| { | |
| "paper_id": "U12-1012", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T03:07:30.548211Z" | |
| }, | |
| "title": "In Your Eyes: Identifying Clich\u00e9s in Song Lyrics", | |
| "authors": [ | |
| { | |
| "first": "Alex", | |
| "middle": [ | |
| "G" | |
| ], | |
| "last": "Smith", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "RMIT University GPO Box 2476V", | |
| "location": { | |
| "postCode": "3001", | |
| "settlement": "Melbourne", | |
| "country": "Australia" | |
| } | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "Christopher", | |
| "middle": [ | |
| "X S" | |
| ], | |
| "last": "Zee", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "RMIT University GPO Box 2476V", | |
| "location": { | |
| "postCode": "3001", | |
| "settlement": "Melbourne", | |
| "country": "Australia" | |
| } | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "Alexandra", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Uitdenbogerd", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "RMIT University GPO Box 2476V", | |
| "location": { | |
| "postCode": "3001", | |
| "settlement": "Melbourne", | |
| "country": "Australia" | |
| } | |
| }, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "We investigated methods for the discovery of clich\u00e9s from song lyrics. Trigrams and rhyme features were extracted from a collection of lyrics and ranked using term-weighting techniques such as tf-idf. These attributes were also examined over both time and genre. We present an application to produce a clich\u00e9 score for lyrics based on these findings and show that number one hits are substantially more clich\u00e9d than the average published song.", | |
| "pdf_parse": { | |
| "paper_id": "U12-1012", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "We investigated methods for the discovery of clich\u00e9s from song lyrics. Trigrams and rhyme features were extracted from a collection of lyrics and ranked using term-weighting techniques such as tf-idf. These attributes were also examined over both time and genre. We present an application to produce a clich\u00e9 score for lyrics based on these findings and show that number one hits are substantially more clich\u00e9d than the average published song.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "Song lyrics can be inspiring, moving, energetic and heart wrenching pieces of modern poetry. Other times, we find lyrics to be boring and uninspired, or clich\u00e9d. Some lyricists may aim to write truly original lyrics, while others are after a number one on the charts. The authors of The Manual (Drummond and Cauty, 1988) , who have several hits to their credit, state that to succeed in achieving a number one hit one needs to \"stick to the clich\u00e9s\" because \"they deal with the emotional topics we all feel\".", | |
| "cite_spans": [ | |
| { | |
| "start": 294, | |
| "end": 320, | |
| "text": "(Drummond and Cauty, 1988)", | |
| "ref_id": "BIBREF3" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Credits 2 Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "Despite dictionary definitions, it isn't easy to pinpoint what is clich\u00e9 and what isn't. Dillon (2006) explains that linguists tend to prefer the term idiom or fixed expression. He also points out the subjective nature of the decision as to whether a phrase is a clich\u00e9, illustrating this with some frequently used phrases that are not considered clich\u00e9d, and other phrases such as 'armed to the teeth' that are, despite their relative infrequent appearance within corpora.", | |
| "cite_spans": [ | |
| { | |
| "start": 89, | |
| "end": 102, | |
| "text": "Dillon (2006)", | |
| "ref_id": "BIBREF2" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Credits 2 Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "There is also a temporal component to whether something is clich\u00e9, since an expression would not be considered clich\u00e9 on its first use, but only after widespread adoption. For song lyrics, clich\u00e9s can arise due to the perceived need to make rhymes. Some words have limited possibilities for rhyme, and so using exact rhyme makes clich\u00e9 more likely. Early songwriters believed that a good song must have perfect rhyme in its lyrics. However, recent thought is that alternatives, such as assonance and additive or subtractive rhymes, are valid alternatives in order to avoid clich\u00e9d writing (Pattison, 1991) .", | |
| "cite_spans": [ | |
| { | |
| "start": 589, | |
| "end": 605, | |
| "text": "(Pattison, 1991)", | |
| "ref_id": "BIBREF16" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Credits 2 Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "In this paper we use an information retrieval approach to defining what is clich\u00e9d in song lyrics, by using human judgements. We use statistical measures to build ranked lists of clich\u00e9d trigrams and rhymes, then combine these results to produce an overall clich\u00e9 score for a song's lyrics. A simple count of the occurrences of terms in song lyrics, ranked according to frequency is likely to produce generic common phrases rather than lyric-specific terms. Therefore we investigated means of detecting typical rhymes and phrases in lyrics using a term-weighting technique. We examined trends in these attributes over musical genre and time. Using our results, we developed a clich\u00e9 score for song lyrics.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Credits 2 Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "The remainder of this paper is structured as follows: first, we discuss related work, then describe the data collection and preparation process. Next, our rhyme and collocation techniques and results are shown. Finally, we present our application for clich\u00e9 scoring.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Credits 2 Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "There are several areas of research that are relevant to our topic, such as other studies of lyrics, analysis of text, and work on rhyme. However, we have not found any work specifically on identifying clich\u00e9s in either songs or other works.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Related Work", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "Song lyrics have previously been studied for a variety of applications, including determining artist similarity (Logan et al., 2005) , genre classification (Mayer et al., 2008) and topic detection (Kleedorfer et al., 2008) . Whissell (1996) combined traditional stylometric measures with a technique for emotional description in order to successfully differentiate between lyrics written by Beatles Paul McCartney and John Lennon. In addition, several studies have recently appeared for retrieving songs based on misheard lyrics (Ring and Uitdenbogerd, 2009; Xu et al., 2009; Hirjee and Brown, 2010b) .", | |
| "cite_spans": [ | |
| { | |
| "start": 112, | |
| "end": 132, | |
| "text": "(Logan et al., 2005)", | |
| "ref_id": "BIBREF13" | |
| }, | |
| { | |
| "start": 156, | |
| "end": 176, | |
| "text": "(Mayer et al., 2008)", | |
| "ref_id": "BIBREF15" | |
| }, | |
| { | |
| "start": 197, | |
| "end": 222, | |
| "text": "(Kleedorfer et al., 2008)", | |
