RH4MXM)) #ffd400
			- *What are event types? And what about actors' roles?*
			- - “During the clustering and mining phase, we calculate the overall entropy of the mind process models as shown on the right half of Figure 1(c)” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=LULMWFCI)) #ff6666
			- *An explanatory example would be great here! Moreover, what is the mind process model? Ah ok, it's mined not mind!*
			- - “following Sec. IV” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=Z7CU9Z8W)) #ff6666
			- - “We perform process clustering and mining, and process model measuring for different clustering parameters, and then choose the optimal parameters.” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=SWS2MRMP)) #ffd400
			- *It's not clear what you want to do. It's necessary an explanatory/motivating example.*
			- - “count of occurrences of the corresponding direct follow relationship” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=R6TBU9IC)) #5fb236
			- - “computational efficient way to represent the processes” ([“icse2025-paper149.pdf”, p. 3](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=3&annotation=2XI68E4S)) #5fb236
			- - “A lower (higher) source entropy corresponds to less (more) representative sequences which means a higher model certainty (uncertainty)” ([“icse2025-paper149.pdf”, p. 4](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=4&annotation=X3AMTWNM)) #a28ae5
			- - “we temporally segment a project’s issues into a series of snapshots with a non-overlapping sliding window of one month.” ([“icse2025-paper149.pdf”, p. 4](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=4&annotation=5GNNULGL)) #a28ae5
			- - “Extract Issue Event Sequence from Raw Data” ([“icse2025-paper149.pdf”, p. 4](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=4&annotation=8LE56849)) #5fb236
			- - “we process raw data and turn it into event sequence.” ([“icse2025-paper149.pdf”, p. 4](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=4&annotation=R72DZ69U)) #5fb236
			- - “SubscribedEvent and MentionedEvent always show together in consecutive order, which is by ‘@’ someone in the comment” ([“icse2025-paper149.pdf”, p. 4](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=4&annotation=2IJNUFZI)) #ffd400
			- *???*
			- - “For the above case, both Label and Unlabel are marked as the Label event.” ([“icse2025-paper149.pdf”, p. 4](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=4&annotation=CA9FWMPH)) #ffd400
			- *???*
			- - “We define parallel events as events that take place at the same time, and are carried out by the same person” ([“icse2025-paper149.pdf”, p. 4](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=4&annotation=3H34XLDC)) #ffd400
			- *Examples? The same person that does different actions at the same time? This is not clear.*
			- - “We use direct follow probability matrices instead of more complex models like Petri-Nets [17] because we can directly use them as feature matrices for clustering, and calculate the entropy-based metrics” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=MHKPUV84)) #a28ae5
			- - “Build a DFG-Based Model for a Single Issue:” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=CRLMHNAK)) #ffd400
			- *It's not clear how this works in practice. An issue consists of a sequence of events, and it is not clear how the probability concept comes to the place here.*
			- - “issues that involve the same sets of events, but different in event orders including cases of parallel events.” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=9XDK24IC)) #a28ae5
			- - “It is straightforward to extend our approach to build a model for a group of issues by extending the count()” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=VXAF9GGI)) #5fb236
			- - “groups of issues” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=Q84ARQ86)) #ffd400
			- - “Sec III-C” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=IVU4FWIY)) #ffd400
			- *We are in Sec III-C. This reference is not needed.*
			- - “Manhattan distance is used to calculate distance between data points in order to measure the difference between process models” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=EJC8FRT7)) #5fb236
			- - “d” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=NVH3AQ7Z)) #5fb236
			- - “We adopt the “trace clustering and mining” paradigm [22] to discover the underlying process models of issue event sequences extracted from raw issue timeline data in OSS repositories.” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=GV4G4ULY)) #5fb236
			- - “have” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=DL5XFL3H)) #ff6666
			- *has?*
			- - “Given the set of data points D to be clustered” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=GV4LYTIS)) #ffd400
