Strength of Relationship Between Multi-labeled Data and Labels - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2015

Strength of Relationship Between Multi-labeled Data and Labels

Masahiro Kuzunishi
  • Fonction : Auteur
  • PersonId : 1001258
Tetsuya Furukawa
  • Fonction : Auteur
  • PersonId : 1001259

Résumé

Collected data must be organized properly to utilize well and classification of data is one of the efficient methods. Individual data or an object is classified to categories and annotated with labels of those categories. Giving ranks to labels of objects in order to express how close objects are to the categories enables us to use objects more precisely. When target objects are identified by a set of labels $\mathcal{L}$, there are various strength of relationship between objects and $\mathcal{L}$. This paper proposes criteria for objects with two rank labels, primary and secondary labels, such as a label relates to $\mathcal{L}$, a primary label relates to $\mathcal{L}$, every primary label relates to $\mathcal{L}$, and every label relates to $\mathcal{L}$. The strongest criterion which an object satisfies is the level of the object to express the degree of the strength of relationship between the object and $\mathcal{L}$. The results for two rank objects are extended to k rank objects.
Fichier principal
Vignette du fichier
978-3-319-24315-3_10_Chapter.pdf (89.85 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01466209 , version 1 (13-02-2017)

Licence

Paternité

Identifiants

Citer

Masahiro Kuzunishi, Tetsuya Furukawa. Strength of Relationship Between Multi-labeled Data and Labels. 3rd International Conference on Information and Communication Technology-EurAsia (ICT-EURASIA) and 9th International Conference on Research and Practical Issues of Enterprise Information Systems (CONFENIS), Oct 2015, Daejon, South Korea. pp.99-108, ⟨10.1007/978-3-319-24315-3_10⟩. ⟨hal-01466209⟩
44 Consultations
97 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More