| "ref_id": "BIBREF9" | |
| }, | |
| { | |
| "start": 225, | |
| "end": 240, | |
| "text": "Whissell (1996)", | |
| "ref_id": "BIBREF21" | |
| }, | |
| { | |
| "start": 529, | |
| "end": 558, | |
| "text": "(Ring and Uitdenbogerd, 2009;", | |
| "ref_id": "BIBREF18" | |
| }, | |
| { | |
| "start": 559, | |
| "end": 575, | |
| "text": "Xu et al., 2009;", | |
| "ref_id": "BIBREF22" | |
| }, | |
| { | |
| "start": 576, | |
| "end": 600, | |
| "text": "Hirjee and Brown, 2010b)", | |
| "ref_id": "BIBREF7" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Related Work", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "Rhyme in poetry has been studied statistically for many years (for example a study of Shakespeare and Swinburne (Skinner, 1941) ). More recently Hirjee and Brown (2009) (Hirjee, 2010) introduced a probabilistic scoring model to identify rhymes in song lyrics based on prior study of the complex rhyme strategies found in Hip-Hop. They define a normal or 'perfect' rhyme as 'two syllables that share the same nucleus (vowel) and coda (ending consonants). They found this method more accurate than rule-based methods and developed a Rhyme Analyzer application based upon their model (Hirjee and Brown, 2010a) .", | |
| "cite_spans": [ | |
| { | |
| "start": 112, | |
| "end": 127, | |
| "text": "(Skinner, 1941)", | |
| "ref_id": "BIBREF20" | |
| }, | |
| { | |
| "start": 145, | |
| "end": 168, | |
| "text": "Hirjee and Brown (2009)", | |
| "ref_id": "BIBREF5" | |
| }, | |
| { | |
| "start": 169, | |
| "end": 183, | |
| "text": "(Hirjee, 2010)", | |
| "ref_id": "BIBREF8" | |
| }, | |
| { | |
| "start": 581, | |
| "end": 606, | |
| "text": "(Hirjee and Brown, 2010a)", | |
| "ref_id": "BIBREF6" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Related Work", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "In our work, we considered the use of collocation extraction for finding words that frequently appear together in lyrics. Smadja (1993) described several techniques for collocation extraction and implemented these in the Xtract tool. The precision of this tool is tested on a corpus of stock market reports and found to be 80% accurate. Lin (1998) defined a method for retrieving two word collocations using a broad coverage parser and applied this to compute word similarties. Word n-grams have been used elsewhere as features for authorship attribution of text ( Solorio, 2011; Koppel et al., 2009) , and source code (Burrows et al., 2009) .", | |
| "cite_spans": [ | |
| { | |
| "start": 337, | |
| "end": 347, | |
| "text": "Lin (1998)", | |
| "ref_id": "BIBREF12" | |
| }, | |
| { | |
| "start": 563, | |
| "end": 564, | |
| "text": "(", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 580, | |
| "end": 600, | |
| "text": "Koppel et al., 2009)", | |
| "ref_id": "BIBREF11" | |
| }, | |
| { | |
| "start": 619, | |
| "end": 641, | |
| "text": "(Burrows et al., 2009)", | |
| "ref_id": "BIBREF1" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Related Work", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "Our aim was to detect and measure clich\u00e9s in song lyrics. In normal text, clich\u00e9s are likely to be stock phrases, such as \"par for the course\". Frequently used phrases can be found by looking at n-grams or collocated terms in text. The second source of clich\u00e9 in song lyrics arises from rhyme pairs. Due to the typical subject matter of pop songs, and the tendency toward perfect rhyme use, particular rhyme pairs are likely to be common. Our approach was to obtain a large collection of song lyrics, observe the effect of different formulations for ranking rhyme pairs and collocated terms, then create a clich\u00e9 measure for songs based on the most promising ranking formulae. These were to be evaluated with human judgements.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Experiments", | |
| "sec_num": "4" | |
| }, | |
| { | |
| "text": "The collection was composed of a subset of lyrics gathered from online lyrics database LyricWiki 1 using a web crawler. Song title and artist meta-data were also retrieved. All lyrics were converted to lower case and punctuation removed. Digits between one and ten were replaced with their written equivalent. Duplicate lyrics were found to be a problem, for example the song 'A-OK' by 'Face to Face' was submitted three times under different names as 'A.O.K', 'A-Ok' and 'AOK'. Variations between lyrics included case, punc- tuation and white space. Removing these distinctions allowed us to identify and discard many duplicates. This process resulted in a collection of 39,236 items by 1847 artists. As our focus was on English lyrics, therefore the world music items were also discarded, removing the majority of foreign language lyrics. This reduced our collection to 34,855 items by 1590 artists.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Lyrics Collection", | |
| "sec_num": "4.1" | |
| }, | |
| { | |
| "text": "Using collected meta-data, we were able to query music statistics website last.fm 2 to determine the genre of each artist. We considered the top three genre tags for each and performed a broad general categorisation of each. This was done by checking if the artist was tagged as one of our pre-defined genres. If not, we checked the tags against a list of subgenres; for example 'thrash metal' was classified as 'metal' and 'house' fit into 'electronic' music. This resulted in the genre distribution shown in Table 1 .", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 510, | |
| "end": 517, | |
| "text": "Table 1", | |
| "ref_id": null | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Genre distribution", | |
| "sec_num": "4.2" | |
| }, | |
| { | |
| "text": "In order to find typical rhymes and phrases, we applied the term-weighting scheme tf-idf (Salton and Buckley, 1988) to our collection. As a bag-of-words approach, tf-idf considers terms with no regard to the order in which they appear in a document. The objective of this scheme was to highlight those terms that occur more frequently in a given document, but less often in the remainder of the corpus.", | |
| "cite_spans": [ | |
| { | |
| "start": 89, | |
| "end": 115, | |
| "text": "(Salton and Buckley, 1988)", | |
| "ref_id": "BIBREF19" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Lyric Attributes", | |
| "sec_num": "4.3" | |
| }, | |
| { | |
| "text": "The term frequency tf for a document is given by the number of times the term appears in the document. The number of documents in the corpus containing the term determines document frequency, df. With the corpus size denoted D, we calculate a term t's weight by tf (t) \u00d7 ln(D/df (t)).", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Lyric Attributes", | |
| "sec_num": "4.3" | |
| }, | |
| { | |