			- *It is necessary to explain what these datapoints represent in the case at hand (i.e., management of issue process models). In other words it is necessary to describe clearly the process that comes from the original issues, and encoding them in a way, which is amenable for clusterization purposes.*
			- - “finding representative and simple process models.” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=YF467W5E)) #5fb236
			- - “In our proposed algorithm, we combine the Silhouette score and information gain score as the measurement for optimal cluster number evaluation.” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=I5P8YSPF)) #a28ae5
			- - “goodness of clustering by taking both within and between cluster distances into account.” ([“icse2025-paper149.pdf”, p. 5](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=5&annotation=2TUKPF6M)) #a28ae5
			- - “N is the total number of issues.” ([“icse2025-paper149.pdf”, p. 6](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=6&annotation=A422VFUP)) #a28ae5
			- - “When Sil = 1, it means the clusters are dense and well separated.” ([“icse2025-paper149.pdf”, p. 6](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=6&annotation=7CDKKPNV)) #a28ae5
			- - “Sil = 0, it means the clusters are not clearly separated [28].” ([“icse2025-paper149.pdf”, p. 6](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=6&annotation=BY4JKZYT)) #a28ae5
			- - “selecting the one with the highest Sil value.” ([“icse2025-paper149.pdf”, p. 6](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=6&annotation=LTP5V5GT)) #a28ae5
			- - “representative yet simple processes” ([“icse2025-paper149.pdf”, p. 6](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=6&annotation=FCR9APZL)) #ffd400
			- *Apart from the definition and usage of different metrics, which underpin the definition of issue clusters, as a reader I'm missing a presentation and discussion of explanatory example showing real issues and corresponding clusters and what's the benefit of having issues that are clustered as proposed by the authors.*
			- - “the local entropy measures the degree of uncertainty with respect to each event’s transition in the mined model.” ([“icse2025-paper149.pdf”, p. 6](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=6&annotation=9SC3P4FX)) #5fb236
			- - “The overall entropy also increases with the increasing frequencies of events whose entropy is higher than the overall entropy” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=E3D6PYFF)) #5fb236
			- - “A higher overall entropy indicates a more complicated process model, meaning that there are more uncertainties present in the model.” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=9DDMB4D9)) #ffd400
			- *Authors assume a correlation of model complexity vs model uncertainties. I can agree with this but the paper does not provide any evidence in terms of explanatory examples.*
			- - “The measured overall process entropy is linked to the organization of an OSS community in issue processing.” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=J6D23VGA)) #ffd400
			- *How can we support statements like this one?*
			- - “organization of a project is associated with its sustainability” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=J2ADZ7TB)) #5fb236
			- - “information entropy to measure the process of issue resolution by taking into account not only the creators and solvers [14], but also the actors’ roles and events, for every step.” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=3D7D23GG)) #a28ae5
			- - “we argue that the proposed overall process entropy also quantifies the level of organization for issue resolution,” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=FHU2H6M8)) #ffd400
			- *This is what you argued. Let's see if it is evaluated.*
			- - “does higher local uncertainty links to longer time consumed for events in a process?” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=5YE9RH6S)) #a28ae5
			- - “2022 APRIL” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=9TIAV7LB)) #ffd400
			- *As I said, this is a resubmission of a paper submitted long time ago somewhere else.*
			- - “Table V, Sec. V-D” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=JWUFIGCA)) #ff6666
			- *I would avoid references to sections/elements which are not presented yet.-*
			- - “reduced” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=KE7GM52W)) #ff6666
			- *reduce*
			- - “proposed entropy metrics about mined issue processes and the overall issue lifetime, as well as the events’ local transition time, respectively, with respect to the three research questions in the Introduction.” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=677XGIQ6)) #5fb236
			- - “abnormal issue recor” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=5X2SA5TP)) #ffd400
