Anti-patterns are poor solution to design and im-plementation problems. Developers may introduce anti-patterns in their software systems because of time pressure, lack of understanding, communication and or-skills. Anti-patterns create problems in software maintenance and development. Database anti-patterns lead to complex and time consuming query process-ing and loss of integrity constraints. Detecting anti-patterns could reduce costs, efforts and resources. Researchers have proposed approaches to detect anti-patterns in software development. But not much research has been done about database anti-patterns. This report presents two approaches to detect schema design anti-patterns in relational database. Our first approach is based on pattern matchingwe look into potential candidates based on schema patterns. Second approach is a machine learning based approach we generate features of possible anti-patterns and build SVMbased classifier to detect them. Here we look into these four anti-patterns a) Multi-valued attribute b) Nave tree based c) Entity Attribute Value and d)Polymorphic Association . We measure precision and recall of each approach and compare the results. SVM-based approach provides more precision and recall with more training dataset.
展开▼