| "text": "We modified Hirjee and Brown's 'Rhyme Analyzer' software in order to gather a set of all rhymes from our LyricWiki dataset. The pairs were then sorted by frequency, with reversed pairs, such as to/do and do/to, being combined. To lower the ranking of pairs that are likely to occur in ordinary text rather than in songs, we used tf-idf values calculated for rhyme pairs extracted from a corpus consisting of eighty-two Project Gutenberg 3 texts. The size of this corpus was approximately the same as the lyric collection. Note that most of the corpus was ordinary prose. Table 2 shows that tf-idf has increased the rank of rhyme pairs such as 'right night' and introduced new ones like 'heart apart' and 'night light'. While not occurring as frequently in the lyrics collection, these pairs are given a greater weight due to their less frequent appearance in the Gutenberg texts. Note also, that the \"rhyme pairs\" found in the Gutenberg texts are not what one would normally think of as rhymes in poetry or songs, even though they have some similarity. This is due to the nature of the rhyme analyser in use, in that it identifies rhymes regardless of where in a line they occur, and also includes other commonly used rhyme-like devices, such as partial rhymes (for example, \"pretty\" and \"little\" (Pattison, 1991) ). The benefit of using the Gutenberg text in this way is that spurious rhymes of high frequency in normal text can easily be filtered out. The technique may also make a rhyme analyser more robust, but that is not our purpose in this paper.", | |
| "cite_spans": [ | |
| { | |
| "start": 1297, | |
| "end": 1313, | |
| "text": "(Pattison, 1991)", | |
| "ref_id": "BIBREF16" | |
| } | |
| ], | |
| "ref_spans": [ | |
| { | |
| "start": 571, | |
| "end": 578, | |
| "text": "Table 2", | |
| "ref_id": "TABREF2" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Rhyme Pairs", | |
| "sec_num": "4.4" | |
| }, | |
| { | |
| "text": "The results of grouping the lyrics by genre and performing tf-idf weighting are shown in ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Rhyme Pairs", | |
| "sec_num": "4.4" | |
| }, | |
| { | |
| "text": "All possible trigrams were extracted from the LyricWiki collection and Gutenberg texts. Again, tf-idf was used to rank the trigrams, with a second list removing trigrams containing terms from the NLTK list of English stopwords (Bird et al., 2009) and repeated syllables such as 'la'. Table 4 , we can see that the difference between frequency and tf-idf in the top fifteen are both positional changes and the introduction of new terms. For example, 'i love you' is ranked fifth by frequency, but fifteenth using tf-idf. Also note how phrases such as 'its time to' and 'i just wanna' rank higher using tf-idf. This shows the influence of the Gutenberg texts -common English phrasing is penalised and lyric-centric terminology emphasised.", | |
| "cite_spans": [ | |
| { | |
| "start": 227, | |
| "end": 246, | |
| "text": "(Bird et al., 2009)", | |
| "ref_id": "BIBREF0" | |
| } | |
| ], | |
| "ref_spans": [ | |
| { | |
| "start": 284, | |
| "end": 291, | |
| "text": "Table 4", | |
| "ref_id": "TABREF5" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Collocations", | |
| "sec_num": "4.5" | |
| }, | |
| { | |
| "text": "Interestingly, the filtered tf-idf scores 'll cool j' the highest. This is the name of a rapper, whose name appears in 136 songs within our collection. Hirjee's work involving hip-hop lyrics found that rappers have a tendency to 'name-drop in their lyrics, including their own names, nicknames, and record label and group names' (Hirjee, 2010) . Examining these lyrics, we determined that many of these occurrences can be attributed to this practice, while others are annotations in the lyrics showing the parts performed by LL Cool J which we did not remove prior to experimentation.", | |
| "cite_spans": [ | |
| { | |
| "start": 329, | |
| "end": 343, | |
| "text": "(Hirjee, 2010)", | |
| "ref_id": "BIBREF8" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Collocations", | |
| "sec_num": "4.5" | |
| }, | |
| { | |
| "text": "Substituting document frequency for term frequency in the lyric collection, as in Table Decade Collection 2000 -2010 55.41% 1990 -2000 33.49% 1980 -1990 7.08% 1970 -1980 2.88% 1960 -1970 0.52% Table 7 : Time distribution of lyrics.", | |
| "cite_spans": [ | |
| { | |
| "start": 108, | |
| "end": 188, | |
| "text": "2000 -2010 55.41% 1990 -2000 33.49% 1980 -1990 7.08% 1970 -1980 2.88% 1960 -1970", | |
| "ref_id": null | |
| } | |
| ], | |
| "ref_spans": [ | |
| { | |
| "start": 82, | |
| "end": 96, | |
| "text": "Table Decade", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 195, | |
| "end": 202, | |
| "text": "Table 7", | |
| "ref_id": null | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Collocations", | |
| "sec_num": "4.5" | |
| }, | |
| { | |
| "text": "6, decreases the weight of trigrams that occur repeatedly in fewer songs. As a result, this 'dfidf' should increase the quality of the ranked list. We see that the syllable repetition is largely absent from the top ranking terms and among other positional changes, the phrase 'rock n roll' drops from second place to thirteenth.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Collocations", | |
| "sec_num": "4.5" | |
| }, | |
| { | |
| "text": "Several interesting trends are present in Table 5, which shows df-idf ranked trigrams by genre. Firstly, Hip-hop shows a tendency to use coarse language more frequently and genre-specific phrasing like 'in the club' and 'in the hood'. Repeated terms as in 'oh oh oh' and 'yeah yeah yeah' were more prevalent in pop and rock music. Such vocal hooks may be attempts to create catchy lyrics to sing along to. Love appears to be a common theme in pop music, with phrases like 'you and me', 'youre the one' and of course, 'i love you' ranking high. This terminology is shared by the the other genres to a lesser extent, except in the cases of hip-hop, punk and metal, where it seems largely absent. The term 'words music by' within the metal category is the result of author attribution annotations within the lyrics.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Collocations", | |
| "sec_num": "4.5" | |
| }, | |
| { | |
| "text": "There is a temporal component to clich\u00e9s. There was probably a time when the lines \"I'm begging you please, I'm down on my knees\", or the trigram \"end of time\" sounded fresh to the sophisticated audience. Fashions and habits in language also change over time. In this section we examine the rhyme pairs and trigrams across four time periods. We queried the MusicBrainz 4 database using song title and artist in order to determine the first year of release for each song. This method was able to identify 22,419 songs, or 59% of our collection. Given the considerable size of MusicBrainz (10,761,729 tracks and 618,224 artists on 30th March 2011), this relatively low success rate can likely be at-4 http://musicbrainz.org 1990-1995 1995-2000 2000-2005 2005-2010 its time to i got a its time to its time to i got a its time to i got the dont give a i dont wanna i dont wanna dont give a i got the in the club i need to i got a me and my i need to you need to i love you i got a dont give a im in the and you know i need to and you know and i know here we go check it out and i know in the back check it out in the back in the back i got the in the back i dont wanna i got the i try to i dont wanna you need to tributed to incorrect or partial meta-data retrieved from LyricWiki rather than incompleteness of the database. As shown in Table 7 our collection has a significant inclination towards music of the last twenty years, with over half in the last decade. It is suspected that this is again due to the nature of the source database -the users are likely to be younger and submitting lyrics they are more familiar with. Also, the distribution peak corresponds to the Web era, in which it has been easier to share lyrics electronically.", | |