			- *what do you mean? Is there any thresholds used and how/why you defined it?*
			- - “17 projects are selected.” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=GVL7VCTT)) #5fb236
			- - “exclude extreme outliers, issues with issue length or issue lifetime beyond the upper outer fences are removed” ([“icse2025-paper149.pdf”, p. 7](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=7&annotation=GMNSB8UH)) #ff6666
			- *Please rewrite.*
			- - “dependent variables” ([“icse2025-paper149.pdf”, p. 8](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=8&annotation=F9RSU5XD)) #5fb236
			- - “control variables” ([“icse2025-paper149.pdf”, p. 8](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=8&annotation=VKHQMWAU)) #5fb236
			- - “representative yet simple processes” ([“icse2025-paper149.pdf”, p. 8](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=8&annotation=M7JXIWW6)) #ffd400
			- *Why are you targeting them? Simple and representative with respect to what? From which perspective?*
			- - “KMeans++ and the proposed clustering algorithm, do not show significant differences (marked as =).” ([“icse2025-paper149.pdf”, p. 8](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=8&annotation=FTZA4AQ5)) #5fb236
			- - “The results suggest that, under all settings of α, the proposed clustering algorithm always finds issue clusters that correspond to significantly less entropy of the mined processes than the original KMeans++ algorithm with more clusters” ([“icse2025-paper149.pdf”, p. 8](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=8&annotation=Q5S7UREB)) #5fb236
			- - “is effective in helping the mining algorithm to find representative and simple process models that better describe the issues than the traditional, KMeans++ algorithm” ([“icse2025-paper149.pdf”, p. 8](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=8&annotation=799JE2TP)) #ffd400
			- *This is not evident at all!*
			- - “RQ2 as: Higher level of process uncertainty measured by Hprocess is associated with an increased issue lifetime.” ([“icse2025-paper149.pdf”, p. 9](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=9&annotation=NTKEC27J)) #a28ae5
			- - “Hprocess through better organized processes with less uncertainty” ([“icse2025-paper149.pdf”, p. 9](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=9&annotation=LTVNFX5U)) #a28ae5
			- - “we breakdown the overall process entropy into local entropy that measures the uncertainty associated with each event, and studies the correlation between such local entropy and the time spent on each event’s transitions to the direct following, next events.” ([“icse2025-paper149.pdf”, p. 9](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=9&annotation=WFA3NI6P)) #ffd400
			- *I'm surprised that the whole study is done by considering the process model, without considering at all issue descriptions, comments, etc. Moreover, it is not clear why and how it is possible to rule out the complexity behind an issue that can be related to some requested functionalities or difficulties related to the resolution of some bug in the source code.*
			- - “the local uncertainty measured by the proposed local event transition entropy (Hevent) is also positively associated with the events’ transition time during issue processing.” ([“icse2025-paper149.pdf”, p. 10](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=10&annotation=ZKSW4USJ)) #5fb236
			- - “The findings of this paper may not generalise well to other projects.” ([“icse2025-paper149.pdf”, p. 10](zotero://select/library/items/R3KQWKHC)) ([pdf](zotero://open-pdf/library/items/T6XVV34C?page=10&annotation=Q3MMB5HK)) #ffd400
			- *That's the point!*
	- ### [[Comments]]
		- #.tabular
			- Paper summary
				- The paper presents an approach to enhance the understanding of issue resolution processes in OSS projects. It leverages process mining techniques to model these processes and uses an entropy-based metric to quantify their uncertainty. The study utilizes Direct Follow Graphs (DFGs) to represent issue processes and proposes an augmented KMeans++ clustering algorithm to identify representative and simple process models from event logs. By analyzing 17 open-source projects, authors have investigated the correlation between overall process entropy and OSS issue lifetime, as well as the correlation between event local entropy and the time needed to move from one state to the next.
			- Strengths
				- + Interesting approach
				- + Relevant problem
				- + Usage of entropy-based metrics to understand issue resolution processes
			- Weaknesses
				- - The paper lacks detailed examples to illustrate how the proposed models and metrics work in practice, which would help in understanding the practical application of the research.
