From Maisqual Private Wiki
The overall process overview for analysing so-called experience projects and giving the right advice to the user may be described as follows:
- Local projects reflect the local enterprise conventions and usages.
- Domain-specific projects provide external experience on general practices relative to a domain (real-time, embedded, desktop, etc.).
- The quality assessment standards provide different views on the quality, thus improving the measure pertinence.
Practices are entered in the engine as:
- Known good practices from methods (Clean, Agile, Scrum, etc.).
- Practices identified in the experience of analysed projects.
- Practices manually entered in the engine by SQuORING people.
The result is something between a quality model and a method, built from the experience of SQuORING people, analysed projects and known methods.
- Practices are actions:
- Improve test coverage,
- Add more comments,
- Refactor module A,
- Setup Requirements inspection,
- Setup Code inspection, etc.
- Objectives set target to reach for a practice. Not all practices have objectives. Examples are:
- Reach 50% Test Coverage
- Reach 20% Comments Ratio
 Decision Engine
The decision engine mixes the influence of local projects over general practices. It takes as inputs:
- The Maisqual quality models built from several experiences and contexts.
- Local projects experience.
- Domain-specific projects experience.
- Established practices (Agile, Clean, CMMi, etc.).
- The current project's context and history.
The user chooses to improve specifically a given characteristic of quality, and the engine provides the best practices to get there.
The decision engine uses recommender systems to propose an optimised set of actions to the user. An hybrid approach, including both collaborative filtering and content-based recommendations, should be used to:
- Fulfil missing attributes for a better confidence in data.
- Propose actions extracted both from the experience of other projects/people and company's culture.
- ↑ The local projects analysis might be seen as the user personal history and context -- in that case, the user from the data mining approach is the company.