| "cite_spans": [ | |
| { | |
| "start": 722, | |
| "end": 751, | |
| "text": "1990-1995 1995-2000 2000-2005", | |
| "ref_id": null | |
| } | |
| ], | |
| "ref_spans": [ | |
| { | |
| "start": 1333, | |
| "end": 1340, | |
| "text": "Table 7", | |
| "ref_id": null | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Time", | |
| "sec_num": "4.6" | |
| }, | |
| { | |
| "text": "The lyrics were divided into 5 year periods from 1990 to 2010 and 2000 random songs selected from each. Rhyme pairs and trigrams were then found with the aforementioned methods, as shown in Tables 8 and 9 .", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 190, | |
| "end": 204, | |
| "text": "Tables 8 and 9", | |
| "ref_id": "TABREF9" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Time", | |
| "sec_num": "4.6" | |
| }, | |
| { | |
| "text": "We tested several clich\u00e9 measures that combined the two components of our approach, being rhyme pairs and trigrams. We used precomputed tf-idf scores based on the Gutenberg collection for rhyme pairs and df-idf trigram weights. In this model, R and C are the sets of scored rhymes and trigrams respectively. The rhyme pairs and trigrams found in the given lyrics are represented by r and c. The length of the song lyrics in lines is denoted L, and |R| denotes the number of rhyme pairs in the collection. Our ground truth was based on human judgements. One coauthor prepared a list of ten songs, five of which were considered to be clich\u00e9d, and five less typical. The list was subjectively ranked by each coauthor from most to least clich\u00e9d. Spearman's rank-order correlation coefficient was used to compare each author's rankings of the songs. Two authors had a correlation of correlation of 0.79. The third author's rankings had correlations of -0.2 and -0.5 with each of the other authors, leading to a suspicion that the list was numbered in the opposite order. When reversed the correlations were very weak (0.09 and -0.1 respectively). We chose to work with an aver-age of the two more highly correlated sets of judgements.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Clich\u00e9 Scores for Songs", | |
| "sec_num": "4.7" | |
| }, | |
| { | |
| "text": "We report on results for the formulae shown below.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Clich\u00e9 Scores for Songs", | |
| "sec_num": "4.7" | |
| }, | |
| { | |
| "text": "\ufffd R(r) + \ufffd C(c) L (1) P (R(r))+1 |R|+1 + \ufffd C(c) L (2) ( \ufffd (R(r)) + 1 |R| + 1 + \ufffd C(c)) \u00d7 ln(L + 1) (3) ( \ufffd (R(r)) + 1 |R| + 1 \u00d7 \ufffd C(c)) \u00d7 ln(L + 1) (4)", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Clich\u00e9 Scores for Songs", | |
| "sec_num": "4.7" | |
| }, | |
| { | |
| "text": "The lyrics were then ranked according to each equation.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Clich\u00e9 Scores for Songs", | |
| "sec_num": "4.7" | |
| }, | |
| { | |
| "text": "An average rank list was prepared, and as the rankings of the third coauthor were an outlier, they were not included. Kendall's Tau and Spearman's rho were then used to compare this list to the equation rankings. These were chosen as they are useful for measuring correlation between two ordered datasets.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Clich\u00e9 Scores for Songs", | |
| "sec_num": "4.7" | |
| }, | |
| { | |
| "text": "In order to test the accuracy of the application, we randomly selected ten songs from our collection and again subjectively ranked them from most to least clich\u00e9. Table 10 shows that the third equation produces the ranked list that best correlates with the coauthor-generated rankings. Table 11 shows the rankings obtained applying the same formula to the test data. Table 12 shows the correlations obtained when compared to each author's ranked list. The results show a drop to about 50% of the values obtained using the training set.", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 163, | |
| "end": 171, | |
| "text": "Table 10", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 286, | |
| "end": 294, | |
| "text": "Table 11", | |
| "ref_id": "TABREF16" | |
| }, | |
| { | |
| "start": 367, | |
| "end": 375, | |
| "text": "Table 12", | |
| "ref_id": "TABREF2" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Clich\u00e9 Scores for Songs", | |
| "sec_num": "4.7" | |
| }, | |
| { | |
| "text": "The third equation showed the greatest correlation with human ranking of songs by clich\u00e9. It suggests that log normalisation according to song length applied to the trigram score component in isolation, with the rhyme score normalised by the number of rhymes. Dividing the summed score by the length of the song (Equation 1) performed relatively poorly. It is possible that introducing a scaling factor into Equation 3 to modify the relative weights of the rhyme and trigram components may yield better results. Oddly, the somewhat less principled formulation, Equation 2, with its doubled normalisation of the rhyming component outperformed Equation 1. Perhaps this suggests that trigrams should dominate the formula.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Discussion", | |
| "sec_num": "4.9" | |
| }, | |
| { | |
| "text": "The different expectations of what clich\u00e9d lyrics are resulted in three distinct lists. However, there are some common rankings, for example it was unanimous that Walkin' on the Sidewalks by Queens of the Stone Age was the least clich\u00e9d song. In this case, the application ranks did not correlate as well with the experimental lists as the training set. Our judgements about how clich\u00e9d a song is are generally based on what we have heard before. The application has a similar limitation in that it ranks according to the scores from our lyric collection. The discrepancy between the ranked lists may be due to this differerence in lyrical exposure, or more simply, a suboptimal scoring equation.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Discussion", | |
| "sec_num": "4.9" | |
| }, | |
| { | |
| "text": "The list of songs was also more difficult to rank, as the songs in it probably didn't differ greatly in clich\u00e9dness compared to the handselected set. For example, using Equation 3, the range of scores for the training set was 12.2 for Carry, and 5084 for Just a Dream, whereas, the test set had a range from 26.76 to 932.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Discussion", | |
| "sec_num": "4.9" | |
| }, | |
| { | |
| "text": "Another difficulty when making the human judgements was the risk of judging on quality rather than lyric clich\u00e9dness. While a poor quality lyric may be clich\u00e9d, the two attributes do not necessarily always go together.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Discussion", | |
