nclandrei / l5-dissertation Goto Github PK

There are template source files for LaTeX on Moodle for the Level 5 Dissertation Project and Proposal.

1. Identify a selection of metrics, e.g.:

• Readability scores
• Presence of stack traces
• Presence of screenshots (or other image attachments)
• ?? See data quality papers for others
• Links to related documents

What dimensions of an issue to be measured?

E.g. comment thread history? Does the structure of the discussion make a difference?

Topic diversity? Is there a metric for telling how wide ranging a topic is?

Number of comments; does a large number of comments indicate hight or low quality?

Use of markdown format - does presentation make a difference (e.g. paragraph structure, use of bullets...?).

Send a flattering email, stating you found their paper and are interested in their prototype. CC Tim in the email. Also note you appreciate the software may be in a prototype form, but might potentially be interested in developing it further with them.

Relevant papers are the Secret Life of Bugs and this one

@Article{Korkala2014WasteIdentification,
author = {Mikko Korkala and Frank Maurer},
title = {Waste identification as the means for improving communication in globally distributed agile software
development },
journal = {The Journal of Systems and Software},
year = 2014,
volume = 95,
pages = {122--140}}

A qualitative study looking at the causes of communication waste in software projects (different types of waste, which implies additional unnecessary communication costs, e.g. repetition, inaccurate).

That some good insight into potentially measuring cost of poor quality?

Maybe focus on high priority tickets, since duration and cost are likely to have a closer relationship (or equally, look at tickets at the same criticality).

Different types of waste might have different impacts.

Then what are the quality attributes of a ticket in different contexts?

• Stack trace
• Textual description
• Language level
• Spelling/grammar bugs, particulalrly for proper nouns ?
• Screen shots
• Comments, discussion (equally what quality attributes cause discussion).

Other things that could be written future work: sentiment analysis. Apply these techniques to the comments applied to issues?

Reading:

• Sentiment analysis tools and how they work.
• Available data sets

Depends on Metrics identified.

What is meant by 'high or low issue quality?'

Something like "Sufficient ifnormation for the issue to be resolved?"

Overall, this version (e2fd831) looks good with some neat ideas. Some comments on style are in
mprop.pdf