Although very effective in computing accurate recommendations, due to their inner nature, collaborative algorithms work very well with dense matrices but show their limits when they deal with sparse ones. In these cases, using only past ratings may lead to unsatisfactory results in the recommendation list. In this paper we show how to move from a user-item to a user-feature matrix by exploiting original user ratings. We then use matrix factorization techniques to compute recommendations.
Moving from item rating to features relevance in top-N recommendation / Anelli, Vito Walter; Di Noia, Tommaso; Di Sciascio, Eugenio; Lops, Pasquale; Trotta, Joseph. - ELETTRONICO. - 2140:(2018). (Intervento presentato al convegno 9th Italian Information Retrieval Workshop, IIR 2018 tenutosi a Roma, Italy nel May, 28-30, 2018).
Moving from item rating to features relevance in top-N recommendation
Vito Walter Anelli;Tommaso Di Noia;Eugenio Di Sciascio;
2018-01-01
Abstract
Although very effective in computing accurate recommendations, due to their inner nature, collaborative algorithms work very well with dense matrices but show their limits when they deal with sparse ones. In these cases, using only past ratings may lead to unsatisfactory results in the recommendation list. In this paper we show how to move from a user-item to a user-feature matrix by exploiting original user ratings. We then use matrix factorization techniques to compute recommendations.File | Dimensione | Formato | |
---|---|---|---|
paper17.pdf
accesso aperto
Tipologia:
Versione editoriale
Licenza:
Creative commons
Dimensione
410.39 kB
Formato
Adobe PDF
|
410.39 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.