| "sec_num": "4.9" | |
| }, | |
| { | |
| "text": "Our results suggest that there are limitations in how closely human judges agree on how clich\u00e9d songs are relative to each other, which may mean that only a fairly coarse clich\u00e9 measure is possible. Perhaps the use of expert judges, such as professional lyricists or songwriting educators, may result in greater convergence of opinion.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Discussion", | |
| "sec_num": "4.9" | |
| }, | |
| { | |
| "text": "Hits?", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "How Clich\u00e9d Are Number One", | |
| "sec_num": "5" | |
| }, | |
| { | |
| "text": "Building on this result, we compared the scores of popular music from 1990-2010 with our collection. A set of 286 number one hits as determined by the Billboard Hot 100 5 from this time period were retrieved and scored using the aforementioned method. We compared the distribution of scores with those from the LyricWiki collection over the same era. The score distribution is shown in Figure 1, and suggests that number one hits are more clich\u00e9d than other songs on average. There are several possible explanations for this result: it may be that number one Figure 1 : Boxplots showing the quartiles of lyric scores for the lyric collection from the 1990-2010 era, and the corresponding set of number one hits from the era.", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 386, | |
| "end": 392, | |
| "text": "Figure", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 559, | |
| "end": 567, | |
| "text": "Figure 1", | |
| "ref_id": null | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "How Clich\u00e9d Are Number One", | |
| "sec_num": "5" | |
| }, | |
| { | |
| "text": "hits are indeed more typical than other songs, or perhaps that a song that reaches number one influences other lyricists who then create works in a similar style. Earlier attempts to compare number one hits with the full collection of lyrics revealed an increase in clich\u00e9 score over time for hit songs. We believe that this was not so much due to an increase in clich\u00e9 in pop over time but that the language in the lyrics of popular music changes over time, as happens with all natural language.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "How Clich\u00e9d Are Number One", | |
| "sec_num": "5" | |
| }, | |
| { | |
| "text": "We have explored the use of tf-idf weighting to find typical phrases and rhyme pairs in song lyrics. These attributes have been extracted with varying degrees of success, dependent on sample size. The use of a background model of text worked well in removing ordinary lan-guage from the results, and the technique may go towards improving rhyme detection software.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusion", | |
| "sec_num": "6" | |
| }, | |
| { | |
| "text": "An application was developed that estimates how clich\u00e9d given song lyrics are. However, while results were reasonable for distinguishing very clich\u00e9d songs from songs that are fairly free from clich\u00e9, it was less successful with songs that are not at the extremes.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusion", | |
| "sec_num": "6" | |
| }, | |
| { | |
| "text": "Our method of obtaining human judgements was not ideal, consisting of two rankings of ten songs by the research team involved in the project. For our future work we hope to obtain independent judgements, possibly of smaller snippets of songs to make the task easier. As it is unclear how consistent people are in judging the clich\u00e9dness of songs, we expect to collect a larger set of judgements per lyric.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusion", | |
| "sec_num": "6" | |
| }, | |
| { | |
| "text": "There were several instances where annotations in the lyrics influenced our results. Future work would benefit from a larger, more accurately transcribed collection. This could be achieved using Multiple Sequence Alignment as in Knees et al. (2005) . Extending the model beyond trigrams may also yield interesting results.", | |
| "cite_spans": [ | |
| { | |
| "start": 229, | |
| "end": 248, | |
| "text": "Knees et al. (2005)", | |
| "ref_id": "BIBREF10" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusion", | |
| "sec_num": "6" | |
| }, | |
| { | |
| "text": "A comparison of number one hits with a larger collection of lyrics from the same time period revealed that the typical number one hit is more clich\u00e9d, on average. While we have examined the relationship between our clich\u00e9 score and song popularity, it is important to note that there is not necessarily a connection between these factors and writing quality, but this may also be an interesting area to explore.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusion", | |
| "sec_num": "6" | |
| }, | |
| { | |
| "text": "http://lyrics.wikia.com", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "http://www.last.fm", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "http://www.gutenberg.org", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "http://www.billboard.com", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "Natural Language Processing with Python", | |
| "authors": [ | |
| { | |
| "first": "Steven", | |
| "middle": [], | |
| "last": "Bird", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Edward", | |
| "middle": [], | |
| "last": "Loper", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Ewan", | |
| "middle": [], | |
| "last": "Klein", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Steven Bird, Edward Loper, and Ewan Klein. 2009. Natural Language Processing with Python. O'Reilly Media Inc.", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "Application of information retrieval techniques for source code authorship attribution", | |
| "authors": [ | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Burrows", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Uitdenbogerd", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Turpin", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "International Conference on Database Systems for Advanced Applications", | |
| "volume": "14", | |
| "issue": "", | |
| "pages": "699--713", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "S. Burrows, A. L. Uitdenbogerd, and A. Turpin. 2009. Application of information retrieval tech- niques for source code authorship attribution. In Ramamohanarao Kotagiri Xiaofang Zhou, Haruo Yokota and Xuemin Lin, editors, Inter- national Conference on Database Systems for Advanced Applications, volume 14, pages 699- 713, Brisbane, Australia, April.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "Corpus, creativity, clich\u00e9: Where statistics meet aesthetics", | |
| "authors": [ | |
| { | |
| "first": "G", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Dillon", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2006, | |
| "venue": "Journal of Literary Semantics", | |
| "volume": "35", | |
| "issue": "2", | |
| "pages": "97--103", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "G. L. Dillon. 2006. Corpus, creativity, clich\u00e9: Where statistics meet aesthetics. Journal of Literary Semantics, 35(2):97-103.", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "The Manual (How to Have a Number One the Easy Way)", | |
