Abstract | ||
---|---|---|
This paper describes and evaluates a method for computing artist similarity from a set of artist biographies. Theproposed method aims at leveraging semantic information present in these biographies, and can be divided in threemain steps, namely: (1) entity linking, i.e. detecting mentions to named entities in the text and linking them to anexternal knowledge base; (2) deriving a knowledge representation from these mentions in the form of a semanticgraph or a mapping to a vector-space model; and (3) computing semantic similarity between documents. Wetest this approach on a corpus of 188 artist biographies and a slightly larger dataset of 2,336 artists, both gatheredfrom Last.fm. The former is mapped to the MIREX Audio and Music Similarity evaluation dataset, so that its similarityjudgments can be used as ground truth. For the latter dataset we use the similarity between artists as providedby the Last.fm API. Our evaluation results show that an approach that computes similarity over a graph of entitiesand semantic categories clearly outperforms a baseline that exploits word co-occurrences and latent factors. |
Year | Venue | Field |
---|---|---|
2015 | ISMIR | Entity linking,Semantic similarity,Graph,Knowledge representation and reasoning,Information retrieval,Computer science,Semantic information,Exploit,Ground truth,Artificial intelligence,Knowledge base,Machine learning |
DocType | Citations | PageRank |
Conference | 6 | 0.45 |
References | Authors | |
24 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Sergio Oramas | 1 | 57 | 7.35 |
Mohamed Sordo | 2 | 152 | 11.63 |
Luis Espinosa Anke | 3 | 78 | 20.51 |
Xavier Serra | 4 | 1014 | 118.93 |