|
Recommending design patterns using task-based conceptual features |
|---|---|
| รหัสดีโอไอ | |
| Creator | 1. Ekawit Nantajeewarawat 2. Nasith Laosen 3. Nuttapon Sanyawong |
| Title | Recommending design patterns using task-based conceptual features |
| Publisher | Maejo University |
| Publication Year | 2559 |
| Journal Title | Maejo International Journal of Science and Technology |
| Journal Vol. | 10 |
| Journal No. | 1 |
| Page no. | 113 |
| Keyword | design pattern,pattern recommendation,task-based conceptual feature |
| ISSN | 1905-7873 |
| Abstract | Task-based conceptual features (TCFs) represent human knowledge concerning intentions and/or characteristics of design tasks to which a design pattern is applicable. They provide a bridge connecting the usage of a design pattern and the characteristics of a design problem. A method for recommending appropriate design patterns based on TCFs is presented. From grammatical relations between words generated from the textual description of an input design problem, problem keywords are extracted. The obtained problem keywords are matched with clue words of each TCF in order to construct a feature vector representing the input problem. Based on the similarity between the feature vector representing the problem and the TCF-based vector representing each design pattern, design patterns are ranked and recommended. The method is evaluated on a collection of 24 input design problems. The evaluation results show that when the first-level hypernyms and hyponyms obtained from the WordNet ontology are employed for word matching and an appropriate penalty score is assigned to them, design patterns recommended in the top three ranks include the correct design patterns for all 24 problems. |