				- - The study does not consider other potential variables that could impact issue resolution, such as issue descriptions and comments.
				- - The findings are based on a relatively small dataset from GitHub, which may limit their generalizability to other OSS platforms or projects.
			- Detailed comments for authors
				- Novelty: The usage of entropy-based metrics in the considered domain is innovative and provides a quantitative measure for issue resolution processes. However, the novelty would be better highlighted with more real-world examples.
				- Rigor: The methodology is rigorous, but there are sections, particularly around the clustering algorithm and the use of entropy, that could benefit from more detailed explanations and justifications.
				- Relevance: The paper addresses a relevant problem related to the management of OSS projects.
				- Verifiability & Transparency: Authors provide a replication package with the source code and data used for the presented experiments. However, the information in the README file is limited, and there are no instructions on executing the different scripts. Readers are supposed to go through the various files and figure out how to run them on their own.
				- Presentation: The overall presentation is clear, but there are several areas where the paper could be improved for better readability. In particular, it is necessary to include explanatory examples to demonstrate the practical application of the proposed definitions and methods. Key terms and concepts should be explained with fragments of concrete processes.
				- Detailed comments:
					- Concerning the creation of DFG-based models, it's not clear how it works in practice. In practice, an issue consists of a sequence of events (e.g., comments, answers, etc.), it is unclear how the concept of probability is applied to these sequences.
					- Concerning the clustering process, it is necessary to explain what datapoints represent in the case at hand (i.e., management of issue process models). In other words, it is essential to clearly describe the process that takes the original issues and encodes them in a way that is amenable for clusterization purposes.
					- Apart from the definition and usage of different metrics, which underpin the definition of issue clusters, as a reader, I'm missing a presentation and discussion of explanatory examples showing real issues and corresponding clusters and discussing the benefits of having issues that are clustered as proposed by the authors.
					- The authors assume a correlation between model complexity and model uncertainties. While this assumption is plausible, the paper does not provide any evidence or explanatory examples to support this correlation.
					- The focus on finding representative yet simple processes from issues needs more justification. The paper should clarify why these processes are targeted, what "simple and representative" means in this context, and from which perspective these criteria are considered.
					- The study is focused on the process model without considering issue descriptions, comments, and other contextual information. It is unclear how the complexity behind an issue, which could be related to requested functionalities or difficulties in resolving bugs in the source code, is accounted for or ruled out.
					- As also mentioned by the authors, "The findings of this paper may not generalise well to other projects.". This is a critical point of the paper as it affects the applicability of the research findings to a broader range of OSS projects.
	- ### [[REVIEWS/Notes]]
		- In the context of Markov chains, the expression 𝑥log⁡𝑥xlogx often appears when discussing concepts such as entropy and the stationary distribution. Here's a detailed look at where and how this function is relevant:
		- ### Markov Chain Basics
			- A Markov chain is a stochastic process that undergoes transitions from one state to another in a state space. It is characterized by:
				- 1.  **States**: The different possible conditions or positions the system can be in.
				- 2.  **Transition Matrix 𝑃**: A matrix where the entry 𝑃𝑖𝑗 represents the probability of transitioning from state 𝑖 to state 𝑗 in one step.
		- ### Entropy of a Markov Chain
			- Entropy is a measure of uncertainty or randomness. For a Markov chain, entropy can help quantify the unpredictability of the chain's state at any given time. Let's delve into this with more detail:
			- #### Stationary Distribution
				- The stationary distribution 𝜋 is a probability distribution over the states that remains unchanged as the system evolves. In other words, if the Markov chain starts in the stationary distribution, it stays in that distribution after each transition.
				- For a Markov chain with a state space 𝑆 and a transition matrix 𝑃, the stationary distribution 𝜋 is a probability distribution over the states in 𝑆 that satisfies the equation:
					- 𝜋𝑃=𝜋
				- This means that if the Markov chain is in the stationary distribution, the distribution of states remains unchanged after transitions.
