From Maisqual Private Wiki

Jump to: navigation, search

This is the April 2012 checkpoint for Maisqual, following the 2-days meeting we had with Philipppe Preux on April, 12-13th.


[edit] Overall Direction

The overall objective of this work is to correlate software development practices with attributes of quality.

This can be divided into the following distinct steps:

  • Practices identification on the history of the project.
  • Indicators gathering. Indicators are then classified into attributes of quality.
  • Application of data mining algorithms (ACPs, time series) on these to get correlated measures.

I could discuss with Martin Monperrus about software engineering, quality, and metrics.

[edit] Practices identification

Michaela Keller gave us some good starting points for text mining on mailing lists.

The following steps are needed to get statistic data on text (any text: bug comments, mailing lists, forums.

Preprocessing of text:

    • Set all text lower case.
    • Remove ponctuation.
    • [en.wikipedia.org/wiki/Stemming Stemming] with the Porter algorithm.
    • [en.wikipedia.org/wiki/Lemmatisation lemmatisation].
    • Remove Stoplist word.

From there we have a dictionnary of words: <w1, .., wn>.

More resources have been put on a dedicated page on linguistics.

[edit] Maisqual Quality Model

We have to decide on a quality model that will be used to rate product and process quality.

Product quality models are plenty. We choose to rely on ISO 9126 and SQuaRE evaluation models, with some improvements extracted from research papers and other models.

Process quality models are scarce. There are major initiatives, such as CMMi or ISO SPICE 15504, that address the organisational level of process quality/maturity. There is no standard about the process quality at the project level. We choose to borrow some of the most generic Key Process Area to classify process quality, with some improvements from research paper and other models.

[edit] Actions


  • Identify existing european projects
  • Identify and check existing software engineering data repositories (qualoss, ohloh, etc.).
  • Dive further into text mining.
  • Complete and implement the maisqual quality model.


[edit] References

  1. Number of documents where the word is found.
Personal tools