21.01.2014 Views

improving music mood classification using lyrics, audio and social tags

improving music mood classification using lyrics, audio and social tags

improving music mood classification using lyrics, audio and social tags

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

LIST OF TABLES<br />

Table 3.1 Mood categories derived from last.fm <strong>tags</strong> .................................................................. 36<br />

Table 4.1 Contingency table of binary <strong>classification</strong> results ........................................................ 47<br />

Table 5.1 Information of <strong>audio</strong> collections hosted in IMIRSEL .................................................. 54<br />

Table 5.2 Descriptions <strong>and</strong> statistics of the collections ................................................................ 56<br />

Table 5.3 Mood categories <strong>and</strong> song distributions ....................................................................... 60<br />

Table 5.4 Distribution of songs with multiple labels .................................................................... 63<br />

Table 5.5 Genre distribution of songs in the experiment dataset (“Other” includes genres<br />

occurring very infrequently such as “World,” “Folk,” “Easy listening,” <strong>and</strong> “Big b<strong>and</strong>”) .......... 66<br />

Table 5.6 Genre <strong>and</strong> <strong>mood</strong> distribution of positive examples ...................................................... 67<br />

Table 6.1 Summary of basic lyric features ................................................................................... 73<br />

Table 6.2 Text stylistic features evaluated in this research .......................................................... 78<br />

Table 6.3 Summary of linguistic <strong>and</strong> stylistic lyric features ........................................................ 78<br />

Table 6.4 Individual lyric feature type performances ................................................................... 81<br />

Table 6.5 Best performing concatenated lyric feature types ......................................................... 82<br />

Table 6.6 Performance comparison of “Content,” “FW,” “GI,” <strong>and</strong> “TextStyle”........................ 83<br />

Table 6.7 Feature selection for TextStyle ..................................................................................... 84<br />

Table 7.1 Comparisons on accuracies of two hybrid methods ..................................................... 93<br />

Table 7.2 Accuracies of single-source-based <strong>and</strong> hybrid systems ................................................ 94<br />

Table 7.3 Statistical tests on pair-wise system performances ....................................................... 95<br />

Table 7.4 Accuracies of lyric <strong>and</strong> <strong>audio</strong> feature types for individual categories .......................... 98<br />

Table 7.5 Top-ranked content word features for categories where content words significantly<br />

outperformed <strong>audio</strong> ....................................................................................................................... 99<br />

Table 7.6 Top GI features for “aggressive” <strong>mood</strong> category ....................................................... 100<br />

Table 7.7 Top-ranked GI-lex features for categories where GI-lex significantly outperformed<br />

<strong>audio</strong> ............................................................................................................................................ 100<br />

Table 7.8 Top ANEW <strong>and</strong> Affect-lex features for categories where ANEW or Affect-lex<br />

significantly outperformed <strong>audio</strong> ................................................................................................ 101<br />

Table 7.9 Top-ranked text stylistic features for categories where text stylistics significantly<br />

outperformed <strong>audio</strong> ..................................................................................................................... 102<br />

Table 7.10 Top lyric features in “calm” category ....................................................................... 103<br />

xi

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!