				- ### The Equation 𝜋𝑃=𝜋πP=π
					- The equation 𝜋𝑃=𝜋 expresses the property of the stationary distribution. Let's break down what this means:
					  collapsed:: true
						- 𝜋π: This is a row vector (𝜋1𝜋2⋯𝜋𝑛)(π1​​π2​​⋯​πn​​), where 𝜋𝑖πi​ represents the stationary probability of being in state 𝑖i.
						- 𝑃P: This is the transition matrix of the Markov chain, where 𝑃𝑖𝑗Pij​ is the probability of transitioning from state 𝑖i to state 𝑗j.
					- The equation 𝜋𝑃=𝜋 means that multiplying the stationary distribution vector 𝜋  by the transition matrix 𝑃 results in the same stationary distribution vector 𝜋. Mathematically, this can be expanded as:
					- \[
					  \begin{pmatrix}
					  P_{11} & P_{12} & \cdots & P_{1n} \\ P_{21} & P_{22} & \cdots & P_{2n} \\ \vdots & \vdots & \ddots & \vdots \\ P_{n1} & P_{n2} & \cdots & P_{nn} \end{pmatrix} = 
					  \begin{pmatrix} 
					  \pi_1 & \pi_2 & \cdots & \pi_n 
					  \end{pmatrix} 
					  \]
					- This means that for each state \(j\): \[ \sum_{i} \pi_i P_{ij} = \pi_j \] This equation indicates that the total probability flow into state \(j\) (the left-hand side) equals the stationary probability of being in state \(j\) (the right-hand side).
					- ### Significance
					  collapsed:: true
						- **Equilibrium State**: The stationary distribution represents an equilibrium state where the probabilities of being in each state remain constant over time.
						- **Long-Term Behavior**: For an irreducible and aperiodic Markov chain, starting from any initial distribution, the state distribution converges to the stationary distribution as the number of transitions goes to infinity.
					- ### Example Let's consider a Markov chain with two states, \(A\) and \(B\), and the transition matrix \(P\): \[ P = \begin{pmatrix} 0.8 & 0.2 \\ 0.4 & 0.6 \end{pmatrix} \] Suppose the stationary distribution is \(\pi = \begin{pmatrix} \frac{2}{3} & \frac{1}{3} \end{pmatrix}\). To verify that \(\pi\) is indeed the stationary distribution, we check: \[ \pi P = \begin{pmatrix} \frac{2}{3} & \frac{1}{3} \end{pmatrix} \begin{pmatrix} 0.8 & 0.2 \\ 0.4 & 0.6 \end{pmatrix} = \begin{pmatrix} \left(\frac{2}{3} \cdot 0.8 + \frac{1}{3} \cdot 0.4\right) & \left(\frac{2}{3} \cdot 0.2 + \frac{1}{3} \cdot 0.6\right) \end{pmatrix} \] \[ = \begin{pmatrix} \frac{16}{30} & \frac{14}{30} \end{pmatrix} = \begin{pmatrix} \frac{2}{3} & \frac{1}{3} \end{pmatrix} = \pi \] Thus, the distribution \(\pi = \begin{pmatrix} \frac{2}{3} & \frac{1}{3} \end{pmatrix}\) is indeed a stationary distribution because it satisfies \(\pi P = \pi\). In summary, \(\pi P = \pi\) characterizes the stationary distribution as a distribution that remains unchanged under the transition probabilities of the Markov chain, representing the long-term equilibrium state of the system.
			- #### Entropy Rate
				- The entropy rate of a stationary Markov chain quantifies the average amount of uncertainty (or information) produced per time step. It is given by:
					- 𝐻(𝜋,𝑃)=−∑𝑖∈𝑆𝜋(𝑖)∑𝑗∈𝑆𝑃𝑖𝑗log⁡𝑃𝑖𝑗
				- Here, 𝜋(𝑖) is the stationary probability of being in state 𝑖, and 𝑃𝑖𝑗​ is the transition probability from state 𝑖 to state 𝑗. This formula incorporates the 𝑥log⁡𝑥 term through the transition probabilities 𝑃𝑖𝑗log⁡𝑃𝑖𝑗.
