Learning and Using Structures for Constraint Acquisition
Type de document | Site actuel | Cote | Statut | Date de retour prévue | Code à barres | Réservations |
---|---|---|---|---|---|---|
Thèse universitaire | La bibliothèque des sciences de l'ingénieur | TH-005.116 DAO (Parcourir l'étagère) | Disponible | 0000000027908 |
PH.D Université Mohammed V 2016
Most of the existing constraint acquisition systems interact with the user by asking her to classify an example as positive or negative. Such queries do not use the structure of the problem and can thus lead the user to answer a large number of queries. In this thesis, we show that using the structure of the problem may improve the acquisition process a lot. To this end, we introduce two new concept of queries that use the structure. The first one, called generalization query, based on an aggregation of variables into types. The second one, named recommendation query, based on the prediction of missing constraints in the current constraint graph. In addition, we propose several algorithms and strategies to deal with these new kind of queries. We incorporate all our proposed algorithms into QUACQ constraint acquisition system, leading to three boosted versions, namely G-QUACQ, M-QUACQ, and P-QUACQ. The results show that the extended versions improve drastically the basic QUACQ. fr_
Il n'y a pas de commentaire pour ce document.