| "authors": [ | |
| { | |
| "first": "B", | |
| "middle": [], | |
| "last": "Drummond", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "J", | |
| "middle": [], | |
| "last": "Cauty", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1988, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "B. Drummond and J. Cauty. 1988. The Manual (How to Have a Number One the Easy Way).", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "Automatic detection of internal and imperfect rhymes in rap lyrics", | |
| "authors": [ | |
| { | |
| "first": "H", | |
| "middle": [], | |
| "last": "Hirjee", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "D", | |
| "middle": [ | |
| "G" | |
| ], | |
| "last": "Brown", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "Proceedings of the 10th International Society for Music Information Retrieval Conference", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "H. Hirjee and D.G. Brown. 2009. Automatic de- tection of internal and imperfect rhymes in rap lyrics. In Proceedings of the 10th International Society for Music Information Retrieval Con- ference.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "Rhyme Analyzer: An Analysis Tool for Rap Lyrics", | |
| "authors": [ | |
| { | |
| "first": "H", | |
| "middle": [], | |
| "last": "Hirjee", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "D", | |
| "middle": [ | |
| "G" | |
| ], | |
| "last": "Brown", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2010, | |
| "venue": "Proceedings of the 11th International Society for Music Information Retrieval Conference", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "H. Hirjee and D.G. Brown. 2010a. Rhyme An- alyzer: An Analysis Tool for Rap Lyrics. In Proceedings of the 11th International Society for Music Information Retrieval Conference.", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "Solving misheard lyric search queries using a probabilistic model of speech sounds", | |
| "authors": [ | |
| { | |
| "first": "H", | |
| "middle": [], | |
| "last": "Hirjee", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "D", | |
| "middle": [ | |
| "G" | |
| ], | |
| "last": "Brown", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2010, | |
| "venue": "Proc. 11th International Society ofor Music Information Retrieval Conference", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "147--152", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "H. Hirjee and D.G. Brown. 2010b. Solving mis- heard lyric search queries using a probabilistic model of speech sounds. In Proc. 11th Interna- tional Society ofor Music Information Retrieval Conference, pages 147-152, Utrecht, Nether- lands, August. ISMIR.", | |
| "links": null | |
| }, | |
| "BIBREF8": { | |
| "ref_id": "b8", | |
| "title": "Rhyme, rhythm, and rhubarb: Using probabilistic methods to analyze hip hop, poetry, and misheard lyrics", | |
| "authors": [ | |
| { | |
| "first": "Hussein", | |
| "middle": [], | |
| "last": "Hirjee", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2010, | |
| "venue": "Master's thesis", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Hussein Hirjee. 2010. Rhyme, rhythm, and rhubarb: Using probabilistic methods to ana- lyze hip hop, poetry, and misheard lyrics. Mas- ter's thesis, University of Waterloo.", | |
| "links": null | |
| }, | |
| "BIBREF9": { | |
| "ref_id": "b9", | |
| "title": "Oh Oh Oh Whoah! Towards Automatic Topic Detection In Song Lyrics", | |
| "authors": [ | |
| { | |
| "first": "Florian", | |
| "middle": [], | |
| "last": "Kleedorfer", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Peter", | |
| "middle": [], | |
| "last": "Knees", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Tim", | |
| "middle": [], | |
| "last": "Pohle", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2008, | |
| "venue": "Proceedings of the 9th International Conference on Music Information Retrieval (ISMIR 2008)", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "287--292", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Florian Kleedorfer, Peter Knees, and Tim Pohle. 2008. Oh Oh Oh Whoah! Towards Automatic Topic Detection In Song Lyrics. In Proceed- ings of the 9th International Conference on Mu- sic Information Retrieval (ISMIR 2008), pages 287-292, September.", | |
| "links": null | |
| }, | |
| "BIBREF10": { | |
| "ref_id": "b10", | |
| "title": "Multiple lyrics alignment: Automatic retrieval of song lyrics", | |
| "authors": [ | |
| { | |
| "first": "P", | |
| "middle": [], | |
| "last": "Knees", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Schedl", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "G", | |
| "middle": [], | |
| "last": "Widmer", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2005, | |
| "venue": "Proceedings of 6th international conference on music information retrieval (is-mir05)", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "564--569", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "P. Knees, M. Schedl, and G. Widmer. 2005. Mul- tiple lyrics alignment: Automatic retrieval of song lyrics. In Proceedings of 6th international conference on music information retrieval (is- mir05), pages 564-569.", | |
| "links": null | |
| }, | |
| "BIBREF11": { | |
| "ref_id": "b11", | |
| "title": "Computational methods in authorship attribution", | |
| "authors": [ | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Koppel", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "J", | |
| "middle": [], | |
| "last": "Schler", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Argamon", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "Journal of the American Society", | |
| "volume": "60", | |
| "issue": "1", | |
| "pages": "9--26", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "M. Koppel, J. Schler, and S. Argamon. 2009. Computational methods in authorship attribu- tion. Journal of the American Society for in- formation Science and Technology, 60(1):9-26.", | |
| "links": null | |
| }, | |
| "BIBREF12": { | |
| "ref_id": "b12", | |
| "title": "Extracting collocations from text corpora", | |
| "authors": [ | |
| { | |
| "first": "Dekang", | |
| "middle": [], | |
| "last": "Lin", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1998, | |
| "venue": "First Workshop on Computational Terminology", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "57--63", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Dekang Lin. 1998. Extracting collocations from text corpora. In First Workshop on Computa- tional Terminology, pages 57-63.", | |
| "links": null | |
| }, | |
| "BIBREF13": { | |
| "ref_id": "b13", | |
| "title": "Semantic analysis of song lyrics", | |
| "authors": [ | |
| { | |
| "first": "B", | |
| "middle": [], | |
| "last": "Logan", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Kositsky", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "P", | |
| "middle": [], | |
| "last": "Moreno", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2005, | |
| "venue": "Multimedia and Expo, 2004. ICME'04. 2004 IEEE International Conference on", | |