			- #### Interpretation of 𝑥log⁡𝑥xlogx
				- In this context, 𝑥log⁡𝑥xlogx (where 𝑥x represents the transition probability 𝑃𝑖𝑗Pij​) helps measure the contribution to the entropy rate from moving from one state to another. High transition probabilities (close to 1) contribute less to uncertainty than lower probabilities, which is why the log⁡log function (which is negative for probabilities between 0 and 1) is used.
		- ### Example Calculation
			- Consider a simple Markov chain with two states, \(A\) and \(B\), and the following transition matrix \(P\):
			  \[
			  P = \begin{pmatrix}
			  P_{AA} & P_{AB} \\
			  P_{BA} & P_{BB}
			  \end{pmatrix} = \begin{pmatrix}
			  0.8 & 0.2 \\
			  0.4 & 0.6
			  \end{pmatrix}
			  \]
			- Assume the stationary distribution \(\pi\) is given by:
			  \[
			  \pi = \begin{pmatrix}
			  \pi_A & \pi_B
			  \end{pmatrix} = \begin{pmatrix}
			  \frac{2}{3} & \frac{1}{3}
			  \end{pmatrix}
			  \]
			- The entropy rate \(H(\pi, P)\) is calculated as follows:
			  \[
			  H(\pi, P) = - \sum_{i \in \{A, B\}} \pi_i \sum_{j \in \{A, B\}} P_{ij} \log P_{ij}
			  \]
			- Substituting the values, we get:
			  \[
			  H(\pi, P) = - \left( \pi_A \sum_{j \in \{A, B\}} P_{Aj} \log P_{Aj} + \pi_B \sum_{j \in \{A, B\}} P_{Bj} \log P_{Bj} \right)
			  \]
			  \[
			  H(\pi, P) = - \left( \frac{2}{3} \left(0.8 \log 0.8 + 0.2 \log 0.2\right) + \frac{1}{3} \left(0.4 \log 0.4 + 0.6 \log 0.6\right) \right)
			  \]
			- Calculate each term:
			  \[
			  0.8 \log 0.8 \approx 0.8 \cdot (-0.09691) = -0.07753
			  \]
			  \[
			  0.2 \log 0.2 \approx 0.2 \cdot (-0.69897) = -0.13979
			  \]
			  \[
			  0.4 \log 0.4 \approx 0.4 \cdot (-0.39794) = -0.15918
			  \]
			  \[
			  0.6 \log 0.6 \approx 0.6 \cdot (-0.22185) = -0.13311
			  \]
			- Now substitute these back into the equation:
			  \[
			  H(\pi, P) = - \left( \frac{2}{3} \left(-0.07753 - 0.13979\right) + \frac{1}{3} \left(-0.15918 - 0.13311\right) \right)
			  \]
			  \[
			  H(\pi, P) = - \left( \frac{2}{3} \left(-0.21732\right) + \frac{1}{3} \left(-0.29229\right) \right)
			  \]
			  \[
			  H(\pi, P) = - \left( -0.14488 - 0.09743 \right)
			  \]
			  \[
			  H(\pi, P) = 0.24231
			  \]
			- Therefore, the entropy rate of this Markov chain, reflecting the average uncertainty per step, is approximately 0.24231 bits.
			- ```mermaid
			  graph TD;
			      A[State A] -- 0.8 --> A
			      A -- 0.2 --> B
			      B[State B] -- 0.4 --> A
			      B -- 0.6 --> B
			  ```
	- ### ❓️Questions
	  collapsed:: true
		- {{query (and [[question]] [[icse2025-paper149]] )[[question]]}}
		  query-table:: true
		  query-properties:: [:block]
-