| "volume": "2", | |
| "issue": "", | |
| "pages": "827--830", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "B. Logan, A. Kositsky, and P. Moreno. 2005. Se- mantic analysis of song lyrics. In Multimedia and Expo, 2004. ICME'04. 2004 IEEE Interna- tional Conference on, volume 2, pages 827-830. IEEE.", | |
| "links": null | |
| }, | |
| "BIBREF14": { | |
| "ref_id": "b14", | |
| "title": "Foundations of statistical natural language processing", | |
| "authors": [ | |
| { | |
| "first": "D", | |
| "middle": [], | |
| "last": "Christopher", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Hinrich", | |
| "middle": [], | |
| "last": "Manning", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Sch\u00fctze", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1999, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Christopher D. Manning and Hinrich Sch\u00fctze. 1999. Foundations of statistical natural lan- guage processing. MIT Press, Cambridge, MA, USA.", | |
| "links": null | |
| }, | |
| "BIBREF15": { | |
| "ref_id": "b15", | |
| "title": "Rhyme and style features for musical genre classification by song lyrics", | |
| "authors": [ | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Mayer", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Neumayer", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Rauber", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2008, | |
| "venue": "ISMIR 2008: proceedings of the 9th International Conference of Music Information Retrieval", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Mayer, R. Neumayer, and A. Rauber. 2008. Rhyme and style features for musical genre clas- sification by song lyrics. In ISMIR 2008: pro- ceedings of the 9th International Conference of Music Information Retrieval, page 337.", | |
| "links": null | |
| }, | |
| "BIBREF16": { | |
| "ref_id": "b16", | |
| "title": "Songwriting : essential guide to rhyming : a step-by-step guide to better rhyming and lyrics", | |
| "authors": [ | |
| { | |
| "first": "P", | |
| "middle": [], | |
| "last": "Pattison", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1991, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "P. Pattison. 1991. Songwriting : essential guide to rhyming : a step-by-step guide to better rhyming and lyrics. Berklee Press, Boston.", | |
| "links": null | |
| }, | |
| "BIBREF17": { | |
| "ref_id": "b17", | |
| "title": "Authorship attribution of web forum posts", | |
| "authors": [ | |
| { | |
| "first": "S", | |
| "middle": [ | |
| "R" | |
| ], | |
| "last": "Pillay", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "T", | |
| "middle": [], | |
| "last": "Solorio", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2010, | |
| "venue": "eCrime Researchers Summit (eCrime)", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "1--7", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "S.R. Pillay and T. Solorio. 2011. Authorship at- tribution of web forum posts. In eCrime Re- searchers Summit (eCrime), 2010, pages 1-7. IEEE.", | |
| "links": null | |
| }, | |
| "BIBREF18": { | |
| "ref_id": "b18", | |
| "title": "Finding 'Lucy in Disguise': the misheard lyric matching problem", | |
| "authors": [ | |
| { | |
| "first": "N", | |
| "middle": [], | |
| "last": "Ring", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Uitdenbogerd", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "The Fifth Asia Information Retrieval Symposium", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "N. Ring and A. L. Uitdenbogerd. 2009. Finding 'Lucy in Disguise': the misheard lyric matching problem. In The Fifth Asia Information Re- trieval Symposium, Sapporo, Japan, October.", | |
| "links": null | |
| }, | |
| "BIBREF19": { | |
| "ref_id": "b19", | |
| "title": "Term-weighting approaches in automatic text retrieval", | |
| "authors": [ | |
| { | |
| "first": "Gerard", | |
| "middle": [], | |
| "last": "Salton", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Christopher", | |
| "middle": [], | |
| "last": "Buckley", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1988, | |
| "venue": "Inf. Process. Manage", | |
| "volume": "24", | |
| "issue": "", | |
| "pages": "513--523", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Gerard Salton and Christopher Buckley. 1988. Term-weighting approaches in automatic text retrieval. Inf. Process. Manage., 24:513-523, August.", | |
| "links": null | |
| }, | |
| "BIBREF20": { | |
| "ref_id": "b20", | |
| "title": "A quantitative estimate of certain types of sound-patterning in poetry", | |
| "authors": [ | |
| { | |
| "first": "B", | |
| "middle": [ | |
| "F" | |
| ], | |
| "last": "Skinner", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1941, | |
| "venue": "Smadja. 1993. Retrieving collocations from text: Xtract. Computational linguistics", | |
| "volume": "54", | |
| "issue": "", | |
| "pages": "143--177", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "B. F. Skinner. 1941. A quantitative estimate of certain types of sound-patterning in poetry. American Journal of Psychology, 54(1):64-79, January. F. Smadja. 1993. Retrieving collocations from text: Xtract. Computational linguistics, 19(1):143-177.", | |
| "links": null | |
| }, | |
| "BIBREF21": { | |
| "ref_id": "b21", | |
| "title": "Traditional and emotional stylometric analysis of the songs of", | |
| "authors": [ | |
| { | |
| "first": "C", | |
| "middle": [], | |
| "last": "", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1996, | |
| "venue": "Computers and the Humanities", | |
| "volume": "30", | |
| "issue": "3", | |
| "pages": "257--265", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "C. Whissell. 1996. Traditional and emotional sty- lometric analysis of the songs of Beatles Paul McCartney and John Lennon. Computers and the Humanities, 30(3):257-265.", | |
| "links": null | |
| }, | |
| "BIBREF22": { | |
| "ref_id": "b22", | |
| "title": "Robust and fast lyric search based on phonetic confusion matrix", | |
| "authors": [ | |
| { | |
| "first": "X", | |
| "middle": [], | |
| "last": "Xu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Naito", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "T", | |
| "middle": [], | |
| "last": "Kato", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "K", | |
| "middle": [], | |
| "last": "Kawai", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "Proc. 10th International Society ofor Music Information Retrieval Conference", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "417--422", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "X. Xu, M. Naito, T. Kato, and K. Kawai. 2009. Robust and fast lyric search based on phonetic confusion matrix. In K. Hirata and G. Tzane- takis, editors, Proc. 10th International Society ofor Music Information Retrieval Conference, pages 417-422, Kobe, Japan, October. ISMIR.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "TABREF2": { | |
| "text": "", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td>: Top fifteen rhyme pairs ranked by fre-</td></tr><tr><td>quency and tf-idf.</td></tr></table>" | |
| }, | |
| "TABREF3": { | |
| "text": "", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td>Rock</td><td>Hip-Hop</td><td>Electronic</td><td>Punk</td><td>Metal</td><td>Pop</td><td>Alternative</td><td>Folk</td></tr><tr><td>day away</td><td>way day</td><td>stay away</td><td>da na</td><td>night light</td><td>sha la</td><td>sha la</td><td>la da</td></tr><tr><td>way day</td><td>way say</td><td>day away</td><td>day away</td><td>la da</td><td>way say</td><td>insane brain</td><td>light night</td></tr><tr><td>say away</td><td>right night</td><td>control soul</td><td>say away</td><td>day away</td><td>feel real</td><td>alright tonight</td><td>day say</td></tr><tr><td>night light</td><td>good hood</td><td>say way</td><td>day way</td><td>real feel</td><td>say ok</td><td>little pretty</td><td>hand stand</td></tr><tr><td>way say</td><td>away day</td><td>getting better</td><td>way say</td><td>near fear</td><td>said head</td><td>write tonight</td><td>away day</td></tr><tr><td>stay away</td><td>dead head</td><td>way day</td><td>play day</td><td>say away</td><td>oh know</td><td>know oh</td><td>wave brave</td></tr><tr><td>night right</td><td>feel real</td><td>night right</td><td>day say</td><td>head dead</td><td>say day</td><td>way say</td><td>stride fried</td></tr><tr><td>way away</td><td>little bit</td><td>say away</td><td>way away</td><td>soul control</td><td>la da</td><td>la da</td><td>head dead</td></tr><tr><td>oh know</td><td>say play</td><td>heart start</td><td>head dead</td><td>high sky</td><td>right night</td><td>said head</td><td>sunday monday</td></tr><tr><td>da la</td><td>pretty little</td><td>light night</td><td>bed head</td><td>eyes lies</td><td>sha na</td><td>real feel</td><td>radio na</td></tr></table>" | |
| }, | |
| "TABREF4": { | |
| "text": "Genre specific top ten rhyme pairs ranked by tf-idf.", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table/>" | |
| }, | |
| "TABREF5": { | |
| "text": "", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td>provides a compar-</td></tr><tr><td>ison of these techniques with raw frequency</td></tr><tr><td>weighting. Similar attempts using techniques</td></tr><tr><td>such as Pointwise Mutual Information, Stu-</td></tr><tr><td>dent's t-test, the Chi-Squared Test and Like-</td></tr><tr><td>lihood Ratio (Manning and Sch\u00fctze, 1999) did</td></tr><tr><td>not yield promising ranked lists and are not</td></tr><tr><td>included in this paper.</td></tr><tr><td>From</td></tr></table>" | |
| }, | |
| "TABREF7": { | |
| "text": "Top fifteen trigrams, ranked by term frequency and tf-idf.", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td>Rock</td><td>Hip-Hop</td><td>Electronic</td><td>Punk</td><td>Metal</td><td>Pop</td><td>Alternative</td><td>Folk</td></tr><tr><td>its time to</td><td>in the club</td><td>its time to</td><td>its time to</td><td>its time to</td><td>i love you</td><td>and i know</td><td>and i know</td></tr><tr><td>i dont wanna</td><td>give a f**k</td><td>i dont wanna</td><td colspan=\"3\">i dont wanna time has come i dont wanna</td><td>i love you</td><td>you are the</td></tr><tr><td>i just cant</td><td>its time to</td><td>you and me</td><td>and i know</td><td>i can feel</td><td>and i know</td><td>its time to</td><td>i need to</td></tr><tr><td>i dont need</td><td>dont give a</td><td>i need you</td><td>and i dont</td><td>in my mind</td><td>you and me</td><td>the way you</td><td>close my eyes</td></tr><tr><td>i love you</td><td colspan=\"2\">what the f**k cant you see</td><td>dont give a</td><td>its too late</td><td>oh oh oh</td><td>in your eyes</td><td>i dont know</td></tr><tr><td colspan=\"2\">yeah yeah yeah in the hood</td><td>i need to</td><td>i wanna be</td><td colspan=\"2\">close my eyes its time to</td><td>i try to</td><td>i love you</td></tr><tr><td>i need to</td><td>i got a</td><td>i love you</td><td>and i cant</td><td>the time has</td><td>i need you</td><td colspan=\"2\">and you know my heart is</td></tr><tr><td>so hard to</td><td>on the block</td><td colspan=\"2\">what you want i dont need</td><td>cant you see</td><td colspan=\"2\">yeah yeah yeah i need you</td><td>let it go</td></tr><tr><td>i need you</td><td>im in the</td><td>you feel the</td><td>i just dont</td><td>be the same</td><td>in your eyes</td><td>and i will</td><td>i need you</td></tr><tr><td>in my eyes</td><td colspan=\"2\">i dont wanna in your eyes</td><td colspan=\"2\">cant you see in the sky</td><td>to make you</td><td>you want me</td><td>i know youre</td></tr></table>" | |
| }, | |
| "TABREF8": { | |
| "text": "", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td>1990-1995</td><td>1995-2000</td><td>2000-2005</td><td>2005-2010</td></tr><tr><td>da na</td><td>day away</td><td>away day</td><td>today away</td></tr><tr><td>way day</td><td>feel real</td><td>me be</td><td>town down</td></tr><tr><td>baby me</td><td>me be</td><td>know go</td><td>me be</td></tr><tr><td>go know</td><td>know go</td><td>wrong song</td><td>go know</td></tr><tr><td>be me</td><td>town down</td><td>day way</td><td>say away</td></tr><tr><td>soul control</td><td>day way</td><td>play day</td><td>right tonight</td></tr><tr><td>know show</td><td>oh know</td><td>right night</td><td>let better</td></tr><tr><td>tight night</td><td>stay away</td><td>say day</td><td>day away</td></tr><tr><td>down town</td><td>find mind</td><td>heart apart</td><td>alright light</td></tr><tr><td>day away</td><td>say away</td><td>say way</td><td>right night</td></tr></table>" | |
| }, | |
| "TABREF9": { | |
| "text": "Top ten rhyme pairs ranked by tf-idf, five year period.", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table/>" | |
| }, | |
| "TABREF10": { | |
| "text": "Top ten trigrams ranked by tf-idf, five year period. Terms that only occur in one genre's top 15 ranked list are emphasised.", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table/>" | |
| }, | |
| "TABREF12": { | |
| "text": "Top fifteen trigrams, ranked by document frequency and df-idf.", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table/>" | |
| }, | |
| "TABREF13": { | |
| "text": ".4222 0.0892 0.5555 0.0253 0.7333 0.0032 0.6888 0.0056 \u03c1 0.5640 0.0897 0.6969 0.0251 0.8787 0.0008 0.8666 0.0012Table 10: Correlation measure results for training list using Kendall's tau (\u03c4 ) and Spearman's rho (\u03c1).", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td/><td>Eq. 1</td><td>Eq. 2</td><td>Eq. 3</td><td>Eq. 4</td></tr><tr><td/><td>p</td><td>p</td><td>p</td><td>p</td></tr><tr><td>\u03c4 0Ex. 1</td><td>Ex. 2</td><td>Ex. 3</td><td/></tr><tr><td>p</td><td>p</td><td>p</td><td/></tr><tr><td colspan=\"3\">\u03c4 0.333 0.180 0.333 0.180 0.244 0.325</td><td/></tr><tr><td colspan=\"3\">\u03c1 0.466 0.174 0.479 0.162 0.345 0.328</td><td/></tr></table>" | |
| }, | |
| "TABREF14": { | |
| "text": "", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td>: Correlation measure results for random</td></tr><tr><td>list, using Kendall's tau (\u03c4 ) and Spearman's rho</td></tr><tr><td>(\u03c1).</td></tr></table>" | |
| }, | |
| "TABREF16": { | |
| "text": "Expected and application rankings for ten randomly selected songs.", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td>LyricWiki Collection</td><td/><td/><td/><td/></tr><tr><td>Number One Hits</td><td/><td/><td/><td/></tr><tr><td>0</td><td>1000</td><td>2000</td><td>3000</td><td>4000</td></tr></table>" | |
| } | |
| } | |
| } | |
| } |